December 4, 2024

Saclung

The Future of Business, Today

Digital finance reduces urban carbon footprint pressure in 277 Chinese cities

Digital finance reduces urban carbon footprint pressure in 277 Chinese cities

Data sources

This article uses Chinese cities as a research sample, taking into account the background of digital finance and China’s carbon emission data, and selects 2011–2020 as the time window for the survey sample. Digital financial data comes from the Digital Finance Research Center of Peking University, and other city-related data comes from the “China Statistical Yearbook” and relevant data released by various city governments. Enterprise data comes from the CSMAR database. Additionally, after data collection integration, we removed samples with significant missing values and used linear interpolation to fill in some cities with fewer missing values. After data cleaning, we obtained 10 years of data for 277 cities, forming a sample of 2770 observations.

Variable calculation

  1. 1.

    Independent variable The independent variable in this paper is the Urban Carbon Footprint Pressure (CCF). Urban carbon footprint refers broadly to the pressure of carbon dioxide emissions faced by a city. This pressure arises from both natural and social factors. Prior to measuring carbon footprint pressure, this study reviews existing methods of urban carbon footprint measurement. Moran developed a networked global carbon footprint model, identifying urban–rural consumption patterns and purchasing power as primary predictors of urban residents’ carbon footprints26. Galli measured regional ecological carbon footprints based on carbon emissions produced by residents’ use of natural resources and generation of services27. Lombardi provided a phased summary of urban carbon footprint measurement techniques. They highlighted the determination of regional carbon emissions and the definition of ‘urban’ (or community) as primary considerations28. Some scholars argue that if the ratio of carbon emissions to economic output in a city approaches 1, it suggests a balanced development of environment and economy. A ratio exceeding 1 indicates that the city is overburdened with carbon29,30. Similarly, urban carbon footprint pressure specifically stems from increased carbon emissions and decreased vegetation absorption capacity. Hence, the measurement of urban carbon footprint pressure should consider both reducing emissions and enhancing carbon sinks. To this end, this study employs Formula (3) to assess urban carbon footprint pressure.

    $$ CCF_i = CE_i /CVS_i $$

    (1)

    In Formula (1), CCFi represents the Urban Carbon Footprint Pressure; CEi denotes the city’s carbon dioxide emission equivalents; and CVSi signifies the carbon absorption capacity of urban vegetation. This method of measuring urban carbon footprints has been widely applied in existing research31,32. Urban carbon emission data are calculated according to the “International Standard for Urban Greenhouse Gas Accounting.” The steps are as follows: First, establish the accounting boundaries. This paper considers Chinese cities as the accounting units for carbon emissions, using municipal administrative divisions as the basis for accounting boundaries. Second, identify the sources of carbon dioxide emissions. Following the “IPCC Guidelines for National Greenhouse Gas Inventories,” greenhouse gas emission sources are categorized into five main types: energy activities, industrial production, agricultural activities, land use changes, and waste disposal. The third step involves data collection and the calculation of urban carbon dioxide emissions using the emission factor method.

    $$ GHG_k = AD_k \times EF_k $$

    (2)

    In Formula (2), GHGk represents the greenhouse gas carbon dioxide (CO2). ADk refers to the activity data for production or consumption activities that cause CO2 emissions, where k represents the activities generating CO2. EFk denotes the emission factor for carbon. Additionally, this study builds on previous research by using field observation methods to collect and calculate data on carbon sinks33.

  2. 2.

    Dependen variable The dependent variable in this paper is Digital Finance (DIF). In existing research, the most widely used measure of digital finance is the Digital Inclusive Finance Index compiled by the Digital Finance Research Center of Peking University34,35. This study selects this index as the basis for measuring digital finance for two reasons: firstly, the Digital Finance Research Center of Peking University collaborates with Ant Financial to organize and compile the data, ensuring its scientific rigor and reliability. Secondly, Ant Financial utilizes internal digital technologies to record comprehensive and objective data on users’ electronic transactions. Furthermore, to study the impact of digital finance on urban carbon footprint pressure from multiple dimensions, we utilize the secondary indicators of the digital finance index for multidimensional measurement. These secondary indicators include the degree of digitalization in digital finance (DIG), the breadth of coverage of digital finance (COV), and the depth of usage of digital finance (USE).

  3. 3.

    Control variables To reduce the influence of confounding factors such as regional economic development disparities, educational differences, and technological level disparities, and to more accurately study the relationship between digital finance and urban carbon footprint pressure, this study introduces the following control variables: Economic development level. Foreign trade dependence Educational foundation. Technology level. Urbanization degree.

    Table 1 presents the computational explanations and descriptive statistics for the variables in question. According to the descriptive results, the maximum value of urban carbon footprint pressure is 16.51, the minimum value is 0.009, and the mean is 1.363. The mean significantly exceeds the minimum value and is less than one-tenth of the maximum value, indicating substantial disparities in carbon footprint pressure among cities. This further underscores the necessity of this study. The descriptive statistics for other data fall within reasonable ranges, with no evident outliers.

    Table 1 Definition and descriptive statistics of variables. SD is the standard deviation.

Model construction

To study the impact of digital finance on urban carbon footprint pressure, this study established the following estimation model:

$$ CCF_i,t = \alpha_0 + \alpha_1 DIF_i,t + \alpha_2 Control_i,t + \gamma_i + \theta_t + \varepsilon_i,t $$

(3)

In formula (2), CCFi,t represents the urban carbon footprint pressure, while DIFi,t serves as the core explanatory variable for digital finance. Controli,t comprises the control variables necessary for this study. γi denotes the fixed effects for individual cities, and θt encapsulates the fixed effects over time and years. Controlling for individual and temporal variations mitigates biases in the estimation results caused by individual differences and time trends. εi,t is the random disturbance. Here, i stands for the city, and t represents the year. Particular attention is given to the coefficient α1 of digital finance. If this coefficient is significantly negative, it suggests that digital finance effectively reduces the pressure of urban carbon footprints. Conversely, a positive coefficient would indicate that digital finance exacerbates this pressure. Additionally, to delve deeper into the intrinsic mechanism by which digital finance impacts urban carbon footprints, this study introduces interaction terms between the independent variables and mechanism variables based on Model (3), thereby constructing a mechanism testing model.

$$ CCF_i,t = \beta_0 + \beta_1 DIF_i,t + \beta_2 DIF_i,t \times M_i,t + \beta_3 Control_i,t + \gamma_i + \theta_t + \varepsilon_i,t $$

(4)

In Formula (4), Mi,t represents the mechanism variable, which is the coefficient for the interaction term, with all other variables consistent with those in Model (3). Specifically, if the coefficient for digital finance is significantly negative, and the coefficient for the interaction term is also significantly negative, it indicates that digital finance can alleviate urban carbon footprint pressure through this specific mechanism. This suggests a synergistic effect where digital finance not only directly reduces carbon footprint pressures but also enhances the effect through the mechanism variable, thereby further contributing to sustainability in urban environments.

Data validation

Differences in distribution of urban carbon footprint pressure

To more effectively illustrate the spatial distribution of urban carbon footprint pressures, we utilized ArcGIS software to visualize the data from 2011 and 2020. Figures 2 and 3 respectively display the spatial distribution of carbon footprint pressures in Chinese cities for these years. Observations from these figures reveal a declining trend in urban carbon footprint pressures, indicating significant progress in emission reduction and carbon sinks efforts in recent years. Specifically, several key observations are noted:

Figure 2
figure 2

Distribution of city carbon footprint pressure in 2011.

Figure 3
figure 3

Distribution of city carbon footprint pressure in 2020.

Spatial differences

There are clear spatial disparities in the intensity of urban carbon footprint pressures. From 2011 to 2020, these disparities are evident across different regions. Central China shows minimal changes with most cities maintaining low levels of pressure (below 1.6831). In contrast, cities in Southern and Northern China exhibit significant fluctuations. Notably, some cities in the north have experienced an increase in carbon footprint pressures, likely due to the high energy demands for heating in winter and cooling in summer, typically relying on energy-intensive systems which further contribute to carbon emissions. Conversely, many cities in the south have shown a marked decrease, particularly along the middle and lower reaches of the Yangtze River. This north–south disparity may be attributed to the colder winters and hotter summers in the north, as compared to the rich forest resources in the south where efforts to protect and restore forests help absorb atmospheric CO2 and increase regional carbon sinks capacity, thus alleviating urban carbon footprint pressures.

Spatial clustering

The distribution of urban carbon footprint pressures exhibits clear spatial clustering. For instance, in the southern region, cities such as Nanping, Sanming, Longyan, and Meizhou show contiguous characteristics of carbon pressures. In central China, the carbon pressures are interconnected and generally lower. This clustering is likely due to the uneven distribution of industries across China, with some regions hosting large-scale manufacturing and heavy industries, while others focus more on services and light industries. The high carbon emissions associated with manufacturing activities thus lead to clustered patterns in carbon footprint pressures.

This analysis not only highlights the regional variations and trends in carbon footprint pressures but also emphasizes the influence of industrial distribution and environmental policies on urban carbon management.

Trends in urban carbon footprint pressures

To analyze the trends in the distribution and centroid migration of urban carbon footprints, this study employed standard deviation ellipses in its mapping analysis. Figure 4 illustrates the ellipses for urban carbon footprints pressures in 2011, 2015, and 2020, along with their centroid migration trajectories. Based on the observations from Fig. 4, we can draw several conclusions:

Figure 4
figure 4

Standard deviation elliptical distribution and center of gravity migration trajectory chart.

Firstly, the spatial distribution of urban carbon footprint pressures in China for the years 2011, 2015, and 2020 predominantly followed a “northeast-southwest” orientation. This pattern is consistent with the trajectory of the Hu Line, indicating that urban carbon footprint pressures align with levels of economic development and population distribution.

Secondly, there was a noticeable trend of decreasing minor axes and increasing major axes of the standard deviation ellipses as the years progressed. This shift from a concentrated to a polarized spatial distribution of urban carbon footprint pressures highlights the stark regional differences in these pressures.Thirdly, further analysis reveals that between 2011 and 2020, the migration route of the urban carbon footprint pressure centroid initially moved towards the southwest before shifting to the northeast. This shift was largely driven by policies introduced by the Chinese government to reduce environmental pollution in the eastern regions. These policies led to the relocation of secondary industries to the southwest, thereby intensifying urban carbon footprint pressures in that region. Moreover, from 2015 to 2020, the centroid of urban carbon footprint pressures moved towards the northeast, likely due to the presence of heavy industries and manufacturing sectors in the northern regions, which are typically associated with high carbon emissions. Additionally, these industries often face significant challenges in achieving green transformation, complicating efforts towards carbon neutrality and exacerbating urban carbon footprint issues.

These findings confirm that urban carbon footprint pressure is driven by factors such as economic development and population activity. Furthermore, they underline the importance of using financial instruments to achieve coordinated development of the economy and the environment in today’s era. This emphasizes the need for digital financial developments that not only promote economic growth but also ensure environmental sustainability, thus addressing the twin challenges of economic and ecological health in urban environments.

Empirical analysis

Basic estimation model

Table 2 presents the baseline regression results, where the dependent variable is urban carbon footprint pressure. Columns (1) and (2) utilize a two-way fixed effects model for regression. To reduce bias from model selection, columns (3) and (4) adopt a random effects model, assuming that all individuals share the same intercept, and differences between individuals are random, primarily reflected in the settings of random disturbance terms. Additionally, columns (1) and (3) of Table 2 do not include control variables, while columns (2) and (4) incorporate them.Observing the baseline regression results, under the fixed effects model, the coefficient of digital finance is significantly negative at the 1% level, with values of − 0.4112 and − 0.3455 when control variables are not included and included, respectively. This indicates that digital finance can effectively reduce the pressure of urban carbon footprints, thus confirming Hypothesis 1. Furthermore, under the random effects model, the coefficient of digital finance remains significantly negative at the same 1% level. This result strengthens the evidence of digital finance’s role in mitigating urban carbon footprint pressures.

Table 2 Benchmark regression results.

Multi-dimensional analysis of digital finance

To delve deeper into the impact of various developmental dimensions of digital finance on urban carbon footprint pressures, secondary indicators of the digital finance index—coverage rate, usage depth, and degree of digitalization—were used as independent variables for estimation and testing. Table 3 reports the test results. Notably, the coefficient for digital finance coverage rate is significantly negative at the 1% level, with a value of − 0.9385, indicating that an expanded coverage of digital finance effectively suppresses urban carbon footprint pressures. However, it is important to note that the usage depth and degree of digitalization reported in columns (2) and (3) of Table 3 did not pass the significance tests. This result suggests that the insufficient depth of use and level of digitalization are key obstacles in the role of digital finance in reducing urban carbon footprint pressures. Several reasons may account for these outcomes:

Table 3 Multidimensional regression results.

First, a significant portion of the population lacks understanding and awareness of digital finance products and services. Many urban residents, particularly those less familiar with digital technologies, face challenges in understanding the operation, benefits, and potential risks of digital finance. This lack of understanding constitutes both a psychological and practical barrier that hinders the adoption and optimal use of digital finance services, thereby restraining their role in mitigating carbon footprint pressures. Second, issues related to personal data security and privacy dampen users’ trust in digital finance, hindering their engagement with these services. This problem is exacerbated by notable cyber attacks and insufficient regulatory frameworks, failing to instill confidence in potential users. This phenomenon further impedes the alleviating impact of deep usage of digital finance on urban carbon footprint pressures. Lastly, many urban areas, especially among low-income groups, still have limited access to digital technologies. The digital divide encompasses not only access issues but also variations in connection quality and the affordability of digital tools. These limitations restrict a significant portion of the urban population from participating in the digital finance ecosystem, thus limiting its potential to mitigate carbon footprint pressures effectively.

Robustness analysis

  1. 1.

    Instrumental variable method. In the baseline regression, even though time variations and individual differences are controlled, the potential correlation between the independent variables and the error terms still results in endogeneity issues. Therefore, this study adopts the instrumental variable (IV) method to decompose the endogenous variable into endogenous and exogenous components to eliminate possible endogenous effects. Building on the research by Li18, this study employs the Internet penetration rate as an instrumental variable. Columns (1) and (2) of Table 4 report the results of the two-stage least squares (2SLS) instrumental variable method, where column (1) shows the first-stage regression results and column (2) shows the second-stage regression results in detail. In addition, the instrumental variable (Inter) passed the Wald F-value test and the weak instrumental variable test, confirming that Internet penetration rate is an effective instrumental variable. The regression results show that even after further eliminating the endogenous effects, digital finance still has a suppressive effect on urban carbon footprint pressure.

  2. 2.

    Exclusion of special samples method. We recognize that centrally administered municipalities typically possess more fiscal resources and national support compared to other towns and cities. These advantages often enable such municipalities to develop more rapidly in economic and educational terms. To mitigate the impact of these factors, this paper re-evaluates the data excluding samples from centrally administered municipalities and performs linear estimation again. Column (3) of Table 4 presents the results after the exclusion of these municipalities. The coefficient of digital finance remains significantly negative at the 1% level, indicating that digital finance continues to alleviate urban carbon footprint pressures even when excluding the special influences of municipal authorities.

  3. 3.

    Eliminate bias in pilot cities. Pilot cities for carbon emission reduction are subject to more stringent policies and regulations aimed at decreasing emissions, coupled with policies that encourage such reductions. These factors could potentially confound the positive effects of digital finance. To avoid the influence of pilot cities and to enhance the generality and purity of the sample characteristics, this study excludes cities identified as carbon emission pilot cities during the survey period in its robustness tests. Specifically, data from 11 cities over five years were excluded, resulting in 55 observational samples. Column (4) of Table 4 presents the regression results after the exclusion of pilot cities. The coefficient for digital finance remains significantly negative, indicating that the baseline regression remains robust after removing the effects of the pilot cities.

  4. 4.

    Evaluating the impact of external policies on digital finance. To assess the influence of external policies or events on digital finance and thereby isolate the effects of policy from those due to endogeneity issues, consider the 2016 initiative under the “G20 High-Level Principles on Digital Inclusive Finance,” where China advocated the use of digital technologies to promote inclusive finance. This policy greatly accelerated the development of digital finance. However, the establishment of financial service infrastructures has also introduced new carbon emission challenges. Thus, the question arises: does this policy impact the effectiveness of digital finance? Column (5) of Table 4 shows the test results for exogenous policy impact. Since the policy was implemented in 2016, the period from 2016 to 2020 serves as the regression test window. The results indicate that the coefficient of digital finance is significantly negative at the 1% level, suggesting that digital finance continues to mitigate urban carbon footprint pressures even under the influence of this policy.

  5. 5.

    Validation using firm-level data in baseline regression. In our baseline regression, we utilized city-level data for analysis. To enhance the robustness of our research findings, we now validate these results using firm-level data. First, we assess the level of digital finance in firms based on the development of digital finance in their respective registered locations. We then substitute firm-level carbon emissions for urban carbon footprint pressures, as carbon sinks is typically measured at the macro level and cannot be precisely quantified by firms. Table 4, Column (6) displays the regression results at the firm level. The results indicate that the coefficient of digital finance is significantly negative at the 1% level, demonstrating that the integration of digital finance effectively reduces carbon emissions in firm production and operations. This suggests that the application of digital finance at the firm level can alleviate carbon footprint pressures, further confirming the results of the baseline regression.

  6. 6.

    Adding control variables. Considering that existing control variables already address aspects of urban economy, education, technology, and urbanization, there could still be biases due to omitted variables, especially those related to urban carbon footprints. To address this, this study introduces additional control variables: per capita green space and per capita park area. As primary sources of urban carbon sinks, green spaces and parks play a significant role in the urban carbon cycle and climate change mitigation. Incorporating these variables aims to more directly mitigate the omission of relevant environmental features. Table 5, Column (1), reports the regression results with these characteristic control variables added. The coefficient of digital finance is significantly positive, indicating that the baseline regression results remain valid even when controlling for environmental characteristics.

  7. 7.

    Variable measurement substitution method. Different measurement methods can introduce varying degrees of error and uncertainty. Substituting the measurement approaches for both independent and dependent variables can help reduce these impacts. The foundation of digital finance is the digital transformation of financial services, where finance essentially facilitates the conversion of savings into investments by individuals and households, allows enterprises to raise funds for expansion, and enables governments to fund public projects. Thus, this study substitutes the level of financial development for digital finance, measuring urban financial development using the logarithm of city residents’ savings and financial management. Furthermore, since the root cause of urban carbon footprint pressure is excessive carbon emissions, Carbon Dioxide Equivalent (CDE) is used as a substitute for urban carbon footprint pressure as the dependent variable. Columns (2) and (3) of Table 5 display the test results using the substituted variable measurement methods. The results indicate that the substituted variables still pass the significance tests, further enhancing the robustness of the baseline regression results.

Table 4 Instrumental variable method and replacement sample test results.
Table 5 Mechanism test results.

Influence path analysis

  1. 1.

    Reducing the number of bank branches. The determinants of the number of bank branches in cities are highly complex, influenced by various factors such as market conditions, economic conditions, technology, and policies36. Therefore, this study selects the number of bank branches (Bank) as one of the mechanism variables and measures it by the number of bank branches in the city. Additionally, we use Model (2) to test this mechanism, with the estimated results shown in column (1) of Table 5. The interaction term between digital finance and the number of bank branches is significantly positive at the 1% level, while the coefficient for digital finance is significantly negative at the 1% level.This result indicates that digital finance can play an important role in mitigating urban carbon footprint pressure by reducing the number of bank branches in the city. Hypothesis 2 is thus confirmed.

  2. 2.

    Increasing residents’ environmental awareness. With the iterative upgrading of digital technology, the internet can use big data to timely capture residents’ attention to environmental causes, reflecting their preferences and intentions towards natural resource conservation or green living. Moreover, residents’ attention to the environment can effectively reduce the likelihood of polluting enterprises entering the market, thus controlling pollution emissions at the source37. To test Hypothesis 3, this study quantifies residents’ environmental awareness (PEC). We use Baidu search engine queries for environmental keywords to measure residents’ environmental awareness. This method has been widely applied in existing research38,39. Column (2) of Table 5 reports the potential mechanism of residents’ environmental awareness. The coefficient for digital finance is significantly negative at the 1% level, and the coefficient for the interaction term is also significantly negative, indicating that the application of digital finance has increased residents’ environmental awareness, thereby playing an important role in reducing urban carbon footprint pressure. Hypothesis 3 is thus confirmed.

Nonlinear analysis

Further reflection reveals that the development of digital finance relies on the iterative advancement of digital technology, which, as a form of production technology innovation, exhibits marginal increasing returns. In the early stages of technological innovation, the learning curve effect typically emerges. In simple terms, as the frequency of use increases and experience accumulates, production efficiency and technological application capabilities gradually improve. Initial investments and learning costs are high, but over time, these costs decrease, making the application of technology more efficient40. These marginal changes subsequently affect digital finance, potentially resulting in diminishing returns. Beyond technological factors, the nonlinear effects of digital finance are also evident in the snowball effect of data collection and retrieval, where data accumulation may lead to marginal increasing returns. Consequently, due to technological or market factors, the impact of digital finance on urban carbon footprints may also be nonlinear.Based on this, our study incorporates a quadratic term for digital finance into the original Model (1) to capture the nonlinear relationship, aiming to further understand the complex relationship between digital finance and urban carbon footprint pressure. If the coefficient of the quadratic term passes the significance test, it indicates the presence of nonlinear effects.

Table 6 reports the results of the nonlinear effect test. The coefficient for the linear term of digital finance is significantly negative at the 1% level, while the coefficient for the quadratic term of digital finance is significantly positive at the 1% level. This result indicates that digital finance exerts a nonlinear effect on urban carbon footprint pressure. To further verify the nature of this nonlinear relationship, this study calculates the nonlinear inflection point. The calculation reveals that the nonlinear inflection point is 3.9, whereas the maximum value of digital finance is approximately 3.3. This result suggests that the impact of digital finance follows a linear relationship and shows a monotonic change within the investigated range. However, when the level of digital finance exceeds the inflection point, a nonlinear relationship emerges.

To visually illustrate this change, the study visualizes the nonlinear relationship. Figure 5 shows the visualized nonlinear variation. Within the investigated range of digital finance, as digital finance increases, the pressure on urban carbon footprints gradually decreases. However, the suppression rate gradually diminishes according to the curve’s variation amplitude. This phenomenon could be because the deeper development of digital finance leads urban residents to rely more on electronic devices such as smartphones, tablets, and computers, resulting in a sharp increase in electricity consumption, which is detrimental to alleviating urban carbon footprint pressure. Therefore, it is necessary to seek a balance between the efficiency brought by digital finance and its negative impacts to achieve long-term green and low-carbon development.

Figure 5
figure 5

Nonlinear change trend chart.

Heterogeneity analysis

Regional differences

Considering the potential differences in natural and human factors across various locations is crucial. Natural factors include climate, natural resources, topographical features, and ecosystem conditions, while human factors encompass economic disparities, cultural customs, urbanization differences, and policy variations. These differences can impact the influence of digital finance on urban carbon footprint pressure in different ways. This study examines the inhibitory effect of digital finance on urban carbon footprint pressure in different regional contexts. Table 7 reports the sub-sample regression results for China’s three major regions. From the regression results, it is evident that the inhibitory effect of digital finance on urban carbon footprint pressure is most significant in the central region, followed by the eastern region, while the effect in the western region is not significant. The possible reason for this is that compared to the eastern and central regions, the western region lags economically, with relatively lower levels of urban infrastructure, technology, and digital maturity. The application of digital finance technology is limited in this area. This limitation hinders the full potential of digital finance in energy management, resource allocation, and carbon footprint reduction. Consequently, the alleviation effect on urban carbon footprint pressure is not evident in the western region.

Table 7 Regional heterogeneity.

Differences in sunlight duration

The primary source of carbon sinks in urban areas is vegetation, which absorbs atmospheric carbon dioxide through photosynthesis and stores it as organic matter. The main driver of photosynthesis is solar radiation; therefore, sunlight duration is a key factor influencing the carbon sinks capacity of urban vegetation. This study considers that due to the significant latitudinal differences between northern and southern China, the solar altitude angle varies greatly among cities, resulting in different sunlight durations. Additionally, diverse topographies, such as mountainous areas and plateaus, create shadow effects, leading to shorter sunlight durations in some cities, while sunlight duration is relatively longer in foothill or plain areas.Based on these factors, we explore the differential effects of digital finance on alleviating urban carbon balance pressure under varying sunlight durations. It is important to note that urban sunlight duration is calculated based on the interval between sunrise and sunset. Sunrise is defined as the moment when the solar altitude angle exceeds 0 degrees, and sunset is defined as the moment when it falls below 0 degrees. The solar altitude angle is determined by the city’s dimensions, solar declination, and solar hour angle. Table 8 reports the regression results for samples with long and short sunlight durations. We use the 50th percentile of all samples as the critical value. Samples exceeding this critical value are categorized as having long sunlight durations, while those below it are categorized as having short sunlight durations. In the regression for long sunlight duration samples, digital finance significantly alleviates urban carbon footprint pressure. However, in the regression for short sunlight duration samples, digital finance increases urban carbon footprint pressure, indicating that shorter sunlight durations are unfavorable for reducing carbon footprint pressure.

Table 8 Daylight duration heterogeneity.

link

Copyright © All rights reserved. | Newsphere by AF themes.