This study retrieved data from the WoS database (with SCI-EXPANDED and SSCI only) created by Clarivate Analytics on 1 January 2023. The study topic was selected as Topic: “Green Finance” OR “Green Financial Effectiveness” and refined by Language: English AND Document types: Article OR Review Article OR Meeting Abstract. The time frame chosen for this paper was 2011–2022 because the first article was published in 2011.We refined and retrieved 533 records using the above settings. The literature data download completed the first step in the process of creating the database. Next, we imported the data into CiteSpace. CiteSpace can identify hot research topics and cutting-edge research in the field of green finance based on the frequency of popular keywords used in journal articles (Jia et al., 2019). Bursts of certain keywords can be used to analyze the evolution of green finance research and identify the latest research trends (Ding and Zhong, 2022). An outburst is a significant change in the value of a variable over a relatively short period of time (Jia et al., 2020). The value of a variable can change significantly in a relatively short period of time (Li and Wang, 2016). CiteSpace considered this type of change to be a way of identifying research frontiers (Ying et al., 2022).

As shown in Figure 1, co-occurring network analyses were performed using keywords such as “node type.” It indicated that recent popular research topics in green finance included green credit, green innovation, driving factor, climate change, performance, and economic growth. Topics with high centrality included energy, environmental performance, policy, carbon emission reduction, finance, economic development, determinant, technology, and management. As can also be seen from the visualization, green finance was linked to current political hotspots such as carbon finance and sustainable development and has been used by many scholars for cross-sectional research because of the policy-oriented nature of the subject itself. Overall, from the results analyzed by CiteSpace, the research in this field was divided into the following two categories, the context of the real-life contextual application of green finance and the factors influencing green finance.

According to the literature described above, and the statistical analysis of scholars' use of each factor, this study have developed a research framework to explain the influence on the performance of green finance, as shown in Figure 2. We propose that regional logistics, carbon emission index, regional finance, economic level, and environmental regulation will have an impact on green finance. Although scholars have analyzed each of these five factors having an influence on green finance individually, the causal relationship between green financial performance and these conditional variables is very complex, and it is not inevitable that a single variable plays a role in causing green finance to have high or non-high performance. So the study need to further explore the configurations of improving green financial performance and the differences in the green financial performance of causal factors in allocation terms.

Qualitative comparative analysis (QCA) is an empirical research method based on the pooling theory proposed by Charles C. Ragin, an American sociologist, and is generally applied to small and medium-sized samples (Du and Kim, 2021). QCA aims to focus on the grouping of configurations that arise when problematic outcomes occur and to analyze multiple configurations that lead to the same outcome from a pooling perspective by looking at real-life situational cases behind different configuration representatives (Chuah et al., 2021). QCA is considered a new type of research method between qualitative and quantitative (Mei et al., 2022). In addition, QCA is divided into three specific operations based on the type of variables: crispy set (csQCA), fuzzy set (fsQCA), and multi-set (mvQCA). Among them, csQCA is only used to deal with dichotomous variables (Chen et al., 2021), mvQCA is suitable for dealing with multi-category variables (Zhang and Long, 2022), while fsQCA can solve problems such as partial affiliation or causal changes (Ren et al., 2016).

The fsQCA method was chosen for three main reasons. First, the fsQCA method can reorganize many influencing factors to form a condition configuration and enable a holistic analysis to investigate how the combination of antecedent conditions leads to continuous changes in the outcome variables (Chu et al., 2019) and to explore the conditions and combinations of conditions that have explanatory power on the outcome variables (Zhong Weishun et al., 2022). The fsQCA approach can highlight the combined effect of multiple factors and the effectiveness mechanism driving regional green regional finance. Second, fsQCA is based on the principle of equivalence (Cheng et al., 2019), which summarizes various combinations of conditions that may lead to the same outcome (Xie et al., 2016). The fsQCA method has the advantage of considering multiple condition configurations leading to a specific outcome (Cheng et al., 2022), and it can focus on the diversity of causal relationships (Ren et al., 2016). Although the history of green regional finance varies across Chinese provinces and regions, the impacts of changes in the development of various factors on their ultimate performance may be somewhat consistent (Chuah et al., 2021). fsQCA can be applied to explore which condition configurations can lead to the improvement of regional green finance from a holistic perspective and to uncover the group causes of the relative lag of green regional finance in individual Chinese provinces. The fsQCA method can be applied to explore which condition configurations can be inferred to cause the improvement of regional green finance, as well as the condition configurations that cause the relative lag of green regional finance in individual provinces in China.

Therefore, in order to better elaborate the green regional finance configurations of Chinese provinces and regions and understand the utility mechanisms driving local green regional finance, this paper conducted an exploratory study using fsQCA.

Result variable: In the variable design process, we tried to ensure that the variables were as objective as possible and that the measures were well documented (Du and Kim, 2021). Therefore, we used the data from the green finance index for each region in the China Local Green Finance Development Report (2021) published by the International Institute of Green Finance of the Central University of Finance and Economics to measure the outcome variable. This report provides data tracking and an annual evaluation of 31 Chinese provinces in terms of green finance policy system construction, market product and service innovation, and regional exchange and cooperation. The report also analyzes the correlation between local green finance and local government financial capacity, macroeconomic development, financial system development, social and environmental conditions, and ecological environment quality. Its data meet authority, credibility, and objectivity (Cheng et al., 2022).

Conditional variables: Consideration of the indicators for the antecedent variables mainly followed the principle of diversification to ensure the reliability of the variables (Zhang and Long, 2022). Based on the literature and the specific conditions of each province and region, the conditional variables affecting the effectiveness of green finance in each province and region were set as five indicators: regional logistics, carbon emission index, regional finance, economic level, and environmental regulation. The data were processed using nine indicators with standardization and the principal component analysis method as a measure of the regional logistics variable, including the regional logistics value added, regional population, regional GDP index, regional total employment, regional logistics employment, per capita cargo turnover, regional total retail sales of consumer goods, regional total investment in fixed assets in logistics, and regional total freight volume. The carbon emission index was applied to calculate the carbon dioxide emissions from energy consumption in each province as a measure of the carbon emission index variables. The regional financial data from the Report on Local Green Financial Development in China (2021) published by the International Institute of Green Finance of the Central University of Finance and Economics were used as a measure of the criteria of the regional financial variable. For the economic level variable, the economic level output value of each province was used as a measure. Using industrial SO₂, industrial wastewater, and industrial soot emissions per unit of output value, a composite index was calculated for measuring the environmental regulation variable.

In order to explore the principle and inner mechanism of the green regional financial configurations among the regions, this paper considered 34 provincial-level administrative regions in China as the research objects. Because of the wide scope of data collection, there were gaps and deficiencies in the data of some provinces. In order to ensure the relative completeness and relevance of the final empirical findings (Zhong Weishun et al., 2022), 30 provinces (except Tibet, Hong Kong, Macao, and Taiwan) were finally selected as the research sample. In total, we spent more than 3 months collecting and processing data, and these raw data were obtained from the China Local Green Finance Development Report (2021), China Statistical Yearbook (2022), China Urban Statistical Yearbook (2022), China Energy Statistical Yearbook (2022), and the statistical yearbooks of each province in 2022.

Before using fsQCA to analyze the data, all variables were calibrated (Mei et al., 2022). The calibration was set at three anchor points according to the theoretical knowledge and combined with the actual situation: fully in, crossover, and fully out, and the calibrated variables were between 0 and 1 (Du and Kim, 2021). In this paper, the upper quartile (75% quartile value), the middle value (50% quartile value), and the lower quartile (25% quartile value) of the outcome variables and the five conditional variables for measuring the high performance of green finance in each province and region of China were selected as the three anchor points (Chuah et al., 2021). The three anchor points for measuring the absence of green finance high performance in each province and region of China were set as the opposite of the above (Chu et al., 2019). The three anchor points for each calibration standard are shown in Table 2.

In this paper, FsQCA 3.0 software was used to test the consistency and coverage of variables to measure the sufficiency and necessity of the relationship between variables (the symbol “∼” indicates the absence of the given variable). Consistency refers to the extent to which all samples included in the analysis conform to a given condition or combination of conditions that affect the occurrence of the result variable (Mei et al., 2022). Coverage refers to the extent to which a given condition or combination of conditions has explanatory power over the result variable (Zhang and Long, 2022). For single condition variable necessity analysis, if the necessary consistency of a conditional variable is greater than 0.9, the factor is necessary and the resulting necessary condition will be excluded in the subsequent sufficient condition analysis, and if it is less than 0.9, it cannot be considered necessary (Cheng et al., 2022). For the single condition variable sufficiency analysis, a factor is sufficient if the sufficient consistency of a conditional variable is greater than 0.8, indicating that this single factor constitutes a configuration to explain the results (Ren et al., 2016). As shown in Table 3, the results of the necessary condition test indicated that the necessary consistency of all conditional variables was below 0.9, indicating that the single conditional variable lacked sufficient explanatory power for the result variable and that no single variable was necessary to influence the result variable. The results of the sufficient condition test indicated that the sufficient consistency of all the single conditional variables was also less than 0.8, which was also insufficient to constitute a sufficient condition to influence the result variable, indicating that no single conditional variable constituted a configuration to explain the result variable.

After analyzing the data, the fsQCA 3.0 software presented three different solutions: complex, intermediate, and parsimonious (Wang Yunbo et al., 2022). The complex solution does not contain logical residuals (Liang et al., 2022), the intermediate solution contains only logical residuals that match the theoretical basis and the actual situation (Diao and Liu, 2021), and the simple solution contains all logical residuals but does not evaluate their reasonableness (Mei et al., 2022). Compared with the other two solutions, the intermediate solution was less complex, reasonable, and did not eliminate the necessary conditions, so the intermediate solution was chosen for the subsequent study (Chen et al., 2021). In addition, the intermediate solution determined the peripheral and core conditions that were useful for the outcome variable (Chu et al., 2019). If they do not exist in the simple solution but exist in the intermediate solution, they are peripheral conditions (Cheng et al., 2022). If they exist in both solutions, they are core conditions (Ren et al., 2016). The core condition has a greater effect on the outcome variable, and the peripheral condition plays a secondary role (Xie et al., 2016).

We adopted default thresholds to filter the data, with the case frequency benchmark set as 1 and the raw consistency benchmark as 0.8 (Du and Kim, 2021). This paper first assumed that the presence of all the conditional variables was likely to have a positive impact on green finance performance (Chu et al., 2019), which led to four configurations that produced high green finance performance (as shown in Table 4). The consistency of the results was 0.706, indicating that 70.6% of the provinces and regions that satisfied the four configurations had a high level of green finance performance. The coverage of the results was 0.661, indicating that the three types of configurations explained 66.1% of the provinces and regions with better green regional finance. Second, if the absence of the conditional variables had a negative impact on green finance performance (Mei et al., 2022), four types of configurations were derived to represent the absence of high green finance performance, and these four types of configurations explained 82.1% of the provinces and regions with a low performance of green regional finance (as shown in Table 5). For all cases that satisfied these four types of configurations, 79.3% of the provincial and regional cases showed low levels. In order to better compare the differences between the different configurations at various levels, we explored the four types of configurations that determined the emergence of high green finance performance and the four types of configurations that determined the absence of high green finance performance.

Robustness checks are an integral part of the analysis when using the fsQCA method (Mei et al., 2022). In this paper, we prioritized a specific approach in set theory for robustness checks (Du and Kim, 2021). First, using the varying consistency benchmark test, the raw consistency benchmark of 0.8 was reduced to 0.77, and the results indicated similarity to the previous condition configurations. Second, the frequency benchmark was adjusted by increasing the original benchmark from 1 to 2 (Ren et al., 2016), and the same results and consistent conclusions were still obtained. Therefore, it can be concluded that the findings of this paper were robust.

By comparing the combined relationship between different configurations, a substitution relationship between conditional variables can be derived to help provinces and regions improve the performance of green finance find a more suitable solution for themselves.

By comparing the four condition configurations that produce high green finance performance, three substitution relationships were found. First, comparing H1 with H2, we found that regional logistics (RL) can be substituted with regional finance (RF) and economic level (EL). This means that if regions with better control of carbon emissions, regardless of whether the environment is well-regulated or not, meet the conditions of superior logistics and transportation, or they meet the combination of “high financial development + high economic level,” they can generate a high performance of green finance. This had similarities and differences with the findings of a previous study (Jinru et al., 2022). The similarity was that RL did have an impact on green finance in each province and region. The difference was that, in the previous study, the RL factor influenced the green finance of each province and region only through its influence on environmental quality, but we found that RL can directly influence the green finance performance of provinces and regions as a core condition. Second, comparing H1 with H4 and H2 with H4, we found that the carbon emission index (CEI) could be replaced with RF, EL, and environmental regulation (ER), or with RL and ER. It suggested that if the region has excellent logistics and transportation capacity, it can show a high performance of green finance if the carbon emission is well controlled or the combination of “high financial development + high economic level + good environmental regulation” is satisfied. It also indicated that if the region has high financial development and a high economic level, it can maintain high green finance performance by controlling the carbon emission index or meeting the combination of “high logistics and transportation capacity + good environmental regulation.”

A further comparison of the four condition configurations that produced an absence of high green finance performance revealed a substitution relationship. Comparing AH2 with AH3, we found that an RL absence can be substituted with an EL absence and an ER absence. In other words, if a region has a high index of its own carbon emissions, the surrounding regional transportation pattern does not dominate or is at a low level of economic development, and the government lacks strong controls on environmental management, it will have a negative impact on the local green finance.

Three policy implications for policymakers and policy implementers in the provinces are set out below.

First, China’s provinces and regions must quickly refine the consensus reached in the development of green finance in each region, learn from the valuable experience of local governments in developing green finance, and build a green regional financial system with local characteristics. They must focus on the cities that have made achievements in the field of green finance, guide financial institutions to invest more in green finance, provide better green financial services, focus on creating generic demonstration projects, and enhance the guiding role of advanced demonstration cities. It is also necessary for universities and research institutes to increase support on cultivating excellent financial talents to provide professional vitality for the long-term development of the green financial system in each province.

Second, China’s provinces and regions need to rely on a solid industrial base and actively develop local logistics linkages to strengthen urban infrastructure and transportation. At the same time, governments in each province should also increase local resource environmental protection efforts, promote the development of green branding strategies actively, highlight the green development orientation, realize the optimal allocation of resources, prioritize the increase of green financial resource allocation to environmental protection industries, and standardize the policy and financial support to enterprises. This conclusion is roughly the same as the improved countermeasures proposed in previous studies (Wang Hui et al., 2021; Zhou et al., 2022), all aimed at providing a breeding ground for the development of urban green finance.

Third, based on policy orientation, Chinese provinces and regions should integrate green finance as a measurement indicator into the performance assessment system of local governments to stimulate them to be more proactive as a driving force to promote green finance development because a single policy utility is not enough to promote green finance development. We suggest that local governments introduce more detailed policy guidelines based on the country strategy to further stimulate market dynamics. For instance, a green information-sharing platform should be created to enhance provincial interconnections and find emerging development momentum in the society.