Research on the impact of financial agglomeration on the coordinated development of urban ecological green: based on the empirical comparison of four metropolitan areas in the Yangtze River Delta of China

As key carriers of new urbanization, metropolitan areas should pay more attention to the green and coordinated development of economy, society, and environment. Finance is an important tool to support China’s high-quality development. Exploring the key dynamics and mechanisms of financial agglomeration for a green and coordinated development is important to obtain strategic support for the green development of the Yangtze River Delta metropolitan area. Using panel data of 25 prefecture-level cities in the four major metropolitan areas of the Yangtze River Delta region from 2003 to 2019, a Dubin model of three types of spatial weight matrices is constructed to explore the impact of financial agglomeration on coordinated ecological green development. Results show that, first, the positive local and neighborhood effects of financial agglomeration on the coordinated ecological green development are mainly reflected in the Hefei, Hangzhou, and Shanghai metropolitan areas and that the intensity of such effect decreases sequentially. Information transfer and technology correlation are important mechanisms to improve the local effect of financial agglomeration, and the neighborhood effect varies from one metropolitan area to another. Second, in addition to the adverse impact of the concentration of financial personnel on the ecological green integration, the agglomeration of securities, banking, and insurance industries can release “green driving potential energy,” and the intensity of contribution gradually weakens. Third, the financial agglomeration of Shanghai exerts the strongest radiation effect, those of Nanjing and Hangzhou respectively demonstrate a “U”-shaped and inverted “U”-shaped impact, and that of Hefei does not play a role.


Introduction
Since China's reform and opening up, the Yangtze River Delta has witnessed a rapid urbanization and industrialization in its layout and has transformed into one of the most dynamic economies with the best urban construction and financial conditions in the country. However, certain problems, including factor cost pressure, ecological and environmental pressure, and ecological imbalance, are particularly serious in the region. The overall program of the Yangtze River Delta Ecological Green Integrated Development Demonstration Zone (hereinafter referred to as "the Program") has been designed as a major strategic measure to understand the new urbanization, regional integration, and deep ecological civilization construction of the region since the development of the Yangtze River Economic Belt, the regional integrated construction, and the establishment of city clusters. The guiding ideology of Responsible Editor: Nicholas Apergis the Program points out that the demonstration area should realize an organic unity of green economy, high-quality life, and sustainable development. The green and coordinated development of production, life, and ecology has become critical to the urban economic construction of the Yangtze River Delta. The metropolitan area plays a key role in accelerating regional integration construction and optimizing the new urbanization layout. A comparative study of the key driving factors and their action mechanisms for the coordinated ecological and green development of the Yangtze River Delta metropolitan areas is thus essential to find a strategic support point that suits each of these metropolitan areas.
Issued in 2019, China's "Guidance on Fostering the Development of Modernized Metropolitan Areas" formally defines metropolitan areas as spatial forms of urbanization that are centered on mega-cities or large cities with strong radiationdriven functions, with a 1-h commuting circle serving as the basic scope. Unlike the wide-area urbanization pattern of urban clusters with multiple centers, the new urbanization pattern of metropolitan areas whose development focuses on a single center can better play the role of the central city in economic, social, and ecological radiation and drive ). The conception of and problems faced by metropolitan areas have received scholarly attention as early as the beginning of the twenty-first century, with some scholars comprehensively investigating their conceptual definitions, development plans, driving factors, strategic paths, and policies. Subsequent studies have mostly focused on certain aspects, such as economic issues, social services, environmental constraints, innovation areas, kernel comparisons, government models, and policy effects. In the green coordinated development of the Yangtze River Delta region, which is the focus of this paper, related studies have mostly focused on the spatial and temporal evolutionary trends and the related driving factors (Pan et al. 2020). In their social network analysis, Zhou et al. (2022) found that large cities trigger a siphon effect and that narrowing the way cities regulate environmental protection is conducive to coordinated regional green development. Yang et al. (2020a, b) found that the main driving force for the coordinated green development of the Yangtze River Delta is high-quality economic development, whereas the main obstacle is the development of high-polluting industries. Liao and Li (2022) found that the specialization of green technology innovation not only improves the level of urban green development but also reduces the gap between central and non-central cities in China. These findings provide rich empirical evidence for this paper to explore the coordinated economic, social, and ecological green development of the Yangtze River Delta metropolitan areas.
As the bloodline of the real economy, finance not only influences the efficiency of green economic development through financing constraints but also plays an important role in the infrastructure, public services, and ecological environment of society. The high spatial concentration of the financial industry in central cities is a typical feature and inevitable trend of modern financial industry development (Ye et al. 2018). Some scholars have explored the economic effects of financial agglomeration (FA). At the macro level, FA theoretically promotes sustainable urban economic growth through economies of scale, network economy effects, and environmental improvement . FA also promotes high-quality economic development through environmental improvement effects. At the meso level, Buera et al. (2011) found that FA contributes to the sustainable economic growth of cities through economies of scale and network economies and verified the role of FA in promoting the upgrading of urban industrial structures. At the micro level, Gao and Jin (2022) found that the development of fintech brought about by agglomeration can increase corporate service capabilities and promote enterprise innovation by reshaping financial services. Some scholars have also explored the ecological effect of FA mainly in terms of its direct effect, spatial effect, threshold effect, and mechanism of action. Using the spatial Durbin model, Yuan et al. (2019) detected the significant positive direct effect and spatial spillover effect of FA on urban green development. This spatial effect has a spillover range with the provincial boundary as the proximate value (Yuan et al. 2020a). Tian et al. (2021) found that different characteristics of FA have varying effects on the green development of urban agglomerations and argued that the Yangtze River Delta urban agglomeration cannot rely on traditional FA to promote green development. Feng et al. (2022) identified industrial structure upgrading, labor force upgrading, and technological innovation as important intermediate variables that allow FA to influence urban green development.
In sum, FA can produce greening effects on both urban economy and ecology. However, given the urgent need for an organic unification of green economy, high quality of life, and sustainable development in the Yangtze River Delta, can FA simultaneously unleash the "driving green momentum" of greening production while optimizing society and deepening environmental protection? Does FA affect each financial entity differently? What are the intermediate mechanisms of FA for green and coordinated development? These questions have neither been explored in the literature or analyzed from the metropolitan area perspective. In fact, the Yangtze River Delta cities not only serve as the key planning areas for China's green economic development but are also among those regions with the strongest financial endowment in the country. Therefore, investigating whether financial development under the trend of agglomeration can promote the ecological green coordinated development of the Yangtze River Delta has great practical significance. In addition, new-type urbanization is a necessary path for China to promote modernization. The development of metropolitan areas shows the greatest potential under the vigorous promotion of co-location development and the establishment of a large unified market ). These metropolitan areas not only optimize the spatial structure of the population and economy but also effectively play the role of radiation of single central cities and facilitate the green and coordinated development of urban and rural areas . Six of the 34 metropolitan areas in China are located in the Yangtze River Delta region and are all relatively mature. Therefore, the ecological green coordinated development of the Yangtze River Delta metropolitan area warrants further study. To this end, this paper takes the Yangtze River Delta metropolitan areas as its research objects, investigates the influence of FA and the agglomeration of different subjects on their coordinated ecological and green development, verifies the possible mechanism effect by using an improved spatial weight matrix, and further explores the financial radiation capacity of central cities in each metropolitan area.
The marginal contributions of this paper lie in three aspects. First, this paper explores the influence mechanism of FA on ecologically green coordinated development and the role of information transfer and technology linkage in the path influence, thereby contributing to the literature. Second, by taking Yangtze River Delta metropolitan areas as its research objects, this study enriches the in-depth inquiries into the role of FA in promoting the coordinated green development of small regions and provides an empirical basis for promoting the construction of new urbanization in China. Third, through a comparative analysis of the greening effect of FA and the radiation effect of the central city, this study identifies the pivot and power point in optimizing the green and coordinated development of Yangtze River Delta metropolitan areas.

Analysis of the impact of financial agglomeration on ecological green coordinated development
The basic requirement of ecological green coordinated development is the organic unity of production greening, social optimization, and environmental protection deepening. To analyze the direct impact of FA on ecological green coordinated development, we need to confirm whether FA can bring about simultaneous economic, social, and ecological "greening" effects by maximizing the core advantage of agglomeration. First, FA can release the effect of economies of scale. An increase in financial scale improves the efficiency of financial services, reduces the transaction costs between financial institutions and social entities, and promotes social welfare (Fang et al. 2020). Second, FA can optimize the allocation of resources to improve resource utilization efficiency (Qu et al. 2020), restrain enterprises to engage in green production through financing thresholds, support the development of high-tech industries, force high-polluting enterprises to innovate their technologies, and improve the green efficiency of the economy . Financial agglomeration can guide the benign competition between manufacturing and service industries in the market (Gao et al. 2021), to achieve the rationalization and advanced transformation and upgrading of industrial structure in resource allocation effect, forming an industrial structure effect. Industrial structure upgrading is an important way to block pollution from the source, optimize the living environment, and create higher value for society (Liu et al. 2022). The financial industry itself belongs to the environmental protection industry (Peng et al. 2022). In addition to implicitly raising the environmental awareness of surrounding enterprises, the corporate positioning of this industry is also closely related to social life (Yuan et al. 2020b). Moreover, financial capital supports the construction of new urbanization (Ahmad et al. 2021), financial technologies improve the quality of life of residents (Nasir et al. 2021), and financial jobs attract high-quality labor . In sum, the financial industry promotes the quality of life in cities.
FA also provides paths and financial support for financial institutions to achieve a comprehensive regional coverage through network nodes, and this spider-web distribution realizes a spatial connection among financial institutions and thereby provides a theoretical basis for FA to release spatial spillover effects . The spatial spillover effects of FA are characterized by both the "polarization" and "trickle-down" effects (Yuan et al. 2019). These effects are often related to the degree of FA of the central city, the gap between the FA of the central city and its neighboring cities, and the cooperation and competition among cities in the metropolitan area development mode. At the early stages of agglomeration, cities cannot fully release their agglomeration efficiency and bring about the economies of scale effect of cost reduction advantages. The uncoordinated distribution of financial resources in the economic, social, and ecological fields may even lead to negative feedback (Wang et al. 2022a, b). Furthermore, central cities require more financial resources to support their development, which will lead to the polarization phenomenon where the resources of neighboring cities are crowded out. The failure to close financial gaps with neighboring cities will only strengthen such polarization phenomenon, thereby leading to the negative impact of local FA on the coordinated ecological and green development of neighboring areas. However, when FA takes shape, the financial resources of the central city tend to be saturated, which not only fully releases greening efficiency to strengthen the local effect but also generates the spatial impetus from the center to the periphery to realize the spatial spillover of financial resources and promote the development of neighboring cities (Guo et al. 2020). On the basis of these arguments, the following hypothesis is proposed: H1: FA can influence the coordinated development of local and neighboring ecological greenery without producing any impact at the early stages and produces a positive impact upon taking shape.
Analysis of the mechanism of the role of financial agglomeration in ecological green coordinated development Information transmission and technology correlation are important manifestations of FA releasing network effects. On the one hand, the scale effect of financial nodes set up across regions facilitates information disclosure, information transparency, and green innovation technology diffusion. The effective transfer of information within a region not only strengthens the cooperation and exchange among institutions and improves the efficiency of sharing factors and achievement resources but also enables a faster and more effective screening of enterprise information, facilitates the screening of polluting enterprises and environmental protection enterprises, and restricts polluters through financing threshold constraints . Such information transfer also reduces the financial and time costs of searching for financial information for enterprises, thus allowing them to quickly and timely capture market financing channels and financing conditions, accelerating their green transformation, improving their operational efficiency, and creating more value for the economy, society, and ecology (Feng et al. 2021). However, overemphasizing horizontal comparison, which eventually triggers a vicious competition, can lead to resource mismatch and market failure that will reduce the willingness of enterprises to engage in green transformation, decrease their profits, and give rise to the moral risk of information asymmetry (Chen et al. 2021). On the other hand, FA also produces a significant innovation effect. Financial institutions provide financing support for the development of high-tech industries and the green transformation of polluting enterprises, which can effectively improve production processes, reduce the energy consumption per unit, and realize the accumulation, expansion, and application of green innovative technologies (Cao et al. 2021). In addition to capital-driven innovation, talent serves as an important hub for the production and dissemination of innovative technologies (Zhuang and Chu 2021). The FA area is densely populated with talents and has a rapid flow of factor and information resources that effectively improves the efficiency of innovative technologies on the ground.
The financial information network also promotes the crossregional flow and association of technologies (Hsu et al. 2021). However, the irrational use of innovation capital will reduce social welfare, the over-referencing of innovative technologies will increase energy consumption to a certain extent, and the use of substandard new technologies will introduce new pollution problems (Zhou et al. 2019). The inefficient integration of financial development and technological innovation and the lack of institutional innovation and financial support can also lead to the negative impact of financial development on green development (Cao et al. 2022). On the basis of these arguments, this paper hypothesizes the following: H2: FA can influence the green and coordinated development of local and neighboring areas through information transfer and technology linkages.
The mechanism is illustrated in Fig. 1.

Study design
Variable measurement and description 1. Explanatory variable: Ecological green coordination degree (EGI). The Program clearly states that urban construction should achieve an organic unity of green economy, high-quality life, and sustainable development. The green and coordinated development of urban economy, society, and environment is the basic requirement of integrated development in the metropolitan area. The concept of green development should be integrated into the three major spaces of economic production, residential life, and environmental ecology, and a corresponding index system from the three aspects of green production, green life, and green ecology should be built. A complete evaluation index system can comprehensively, scientifically, and effectively reflect the current degree of coordinated ecological green development for cities. Therefore, based on the in-depth analysis of the guiding ideology of the Program, this study integrates and optimizes the contributions from relevant scholars in building evaluation indexes for the economic, living, environmental, and ecological dimensions and in constructing the three subsystems of production greening, social optimization, and environmental protection deepening (Table 1). This study also measures the EGI of each city based on these three subsystems by using the coupled coordination model (Yang et al. 2020a, b).

Fig. 1
The mechanism of FA affecting EGI (a) Production greening subsystem. The concept of green development of the economy maximizes economic output while consuming less input and minimizing environmental pollution. The core idea of this concept is to improve production efficiency. Green and efficient production methods are necessary for the economy to achieve green development (Qiu et al. 2021). This paper uses the Super-SBM-VRS model with non-expected outputs to measure the production greening index (Yao et al. 2020). This model fits well with the concept of economic green development, including the three first-level indicators of input, expected output, and non-expected output. Labor, capital, innovation, and energy are selected as input indicators along with year-end total employment, capital inventory, government expenditures on science and technology, and total natural gas supply. Among these indicators, capital inventory is measured by the perpetual inventory method following the practice of Han and Yang (2020). Urban GDP is selected as the expected output indicator, whereas industrial wastewater emissions, industrial sulfur dioxide emissions, and industrial smoke (dust) emissions are selected as non-expected output indicators.
(b) Social optimization subsystem. The green transformation of society requires the joint participation of the government and people in practicing a green lifestyle, improving quality of life, and offering greater welfare protection. Therefore, improving the welfare protection of residents is the focus of social optimization and upgrading , whereas improving infrastructure construction and public services plays a key part (Ronchi et al. 2020;Xing et al. 2021). GDP per capita and the employees' average wage are selected as welfare protection indicators, mobile phone subscribers, public bus (electric) passenger volume, and the number of cabs are selected as infrastructure indicators, and hospital beds and library collections are selected as public service indi-cators. A principal component analysis is conducted to measure the social optimization index.
(c) Environmental protection deepening subsystem. Environmental green transformation is established on effective environmental regulation, which requires the establishment of a reasonable current evaluation mechanism and the subsequent construction planning for existing environmental protection projects (OuYang et al. 2016). Current environmental quality is used to evaluate current environmental protection works, whereas pollution treatment efficiency measures future pollution control capacity. Improving environmental quality and pollution control capacity thus plays a key role in environmental protection deepening. The greening coverage rate of the built-up area and park green area are selected as environmental quality indicators, and the solid waste, sewage, and garbage treatment rates are selected as pollution treatment efficiency indicators. 2. Explanatory variable: Financial agglomeration (FA). FA is mainly assessed using the single-indicator method (location entropy index, EG index, etc.) and multi-indicator method (principal component analysis, entropy weight method, etc.). Given the rapid financial development in China, the securities and insurance industries play increasingly important roles in the financial system. Therefore, the use of single-indicator measurement methods can lead to the problem wherein the measured values do not match the real situation and cannot accurately reflect the level of agglomeration. Following Wang et al. (2019), this paper constructs a comprehensive evaluation indicator system that includes different financial industries and financial personnel to measure FA and assigns certain weights to the selected indicators ( Table 2). The deposit and loan balance ratio to built-up area, the deposit and loan balance ratio to population, and the savings ratio to population are selected Internet users per capita and the number of green innovation patent applications, respectively. The mediating variables are integrated into the setting of the spatial weight matrix to improve the traditional geographic distance matrix through the gravity model (Chen et al. 2021). The network information resources or green innovation technologies of each city in the city cluster are correlated with geographical distance, and the existence of the mediating effect is verified by comparing the coefficients of the model regression results under different matrices. Accordingly, two spatial weight matrices, namely, information distance matrix and technology distance matrix, are constructed. On the one hand, the information distance matrix draws on Yuan et al. (2019) to measure the level of information technology by the number of international Internet users per cap- On the other hand, the technology distance matrix draws on Li et al. (2021) to measure the level of innovation technology by the number of green innovation patent applications,W 2 ij = i and j represent different cities in the metropolitan area, T i T j and I i I j denote the strength of inter-city information linkage and technology linkage, respectively, and D ij is the geographical distance between cities. 4. Control variables. On the basis of relevant factors affecting green development, this paper controls for the following variables.
(a) Foreign investment (FDI) is expressed in terms of the actual amount of foreign capital used, with the annual average exchange rate converting US dollar units into RMB units. On the one hand, foreign investment brings more advanced and efficient production technologies to the host country and improves production efficiency to promote the green transformation of industry. On the other hand, excessive and blind references to foreign investment tend to form path dependence that inhibits local technological innovation, results in an excessive consumption of resources, and intensifies environmental pressure .
(b) Drawing on Gu and Shen (2021), human resource quality (LAB) is expressed in terms of the number of students as a percentage of land area. The greening effect of human resources is influenced by their quality and regional distribution. The concentration of high-quality talents improves the efficiency of enterprise operations and regional innovation capacity, realizes the accumulation of human capital and innovative technology spillover to guide the green development of the economy, restrains the behavior of individuals according to the environmental moral norms, teaches the concepts of environmental protection and cultural literacy, and promotes low-carbon green life. However, the development of talents in the center can lead to an uneven regional distribution, thereby crowding the central city and limiting the talent available in surrounding cities, both of which are not conducive to green development (Kang et al. 2022).
(c) Environmental manpower density (EVO) is expressed by the number of employees in the water, environment, and public facilities management industries as a percentage of the total number of unit employees. The greening effect of environmental manpower input is influenced by the level of specialization and the strength of environmental regulation. A balanced distribution of quality and quantity can effectively improve the cityscape of cities and contribute to the healthy development of the ecological environment, whereas an unbalanced distribution exerts a negative impact (Soomro et al. 2021).
(d) Urban construction level (ISL) is expressed as road area per capita. The influence of urban construction level on ecologically green and coordinated development is limited by governmental behavior. Urban construction that is oriented by the concept of green and sustainable development can better expand the bearing capacity of local resources and environment and improve the quality of green life of residents. Conversely, an excessive and blind urban construction will aggravate air and land pollution, intensify energy consumption, and lead to resource wastage ).

Model construction
First, to verify the direct and spatial effects of FA on EGI, this paper uses the spatial Durbin model to compare and analyze the effects of FA in Yangtze River Delta metropolitan areas. To compare the effects of the FA of different subjects on EGI, FA is subdivided into BAN, SEC, INS, and PEO based on the different nature of industries considered in this paper.
where is the spatial autoregressive coefficient, 1 and 3 are the local effect coefficients of FA and control variables on EGI, 2 and 4 are the spatial effect coefficients, and W is the spatial weight matrix that can be classified into three types. In the benchmark regression, the spatial correlation strength between cities is assumed to be only negatively related to the geographical distance, and the reciprocal matrix of geographical distance is used. When verifying the mechanism of information transmission and technology correlation, the improved information distance matrix and technology distance matrix are used. The specific method has been described in the selection of intermediary variables in the previous section and hence will not be repeated here. i and i represent the time and individual fixed effects, and i,t are random error terms.
Second, this study further verifies whether the central cities in the six metropolitan areas have a radiation-driven or absorption-polarizing effect on the other cities. Drawing on Chen and Chen (2012), the geographical distance from the central city to the other cities is incorporated into the FA equation to form the central spillover effect variable (FAC): FAC ct = FA ct ∕d ci , where FA ct denotes the FA level of the central city in year t, and d ci denotes the geographic distance between the central city and other cities. Chen and Chen (2012) pointed out that the effect of the central city on the remaining cities within a wide spatial scope may be nonlinear. The relationship between the central city and neighboring cities is a spiral process from "this dissipates and the other grows" to "synergistic development" (Chen and Li 2017), and this process is related to the development degree of the central city. The promoting, inhibiting, and nonlinear effects of central cities on neighboring cities have also been verified by other scholars (Camagni et al. 2017;Harms 2019;Capello and Camagni 2000;Hughes and Holland 1994). Therefore, linear and nonlinear models are constructed as follows: (1) The central cities of the four metropolitan areas are Shanghai, Nanjing, Hangzhou, and Hefei, and FAC is treated as the spillover effect variable of these central cities.

Data sources
This paper focuses on four major metropolitan areas in the Yangtze River Delta region, namely, Shanghai (including Shanghai, Ningbo, Suzhou, Wuxi, Nantong, Jiaxing, Zhoushan, Changzhou, and Huzhou), Nanjing (including Nanjing, Zhenjiang, Yangzhou, Maanshan, Chuzhou, Wuhu, Xuancheng, and Huaian), Hangzhou (including Hangzhou, Huzhou, Jiaxing, Shaoxing, Quzhou, and Huangshan), and Hefei (including Hefei, Chuzhou, Wuhu, Ma'anshan, Huainan, Liuan, and Bengbu), which cover 25 cities in total. The Su-Xi-Chang and Ningbo metropolitan areas are included in the Shanghai and Hangzhou metropolitan areas and are not considered separately in this paper. The sample data are obtained from the China City Statistical Yearbooks, China City Construction Statistical Yearbooks, Guotaian database, and provincial and municipal statistical yearbooks and span the years 2003 to 2019. To maintain data caliber consistency, all variables are created from the city district data. Those variables that lack city district data are approximated from city-wide data instead. Smoothing is performed for individual missing data.

Moran's I test
Before the spatial econometric analysis, the potential spatial correlation in the EGI of cities in the four metropolitan areas needs to be tested. For this purpose, this study employs Moran's I to calculate the spatial effects of each year under the inverse matrix of geographic distances. Table 3 shows that the Moran's I value of EGI in Shanghai, Nanjing, Hangzhou, and Hefei are all significantly negative at the 10% level from 2003 to 2019, thereby indicating that the EGI of these metropolitan areas has spatial dependence and a significant negative correlation.

Baseline regression
A spatial econometric model needs to be selected prior the spatial econometric analysis. For this purpose, the LM test, LR test, Hausman's test, SDM model simplification test, and SDM fixed effects test are conducted sequentially. The spatial Durbin model (SDM) with time-fixed and entity-fixed effects is eventually selected for the econometric analysis.
The spatial spillover effect under the point estimate of the SDM model is prone to estimation bias, whereas the direct and indirect effects obtained from partial differential decomposition can well explain the local and spatial effects of the independent variables on the dependent variable . Table 4 reports the direct and indirect effects of FA on EGI based on the inverse matrix of geographical distance.
1. Results show that in terms of local effects, the FA of each city in the Shanghai, Hangzhou, and Hefei metropolitan areas significantly contributes to EGI. The intensity of the effect is highest in Hefei and lowest in Shanghai. This observation can be ascribed to the fact that in recent years, the cities in the Hefei metropolitan area have witnessed a rapid financial development and reported a significant agglomeration effect, whereas the FA of cities in the Shanghai metropolitan area shows obvious gaps, indicating a small polycentric cluster development (Wang et al. 2022a, b). The low level of agglomeration cannot fully release the basic financial functions and agglomeration effect, and the existence of polycentric clusters will lead to regional frictions caused by bad competition when the central city absorbs resources, which ultimately weakens the local effect of FA in the Shanghai metropolitan area. The FA of cities in the Nanjing metropolitan has a significant inhibiting effect on EGI, which may be due to the relatively low level of FA of these cities. Specifically, in 2019, the FA of Nanjing was only 0.38, while that of all other cities (except for Yangzhou, with an FA of 0.11) was below 0.1. In addition, these cities are still at the rapid financial development stage, and the agglomeration effect in the entire metropolitan area is too weak to drive a synchronous production greening, social optimization, and environmental protection deepening. 2. In terms of neighborhood effects, the FA of cities within the Shanghai, Hangzhou, and Hefei metropolitan areas can significantly drive the EGI of other cities within these areas. The spillover effect of FA in the Hefei metropolitan area is the strongest, followed by that of the Hangzhou and Shanghai metropolitan areas. Such observation can be ascribed to the speed of financial development and urban green planning and construction in the Hefei metropolitan area, and the geographical distance between Shanghai and the cities in the boundary of the metropolitan area hinders the city from radiating the green development of neighboring cities. The coefficient of the FA spillover effect in the Nanjing metropolitan area is significantly negative, which inhibits the green and coordinated development of its surrounding cities. This finding can be ascribed to the large gap and the weak FA of cities in this metropolitan area. The intense competition among cities for financial resources leads to a serious resource mismatch, which prevents the spatial pattern of Nanjing's radiation-driven development of neighboring cities from opening. A very serious polarization phenomenon is thus observed, which blocks the trickle-down effect of spatial advancement from the inside to the outside. 3. The control variables show that the Shanghai metropolitan area is an intensive area for foreign investment introduction. However, the quality of such investment varies,  and the production processes fail to balance efficiency with equity, both of which are not conducive to EGI. The spatial effect of human resource quality is negative, thereby pointing to an irrational regional distribution and flow of talent. The direct and spatial effects of environmental manpower density are both negative, thereby indicating that the quality and quantity of environmental manpower investment in public services need to be improved. In the Nanjing metropolitan area, improving urban construction can strengthen local ecological and green coordination, but the "suppressing the capital and promoting the neighbors" phenomenon of environmental manpower resources density reveals a mismatch in the spatial distribution of environmental human resources in the area. Meanwhile, in the Hangzhou metropolitan area, the improvement of human resource quality and spatially rational deployment significantly contribute to the ecological greening of the local and neighboring areas. However, the level of urban construction does not generate a positive local effect, thereby suggesting that excessive urbanization has sacrificed the environment to some extent. Foreign investment also fails to trigger a positive spatial effect, and a local pollution out-migration is observed. The Hefei metropolitan area has been attracting foreign investment at a rapid pace in recent years and continuously gathered new energy, high-end equipment, and other emerging foreign investment projects, which play important spatial roles in promoting ecological greening. However, the quality of local environmental protection manpower in the area needs to be strengthened, and its spatial layout of urbanization construction warrants improvement.
Combined with the previous analysis, different financial industries and subjects serve various social objects and perform various financial functions. Therefore, the direction and effect of the concentration of banks, securities, insurance, and financial personnel on EGI in each metropolitan area should be investigated to strengthen the trickle-down effect of the diffusion of financial dominant industries. Table 5 reports the regression results of FA heterogeneity. Results show that (1) in general, the banking, securities, and insurance agglomerations are the most important driving factors of EGI in all metropolitan areas except for Nanjing, with the securities agglomeration exerting the strongest effect (followed by the banking and insurance agglomerations) and the financial personnel agglomeration exerting a negative effect. Such negative effect can be ascribed to the oversaturation of financial personnel in the Yangtze River Delta region, and although the scale is large, the quality of talents varies, which can easily lead to problems, such as mismatch of financial personnel and resources and financial inefficiency, and is not conducive to EGI; and (2) the agglomeration of banking, securities, and insurance industries in the Shanghai metropolitan area can trigger positive local and neighborhood effects, thereby indicating that the financial industry is widely and evenly distributed among the cities within the area. Moreover, the degree of perfection of the financial system and the corresponding agglomeration effect in Shanghai are stronger than those in the other metropolitan areas.
In the Hangzhou metropolitan area, banking and insurance clusters are the main forces of green and coordinated development, but securities clusters do not play a driving role mainly due to the limited number of local securities companies in the area. Positive local effects are observed across all financial sub-sectors in the Hefei metropolitan area, but only the insurance industry can generate positive neighboring effects. Accordingly, the insurance industry clustering contributes most to EGI given that the financial industry in this metropolitan area is still at a rapid growth stage. The local scope of the banking and securities industry can generate a clustering effect, which is not enough to form a spatial driving force. Meanwhile, the insurance industry, which focuses on developing people's livelihood, plays an important role in regional resource allocation. The negative impact of FA in the Nanjing metropolitan area on its EGI is mainly reflected in the agglomeration of the banking and insurance industries, whereas the securities industry does not play a role. These observations can be ascribed to several reasons. First, the obvious gap between the financial industries of cities in this metropolitan area and the uneven distribution of financial institutions lead to an inefficient resource allocation and generate negative local effects. Second, the Nanjing metropolitan area crosses provincial administrative boundaries, is bound by trade and administrative barriers, and lacks a spatial layout of financial institutions and a financial centripetal force, which trigger a serious polarization effect and a negative neighborhood effect.

Robustness test
To ensure the robustness of the regression results, three robustness tests are conducted (Table 6). In the first test, the explanatory variables are replaced. Following the subjective empowerment approach of Wang et al. (2019) in measuring FA, the three indicators of BAN are assigned weights of 0.15, the three indicators of SEC are assigned weights of 0.1, the INS indicator is assigned a weight of 0.15, and the PEO indicator is assigned a weight of 0.1. FA, BAN, and SEC are measured by their weights. After re-running the SDM model, the sign direction of the regression coefficients for the core explanatory variables does not differ from the original model, and while the coefficient magnitude and significance have slightly changed, such change does not affect the conclusions. The regression coefficients of INS and PEO do not change because they only have one secondary variable that does not change. The changes in the regression coefficients of FA, BAN, and SEC can be ascribed to the changes in the weights of their secondary variables. In the second test, a dynamic SDM model is used to solve the endogeneity problem caused by the spatial and time lagged terms of the explanatory variables. In the third test, the spatial lag term of FA by one period is used as the instrumental variable to address the associative endogeneity between FA and EGI (Han and Yang 2020). The regression results for

Action mechanism test
The previous analysis reveals that information transfer and technology spillover are important mechanisms of action of FA for EGI, and the effectiveness of these mechanisms is verified by changing the weight matrix. Table 7 reports the SDM model regression results of FA on the information distance and technology distance matrices. Comparing these results with the original geographic distance inverse matrix results in Table 4 reveals that first, the direct impact coefficient of FA significantly increases, but the spatial impact coefficient does not greatly fluctuate, thereby indicating that with the intervention of information and technology levels, the positive impact of FA on local EGI is generally enhanced but the positive impact on neighboring EGI is not significantly enhanced. The reason is that the linked development of information and innovation technologies is conducive to reducing the friction of geographical distance among cities in the metropolitan area and releasing the green-driving potential energy of FA (Yuan et al. 2019). However, such linked development does not significantly improve the  geographical friction of administrative boundaries across regions, and administrative barriers remain the biggest factor that hinders financial institutions from expanding their businesses outside urban areas. Second, the direct and spatial impact coefficients of FA in the Hangzhou metropolitan area are significantly improved, and both information transfer and innovative technology association eliminate the significant negative effect of FA on EGI. The spatial effect coefficient on neighboring areas also changes from negative to positive yet remains insignificant for the time being. These results suggest that the cities in Zhejiang Province are united and have a strong information technology linkage and that green innovation technology can achieve spillover transmission across regions and strengthen the FA effect. The direct impact coefficient of FA in the Shanghai and Hefei metropolitan areas increases but the spatial impact coefficient decreases, thereby indicating that information transfer and technology linkage strengthen the local effect of FA but weaken its neighboring effect. This finding can be ascribed to the stage of financial development in these metropolitan areas, the regional distribution of their cities, and their current development situation. The Shanghai metropolitan area contains nine cities in Jiangsu, Zhejiang, and Shanghai, with Shanghai being the regional radiation center in the east, Su-Xi-Chang being a small metropolitan area in the west, Ningbo being the main small metropolitan area in the south, and Nantong being the central radiation area in the north. The development in this metropolitan area is extremely uneven. Specifically, the linkage of information networks between cities and the province leads to poor connectivity across financial regions and triggers an obvious intra-regional competition, whereas the linkage between regions with high and low information technology levels intensifies the cannibalization of financial resources by highlevel regions and generates a siphoning effect that weakens the spatial spillover effect of FA. The uneven technology level of cities in the Hefei metropolitan area also explains the weakened neighborhood effect of FA in this area. In the Nanjing metropolitan area, the direct impact coefficient of Fa decreases, whereas its spatial impact coefficient increases, thereby indicating that the spatial association of information and innovative technology plays a key role in weakening the negative impact of FA on the EGI of neighboring areas. However, the problems arising from regional information asymmetry and irrational use of technology only aggravate the negative local effects of FA.
To explore the specific effects of FA of different vectors on EGI in the context of information transfer and innovation linkages, the breakdown of financial subsector agglomeration and financial personnel agglomeration is continued. Table 8 reports the SDM regression results of FA heterogeneity under the information and technological distance matrices. For a more intuitive interpretation of the direct and indirect effects of FA, Table 8 only shows the values of coefficients that pass the 10% significance test, and these values are replaced by " + " signs, with " + " representing values within the 0-0.1 range. Results show that first, under the mechanism of information transmission and technology association, the securities industry agglomeration remains the most important force in releasing green-driving potential, the banking industry agglomeration acts as the middle force, the insurance industry agglomeration acts as the backup force, and the financial personnel agglomeration is still not functional. Second, the local and neighboring effects of the banking and securities agglomeration in the Hangzhou metropolitan area are significantly activated, thereby indicating that the core advantage of the agglomeration effect of the financial industry lies in the linkage and transmission of information and technology among cities in the same area. The significantly lower agglomeration effect of the financial sub-sectors in the Shanghai metropolitan area further verifies that the uneven information technology level of cities in this area exacerbates the siphoning effect of high-level cities on low-level cities. These observations also highlight the importance of information and innovative technologies in small metropolitan areas for individual cities and the ineffectiveness of financial resource allocation due to intense inter-city competition. The local neighborhood effect of financial sub-sector clustering in the Hefei metropolitan area decreases under technology linkage, thereby suggesting that green innovation technology constrains the green and coordinated development of cities in this area.

Further exploration: expansion of spatial perspective
The core cities of the four metropolitan areas also serve as financial center cities in the Yangtze River Delta region. Under the policy directive of implementing ecological and green integrated development in the region, these core cities play a critical role in coordinating the distribution of factor resources in the metropolitan areas, improving economic efficiency, and commanding the green and coordinated development of cities within the region. To complement the findings of previous studies that have verified the existence of local and neighboring effects of FA on EGI, the effects of the FA of Shanghai, Nanjing, Hangzhou, and Hefei, which act as core cities in the Yangtze River Delta, on the EGI of other cities in their respective metropolitan areas are further explored. Table 9 shows the linear and nonlinear regression results of FA. Results show that first, in the linear regression model, the financial center variable is significantly positive for the Shanghai and Hangzhou metropolitan areas, significantly negative for the Nanjing metropolitan area, and insignificant for the Hefei metropolitan area, thereby indicating that Shanghai and Hangzhou play a significant radiationdriving role in the EGI of other cities within their respective metropolitan areas, with Shanghai being stronger than Hangzhou in terms of radiation strength, Nanjing having a negative effect on the EGI of neighboring cities, and Hefei failing to drive the development of greening in the surrounding cities. Second, the nonlinear regression results that include the quadratic term of the financial center variable reveal that the coefficients of both the primary and secondary terms of the Shanghai financial center variable are not significant, thereby confirming that the FA of Shanghai has no nonlinear effect on the EGI of other cities within its metropolitan area. The coefficients of the primary and secondary terms of the Nanjing financial center variable are significantly negative and positive, whereas those of the Hangzhou financial center variable are significantly positive and negative, thereby suggesting that the FA in these metropolitan areas has a "U"-shaped and inverted "U"-shaped nonlinear effects on the EGI of other cities within their respective circles. The inhibiting effect of Nanjing's FA on the green development of the surrounding cities can be eliminated after improving the level of FA, while Hangzhou should be wary of the negative impact of excessive FA on the green development of its surrounding cities.

Conclusion and policy implications
Using panel data from 25 prefecture-level cities in 4 metropolitan areas in China's Yangtze River Delta region from 2003 to 2019, this paper analyzes the direct effect, spatial effect, and intermediate mechanism of FA across different metropolitan areas on EGI by using 3 Durbin models with different spatial weight matrices and further verifies the radiation-driven ability of the central cities in these metropolitan areas.
The main findings are summarized as follows. First, the FA in the Shanghai, Hangzhou, and Hefei metropolitan areas can produce significant local and neighboring effects on EGI, with the FA in Hefei exerting the most intense effect, followed by those in Hangzhou and Shanghai. The FA in the Nanjing metropolitan area negatively affects both the local and neighboring areas. In addition, the direction and extent of the effects of foreign investment, human resource quality, environmental manpower density, and urban construction level on EGI vary from one metropolitan area to another. Second, the agglomeration of banking, securities, and insurance industries serves as the most important force in releasing the potential of FA to promote EGI. The securities agglomeration offers the strongest contribution, followed by the banking and insurance agglomerations. Meanwhile, the financial personnel agglomeration produces a negative effect on EGI. Specifically, the financial sub-sectors in the Shanghai and Hefei metropolitan areas all play an effective role, and the Hangzhou metropolitan area is dominated by banking and insurance. Third, the financial industry and subsectors clustered in each metropolitan area demonstrate different degrees of effectiveness in promoting EGI through the mechanisms of information transfer and technology linkage. The local effect of agglomeration is enhanced in general, whereas its neighboring effect fluctuates. The agglomeration effect in the Hangzhou metropolitan area significantly increases. The degree of contribution of each financial subsector is still dominated by securities and supplemented by banking and insurance. Fourth, with regard to the greening effect of FA in the central cities of the metropolitan area, Shanghai demonstrates the strongest radiation ability, Nanjing shows a "U" non-linear effect, Hangzhou shows an inverted "U" effect, and Hefei has no impact.
The above findings offer several policy implications. First, given that FA can unleash a "driving green potential," all four metropolitan areas in the Yangtze River Delta should continue building financial infrastructures, strengthen the cooperation and exchange of financial institutions across regions, and avoid vicious competition. The Nanjing metropolitan area has the lowest level of finance and the largest urban gap. The Shanghai, Hangzhou, and Hefei metropolitan areas need to strengthen their ability to select foreign and environmental personnel and optimize their allocation of human resources. The Nanjing metropolitan area also needs to improve its spatial layout of environmental protection manpower, and the Hangzhou and Hefei metropolitan areas need to optimize their spatial layout of urban construction and implement new green urbanization construction. Second, in view of the different roles played by financial industries in the EGI of different metropolitan areas, each metropolitan area should focus on developing advantageous industries and making up for their shortcomings. Specifically, improving the quality and suitability of financial talents plays a key role in optimizing the allocation of financial staffing institutions. The Shanghai, Hangzhou, and Hefei metropolitan areas should simultaneously strengthen the development of their securities, banking, and insurance industries. They need to expand their number of securities institutions to improve the convenience of financing channels, strengthen their banking industries to assume the backbone of "driving green power," relax the financing constraints while strictly screening and monitoring enterprises, actively explore the development of green insurance products, and guide the allocation of long-term and stable insurance funds to the green industry. The Nanjing metropolitan area should pay attention to the regional coordination of its financial industry allocation, the "green driving potential" of securities and personnel into positive, and then overcome the negative impact of banks and insurance. Third, in view of the mechanisms of information transfer and technology linkage, each metropolitan area should strengthen its information exchange and crossregional cooperation among financial institutions, set up an open platform for the exchange of institutional financing and innovative technology, and strengthen the active role of information transfer and technology sharing in releasing the local and neighboring effects of FA. Fourth, in view of the radiation capacity of different central cities, Nanjing and Hangzhou should accelerate their financial infrastructure construction to fully transform the polarization effect into a trickle-down effect and play a radiation-driven role as early as possible. Hefei should not only rapidly increase its financial volume but also focus on its coordinated development with neighboring cities, strengthen its exchanges and cooperation, and avoid the polarization effect.