Impact of green nance on carbon intensity-empirical research based on dynamic spatial Durbin model

3 Abstract 4 Green finance is of great significance in improving the ecological environment and 5 achieving the purpose of energy conservation and emission reduction. In order to 6 explore the influence of green finance on carbon intensity, four indicators of green 7 credit, green securities, green insurance and green investment are adopted to construct 8 the green finance development index in this paper. Based on the panel data of 30 9 provinces in China from 2009 to 2019, a dynamic spatial Durbin model is constructed 10 and the method of partial differential matrix is selected to analyze the influence of green 11 finance on carbon intensity in the short and long terms. The empirical results show that 12 (1) the development of green finance in local area has positive effect on the reduction 13 of carbon intensity. (2) with the significant spatial spillover effect on carbon intensity, 14 green finance can reduce the carbon intensity of the adjacent area and promote the 15 development of low-carbon economy. (3) dynamic test results prove that in terms of 16 direct effect and spatial spillover effect, green finance has a greater long-term effect on 17 carbon intensity.


Introduction
Human survival has been seriously threatened by global warming caused by  China, as the world's largest CO2 producer, has been actively promoting the 24 development of low-carbon economic to reduce carbon emission . China 25 has proposed to reach the CO2 emission peak by 2030, and achieve the carbon neutrality 26 by 2060, indicating that the ecological civilization construction of China will focus on 27 carbon reduction to promote a comprehensive green transformation on economic and 28 social development (Yang 2016;Tang 2021). 29 As a policy framework system for environmental protection, green finance  Results showed that the improvement of the green finance development index and the 69 increase in non-fossil energy utilization contributed to the reduction of carbon emission 70 intensity. Gianfrate and Peri (2019) believed that the issuance of green bonds by 71 governments was important to mobilize financial resources for the achievement of 72 carbon reduction targets. Glomsrød and Wei (2018)  Therefore, in this paper, green finance development at regional level is researched.

93
Based on four dimensions and six indicators of green credit, green securities, green 94 insurance and green investment, data from 30 provinces in China are selected to 95 construct a comprehensive evaluation system of green finance. The concept of carbon 96 intensity is adopted to represent the carbon emission per unit of GDP. In addition, with 97 the time lag and spatial lag characteristics of carbon emission, the dynamic spatial 98 Dubin model is adopted to measure the impact of green finance development on carbon 99 intensity of local and adjacent areas from both short-term and long-term perspectives. intensities, industry has higher productivity efficiencies, indicating that the industry has 105 entered a mode of low-carbon economic development.

106
Carbon intensity is defined as follows:  The entropy weighting method is adopted to select 6 indicators in 4 dimensions of 124 green credit, green securities, green insurance and green investment for the construction 125 of a comprehensive green finance evaluation system in this paper.

127
(1) Regional economic development level (Redl). The demand of people for 128 environmental quality and the awareness of environmental protection will increase with 129 the improvement of living standards, resulting in a reduction of local carbon intensity 130 (Luo et al. 2017). In this paper, per capita GDP is selected for the measurement of the 131 regional economic development level.

132
(2) Urbanization level (Url). As urbanization rates increase, natural gas is 133 gradually replacing coal as the main energy consumed, and thus reducing carbon 134 intensity. In this paper, the ratio of urban population to total population is adopted to  Table 1.

262
The following spatial dynamic Durbin model is constructed based on the above 263 variables.
where Cit-1 is the time lagged term of carbon intensity; WCit-1 is the time and spatial 265 lagged term of carbon intensity; WCit is the spatial lagged term of carbon intensity; ρ is The regression results present the spatial autoregressive coefficient ρ is 0.101, 279 which is significant at the 10% level. This suggests that the increase of local carbon 280 intensity can lead to an increase in carbon intensity of adjacent areas. In addition, with 281 the characters of time lag and spatial lag, local carbon intensity is influenced by that of 282 the local and adjacent regions in the previous period. As a result, carbon intensity has a 283 certain "cumulative" effect. The regression results also show that the estimated 284 coefficient for the impact of green finance on carbon intensity is -0.396, which is 285 significant at the 1% level, indicating that the development of green finance can reduce 286 carbon intensity. The results of spatial spillover effect are crucial to the spatial Durbin 287 model. Therefore, the spatial spillover effect is analyzed and divided into long-term 288 effect and short-term effect in this paper.

289
According to Elhorst (2014), the basic form of the dynamic space Durbin space The above equation can be translated into the following form： According to the method of Elhorst (2014), the direct effect and spatial spillover 293 effect of X on Y can be solved by partial differential matrix operations. Compared with 294 the static spatial Durbin model, which has only long-term effect, the dynamic spatial 295 Durbin model has both short-term and long-term effects. As a result, at a particular time 296 t, the matrix of partial derivatives of the expected value Y corresponding to the values 297 of X from spatial units 1 to N can be written as: where the average value of diagonal elements is short-term direct effect, while the 299 average value of row sum or column sum of non-diagonal elements is short-term spatial 300 spillover effect, representing the influence of X in a specific spatial unit on Y in other 301 spatial units (Lesage 2009   and 1% levels, respectively. 308 According to the above theory of partial differential matrix operation, the direct 309 effect and spatial spillover effect of green finance on carbon intensity in the short and 310 long term can be calculated, and the results are provided in Table 6. It can be seen from 311 Table 6 that:

312
(1) The decomposition results show that the direct effect, spatial spillover effect 313 and total effect of green finance show a significant negative correlation with carbon 314 intensity in both short term and long term, indicating that the development of green 315 finance can reduce the carbon intensity in both local and adjacent areas.

316
(2) The direct effect of per capita GDP is negative in the short term, and the spatial 317 spillover effect is not significant. However, in the long term, both the direct and indirect

351
Geographical distance spatial weight matrix is adopted in this paper to test the 352 robustness of the result. The regression results show that the dynamic SDM model has 353 the best fitting effect with the utilization of geographic distance spatial weight matrix.

354
There is a significant negative correlation between green finance and carbon intensity,  Table 7. spatial Durbin model is constructed and the method of partial differential matrix is 364 adopted to analyze the influence of green finance on carbon intensity. The main 365 conclusions are as follows:

366
(1) The development of green finance has spatial spillover effect, and can reduce 367 carbon intensity in local and adjacent areas.

368
(2) Green finance has a more significant influence on the direct effect and spatial 369 spillover effect of carbon intensity in the long term.

370
(3) Economic development, industrial structure and foreign investment all have 371 negative influence on carbon intensity. However, the spatial spillover effect of these 372 factors is not obvious.   Competing interests The authors declare that they have no conflict of interest. 411