Coupling refers to the phenomenon by which two or more systems influence each other through interactive mechanisms, and the coupling degree reflects the degree of mutual dependence and mutual restriction among systems (Zhang et al., 2023). However, it cannot reflect the coordinated development level among systems (Sun and Cui, 2018). For example, the coupling degree between GSTIE and DEL is high while the overall level of development is low, which is not the best state of high-quality development. CCD is used to reflect the degree of mutual balance and coordination between different systems (Dong et al., 2021; Yang et al., 2022a). Because traditional CCD has errors in writing, coefficient, weight and model, Wang Shujia and others revised the model, which improves the scientificity and rationality of the research (Wang et al., 2021). Therefore, this paper applied the revised CCD model to assess the coupling coordination development level of the two subsystems of GSTIE and DEL in China’s provinces. The specific calculation formula of the revised CCD model is: C = 1 − U 2 − U 1 × U 1 U 2 D = C × T T = α 1 U 1 + α 2 U 2 (3) where C represents the coupling degree, and C ∈ 0,1 . The greater the C value is, the stronger the interaction between the two systems. U 1 represents the comprehensive evaluation index of the GSTIE system; U 2 represents the comprehensive evaluation index of the DEL subsystem, and max U i is assumed to U 2 ; T indicates the comprehensive coordination index of GSTIE and DEL, α , β is the undetermined coefficient, and α + β = 1 . In addition, it is generally believed that GSTIE is as important as DEL, and therefore α = β = 1 2 ; D refers to the CCD, and D ∈ 0,1 . Specifically, the larger it is, the more harmonious and balanced the GSTIE and DEL subsystems are. In this research, after referring to relevant research (Liu T. et al., 2021), we have divided the CCD into five levels by using the median segmentation method, as shown in Table 2.

The spatial autocorrelation model can observe the degree of aggregation in the study area of an attribute value, mainly including global spatial autocorrelation and local spatial autocorrelation (Yang et al., 2022b). This study applied global spatial autocorrelation to reflect the global spatial association pattern of CCD of the GSTIE and DEL subsystems in the whole research area to determine whether there is a spatial correlation in the space. Global Moran’s I is generally used for the calculation. The specific calculation formula is: I = n × ∑ i = 1 n ∑ j ≠ 1 n W i j ( x i − x ¯ ) ( x j − x ¯ ∑ i = 1 n ∑ j = 1 n W i j × ∑ i = 1 n x i − x ¯ 2 (4) where I denotes Moran’s I, n represents the number of provinces and cities, x i   a n d   x j are the CCD of i province and j province, respectively, and x ¯ is the average value of the CCD; and W i j represents the space weight matrix. Moran’s I value ranges from −1 to 1, with I > 0 indicating a positive spatial correlation. The larger the value is, the more significant the spatial correlation is, with I < 0 indicating a negative spatial correlation. The smaller the value is, the greater the spatial difference is, with I = 0 indicating that there is a randomness of the space.

As the global spatial autocorrelation analysis can only reflect the overall situation of spatial correlation and difference of the study area and cannot sufficiently reflect the coupling and coordinated relationship between local elements and adjacent elements (Yang et al., 2022c; Li et al., 2022). The local spatial autocorrelation analysis can effectively identify the spatial dependence and heterogeneity of CCD. The calculation formula is: I i = x i − x ¯ m 0 ∑ j w i j x j − x ¯ (5) where x i denotes the CCD value in i province, and x ¯ refers to the average CCD value in all provinces; I i > 0 represents the spatial agglomeration (H-H or L-L) with similar coupling coordination values of a province or city, and I i < 0 represents the spatial agglomeration (L-H or H-L) with different coupling coordination values of a province or city.

On the basis of the perspective of configuration, by utilizing the fsQCA method, this study explored how the seven factors—human capital, R&D investment, openness, industrial structure, economic development, environmental regulation, and government subsidies—interacted to jointly affect the coupling and coordinated development between GSTIE and DEL in each province and identified the condition configuration that can achieve the coordinated development of the two subsystems. Unlike traditional statistical analysis and qualitative analysis methodologies, QCA can analyze the complex causal relationship between different combinations of antecedents and results (Liu et al., 2017). As one of the QCA techniques, fsQCA is a novel research methodology between variable and case orientation and goes beyond qualitative and quantitative research (Guo, and Lin, 2022). This method can be effectively applied to explore the synergy and linkage between various factors and can identify different causal paths leading to the same result by comparing and analyzing multiple causal conditions in each case (Mercado-Caruso et al., 2022).

In this study, fsQCA was applied to identify the factors influencing the provincial CCD, which was mainly based on the following considerations. The coupling coordinated development between GSITE and DEL is comprehensively affected by multiple factors, conforming to the basic assumption of fsQCA. Moreover, traditional research methods are mainly applicable for analyzing the net effect of a single influencing factor, whereas fsQCA can explore the complex non-linear and asymmetric causal relationship between multiple factors and results (Sun et al., 2022). In this paper, 30 provinces are taken as research samples, with small sample size, and fsQCA is the best method to perform a comparative analysis of small and mid-sized samples (Yifan and Bei, 2022). Furthermore, in this study, we aim to investigate the influence of different combinations of seven factors of the CCD between GSITE and DEL, as well as the multiple driving paths behind the coordinated development of the two subsystems, and fsQCA can effectively identify the equivalent paths that have different antecedent condition configurations leading to the same result; therefore, it is more suitable to explore the differentiated driving paths of the coordinated development of GSTIE and DEL in various areas.

The measurement results presented in Table 3 indicate that during the research period, the overall level of GSTIE across the country presented an overall upward trend, and the efficiency value rose from 0.684 at the beginning of the period to 0.734 at the end of the period, with an increase of 7.35%. The overall level was high, and the change range was small. This indicates that the development of green science and technology innovation, as an increasingly important topic of national green innovation, has been accelerating in recent years, and its role in promoting energy conservation, emission reduction, and sustainable economic development has become increasingly significant.

To reflect their spatial distribution differences more intuitively, ArcGIS software has been utilized to visualize GSTIE in representative years, as shown in Figure 1. As observed in Figure 1, the spatial change characteristics of GSTIE in 2013, 2015, 2017, and 2019 were relatively similar, showing a spatial distribution pattern of a low level of GSTIE in western provinces and cities and a high level of GSTIE in central and eastern provinces in China. During the investigation period, the efficiency values of GSTIE in Beijing, Jiangsu, Zhejiang, and Qinghai remained unchanged at the level of 1.00, whereas the efficiency values in other provinces and cities fluctuated. Specifically, in 2013, the efficiency values of GSTIE in Hebei, Fujian, Shanxi, Jilin, Henan, Inner Mongolia, Guangxi, Xinjiang, and other regions were below 0.40. By 2015, the efficiency value of Yunnan dropped to below 0.40, and the efficiency values of Liaoning, Anhui, and Heilongjiang rose from below 0.60 to above 0.60. In 2017, the efficiency values of Shanxi, Jilin, and Guangxi broke through the low-efficiency level of below 0.40, and the efficiency values of Jilin and Guangxi reached 1.00, whereas the efficiency values of Hainan and Ningxia decreased from 1.00 to low-efficiency level of below 0.40. By 2019, the efficiency values of Guangxi and Chongqing dropped from 1.00 to below 0.40, those of Hebei, Hainan, and Xinjiang rose from below 0.40 to above 0.40, and those of Hebei and Xinjiang reached 1.00.

As shown in Table 3, from 2013 to 2019, the overall DEL of China showed a decreasing trend, decreasing from 0.2757 in 2013 to 0.2658 in 2019, a decrease of 3.59%, and the average annual decline rate was around 1.00%, suggesting that the development of China’s digital economy lagged behind the development of green technology innovation during the investigation period. From the regional perspective, there was a regional imbalance in DEL, and during the investigation period, DEL in the eastern region was always higher than DEL in the central and western regions, and even far higher than the national average level. This is mainly due to the fact that areas in the east have a relatively superior economic foundation and technological R&D, which provides a good growth environment for the development of the digital economy, and therefore DEL is the highest in this region.

From the perspective of the spatial evolution characteristic of DEL (see Figure 2), China’s DEL showed a similar spatial distribution pattern throughout the study period, and in the four representative years, DEL shows a decreasing trend in the eastern, central, and western regions in succession. During the research period, the DEL of provinces in China was generally at a low level, and there was much room for improvement.

Provinces and cities with high DEL are mainly in Guangdong in the eastern region, while provinces and cities with low DEL are mainly located in Xinjiang, Qinghai, Ningxia, and Gansu in the western region and Jilin in the central region. On the whole, in the four periods, Beijing, Jiangsu, Zhejiang, Shandong, and Guangdong in the eastern region occupy the dominant position in DEL. DEL in these provinces and cities has always been at the forefront of the country. However, for the remaining provinces, DEL is at or below 0.40, and around half is at a low level of 0.20∼0.40.

As can be seen from the calculation results, the average CCD of each province has been classified into five grades, as shown in Table 4. The results revealed that there were certain differences in the CCD among provinces, and most provinces and cities were concentrated in the moderate and high coordination stage, with 22 in total, of which there were 15 high coordination provinces and 7 moderate coordination provinces. High-coordination provinces and cities have the potential to develop into high-quality coordinated provinces and cities, and there is much room for improvement. Only Beijing, Jiangsu, Zhejiang, and Guangdong have been in the stage of high-quality coordination. Inner Mongolia, Hainan, Ningxia, and Xinjiang were in the basic coordination stage; however, there were no low-coordination provinces. On the whole, from 2013 to 2019, the average CCD between GSTIE and DEL in each province was above the level of medium coordination, indicating there has been a good development trend, and mainly showed the distribution situation of “more in the middle, less at both ends,” that is, the number of provinces in the middle state of moderate coordination and high coordination was the highest and there were fewer provinces at both ends of basic coordination and high-quality coordination. In the future, we need to strengthen the relevant construction of GSTIE and DEL of basic and moderate coordination provinces and cities, improve their coupling and coordination level to improve the overall level of CCD in the whole country, and let more provinces and cities develop towards the stage of high coordination and high-quality coordination. In addition, more attention should be paid to the sustainable development of high-coordination provinces so that these provinces can develop into high-quality coordination provinces as soon as possible.

This study selected the starting and ending years and the middle years in the research period, namely, the values of CCD in 2013, 2015, 2017, and 2019, and utilized ArcGIS software to conduct spatial visualization processing to further explore the inter-provincial spatial distribution characteristics of the CCD of GSTIE and DEL (Figure 3).

As illustrated in Figure 3, overall, the national CCD between GSTIE and DEL showed a steadily increasing trend (Figure 3). The spatial distribution of the coupling and coordinated development between GSTIE and DEL was unbalanced, showing the distribution pattern of east > west > central. During the investigation period, the spatial pattern of the CCD of China’s provincial GSTIE and DEL changed significantly, and the CCD in most regions was at or above the level of moderate coordination, which indicated that GSTIE in most provinces and cities of China was well coordinated with DEL. Specifically, in 2013, half of the provinces were in the moderate coordination stage, five provinces were in the stage of high-quality coordination, five provinces were in the stage of high coordination, four provinces—Jilin, Hebei, Ningxia, and Hainan—were in the basic coordination stage, and only Xinjiang in the west was in the low coordination stage and remained unchanged until 2019 when it moved from the low coordination stage and entered the medium coordination stage. In 2015, the number of high-coordination provinces increased by five: Hebei, Shandong, Hainan, Chongqing, Anhui, and Heilongjiang. At the same time, Guizhou moved from the highly coordinated stage and entered the moderate coordination stage. In 2017, there was relatively little change in the spatial distribution pattern of coupling and coordination. Guangxi had two periods of moderate coordination and then entered the high coordination stage. Jilin left the basic coordination stage and entered the moderate coordination stage. In 2019, the number of excellent coordination provinces increased from four to six—Beijing, Jiangsu, Zhejiang, Guangdong, Sichuan, and Shandong—Hebei and Jiangxi rose from a moderate coordination stage in 2017 to a high coordination stage, and Guangxi and Chongqing dropped to the stage of moderate coordination and basic coordination, respectively. At the same time, Xinjiang and Ningxia left the low coordination stage, and there was no low coordination province, which suggested that the CCD of GSTIE and DEL in the two provinces had been improved to a certain extent, and the relationship between the two had become slightly closer.

This paper drew the point map and ridge map of CCD of GSTIE and DEL subsystems of provinces in China (see Figure 4) to further explore the regional distribution characteristics of CCD between the two subsystems.

In Figure 4, it can be seen that there was an imbalance in the coupling and coordinated development between GSTIE and DEL among regions. The regional CCD from high to low was: East China, Central China, South China, Southwest China, Northeast China, North China, and Southwest China, and for every region, there is still substantial room for improvement of the coupling coordination relationship between GSTIE and DEL.

According to the distribution of CCD between GSTIE and DEL among provinces in each region, there was a big difference in the CCD of the provinces and cities in East China, of which the CCD in Jiangsu and Zhejiang provinces was the largest, and the level of CCD in the two provinces fluctuated slightly every year, maintaining the same level. Furthermore, there was a big difference in the CCD among the provinces in North China. The CCD in Beijing fluctuated slightly every year and remained at a high level of 0.80–0.90, and the level of the CCD was the highest in this region and the lowest in Inner Mongolia. There was a small difference in the CCD in the four provinces and cities of Southwest China, among which Sichuan had a higher level than that the other three provinces in this region. Conversely, there was a big difference in the CCD among provinces in Northwest China. The CCD in Ningxia fluctuated sharply every year, followed by Xinjiang, and the level of coupling coordination in the other three provinces showed small fluctuations every year. There was a large difference in the CCD between GSTIE and DEL in the three provinces of Southern China, among which the CCD in Guangdong had a small fluctuation annually and was in the highest position and far higher than that in Guangxi and Hainan, with a maximum gap of 0.76. The CCD was relatively different in the three provinces of Central China, and the CCD in Henan was the lowest, but it was rising as a whole. The CCD in Hubei grew slowly during the investigation period and exceeded that of Hunan. There was a small difference in the CCD among provinces in Northeast China, among which Liaoning and Heilongjiang had small fluctuations in the CCD every year.

Before using fsQCA for analysis, variables need to be classified into two categories: result variables and conditional variables. With reference to the research of Luo L. et al. (2022), all the indicator data included in the variables are the average value of relevant statistical data from 2013 to 2019 (Luo L. et al., 2022).

The coupling and coordinated development between GSTIE and DEL can be affected by a variety of factors. Based on previous research and the actual situation, human capital, capital investment, openness, economic development, industrial structure, government subsidies, and environmental regulation were determined as the conditional variables. The full-time equivalent of R&D personnel is a universal indicator that is applied for the measurement of human capital (HC). Therefore, in this study, the full-time equivalent of R&D personnel of each province was selected as the specific evaluation indicator of human capital. R&D investment is commonly used as an indicator reflecting the financial resource investment in technological innovation, and considering the difference in the scale of economic development in various regions, the proportion of internal R&D expenditure in the gross domestic product (GDP) was selected for the measurement of R&D investment (RD). In this study, the amount of foreign direct investment (FDI) was selected as the proxy variable of the level of openness (OP) (Song et al., 2021). GDP per capita was chosen for the measurement of the economic development (ED) level (Yang et al., 2021). The industrial structure (IS) was measured using the proportion of the value of the tertiary industry value to GDP as a specific measure (Hao et al., 2020). The index of “the proportion of government funds in the internal expenditure of R&D funds” was selected to represent the strength of government subsidies (GS). Environmental regulation (ER) represents the intensity of local environmental regulation in a region. To quantitatively evaluate environmental regulation, this study selected the proportion of industrial pollution control investment in the secondary industry as the proxy variable (Li et al., 2020). The descriptive statistics of all the variables are presented in Table 6.

Calibration, the first step of fsQCA, refers to the process of assigning set membership to a case (Huang et al., 2022). This study calibrated the relevant variables to a fuzzy set by using the direct calibration method. Referring to existing studies (Romero-Castro et al., 2022), the 75%, 50%, and 25% quantile value of the resulting variable and the conditional variable were selected as the three qualitative anchor points of full subordination, intersection, and non-subordination, respectively. The calibration results are presented in Table 7.

The results presented in Table 9 reveal that five configurations could generate a high level of CCD between GSTIE and DEL, and the consistency values of these configurations were above the level of 0.90, indicating that all five configurations were sufficient conditions for high CCD between the two subsystems. The consistency level of the solutions was 0.9263, indicating that all the configurations covering most cases were sufficient for high CCD. In addition, the overall coverage of the solutions was 0.6712, which indicated that all the configurations can explain around 67.12% of the cases with high CCD. Configuration H1a had the same core conditions as configuration H1b, and they can form a second-order equivalent configuration. Accordingly, this paper summarized four configuration paths that can produce the high CCD between GSTIE and DEL.

(1) HC + RD + OP-jointly driven. In configuration H1 (H1a and H1b), the core conditions are HC, RD, and OP, so it was named HC + RD + OP-jointly driven. Configuration H1a showed that the conditional configuration with high HC, high RD, high OP, and non-high ER as core conditions and non-high GS as the edge condition can generate high CCD between GSTIE and DEL. H1a indicated that the areas with insufficient GS and weak ER, relying on rich HC, high RD and high OP, and through independent R&D and introduction and absorption, can achieve high CCD between GSTIE and DEL. In this configuration, the values of consistency and unique coverage were 0.8988 and 0.1765, respectively, which indicated that the configuration could explain around 89.88% of the cases, and around 17.65% of the cases should be interpreted only by this configuration path. The provinces and cities represented by this route include Jiangsu, Zhejiang, Anhui, Guangdong, Hunan, and Hubei. Taking Guangdong as an example, the original driving data showed that the full-time equivalent of R&D personnel in this province ranked first in the whole country, with prominent advantages in HC, and RD and OP ranked among the top five in the country. However, Guangdong was ranked lowest in the country in terms of GS and ER. This showed that the provinces and cities represented by Guangdong, under low GS and weak ER, relying on high HC, high RD and high OP, can drive the coupling coordinated development between GSTIE and DEL, which conformed to the typical characteristics of the configuration solution of HC + RD + OP-co-driven. In configuration H1b, the core conditions included high HC, high RD, high OP, and non-high ER, and the marginal conditions were non-high ED and non-high IS, which indicated that if the level of regional ED is not high, IS is not advanced and ER is not high, it could realize the coordinated development between GSTIE and DEL through the combination of high HC, high RD, and high OP. In this configuration, the consistency value was around 0.9863, and the unique coverage was around 0.0512, meaning that the configuration path could explain approximately 98.63% of the cases, and around 5.12% of the cases could be interpreted only by this configuration path. Provinces with such typical characteristics included Anhui and Sichuan. Taking Sichuan as an example, the original data showed that ED level, advanced IS and the intensity of ER of Sichuan were in the bottom 50% of the country, but in terms of HC, RD, and OP, it ranked in the top 50% of the country, and the CCD of Sichuan ranked sixth in the country. This showed that the provinces represented by Sichuan mainly relied on the combination of high HC, high RD and high level of OP to produce the high level of coupling coordination development between GSTIE and DEL, which conformed to the typical characteristics of the configuration solution of HC + RD + OP-jointly driven.

This study also tested the configurations that produced non-high CCD between GSTIE and DEL, and the results are presented in Table 10. The results showed that six configurations generated non-high CCD, the consistency of the solution was 0.9039, and the coverage was 0.6194, indicating that the interpretation of these six configurations was 90.39% and explained around 61.94% of the cases with non-high CCD.

NH1 showed that non-high RD, non-high OP, non-high IS and non-high ER were the core conditions, indicating that in areas with low RD and OP levels, if the IS was not advanced and ER was weak, even if the HC level was low, it would be impossible to achieve high CCD of GSTIE and DEL. NH2a and NH2b had the same core conditions, namely, non-high RD and non-high OP. NH2a showed that irrespective of whether ER was strong, the lack of high HC, high RD, high OP, high ED, and high IS would not produce high CCD even if there was high GS. NH2b showed that the lack of high HC, high RD, high OP and high ER would not achieve high CCD even though DE, IS and GS were high. In NH3, non-high OP, non-high IS, non-high GS, and non-high ER were the core conditions, and high HC, high RD and high ED were the marginal conditions. NH3 showed that areas with a low level of OP and weak ER, where the IS was not advanced enough and GS were insufficient, even if the ED level was high, and even with a high level of HC and a high level of RD, high CCD between GSTIE and DEL would not be achieved. In NH4, non-high RD and non-high OP were the core conditions, and non-high HC, non-high ED, and high GS were the marginal conditions, which showed that in areas with low levels of OP and ED, the lack of adequate RD and HC, even with high GS and high ER, would still lead to non-high CCD between GSTIE and DEL. NH5 showed that the lack of high RD, high ED, high IS and high GS would eventually lead to non-high CCD between GSTIE and DEL even if HC, OP and ER were high.