Research on the configuration paths of Sustainable Marine Eco-Efficiency: Based on fuzzy-set qualitative comparative analysis of 11 coastal areas in China

2.1 fuzzy-set Qualitative Comparative Analysis

This study tries to analyze the land-based influencing conditions of MEE from the configuration perspective, which makes the fsQCA a suitable method for carrying out our empirical test. QCA was proposed by Ragin in the 1980s. By using QCA, researchers can find out the logical relationship (for instance: what kind of configuration formed by certain condition variables may lead to the emergence or disappearance of certain outcome?) between the matching patterns of different conditions and the outcomes through cross cases comparison, so as to further identify the synergistic effects of multiple condition variables on the premise of recognizing the causal complexity [14].

Comparing with quantitative research based on traditional regression analysis, the advantages of QCA are as follow [16]: (1) Through cross cases comparison, researchers can ensure the external promotion of empirical results to a certain extent on the basis of identifying the mechanism of condition variables. (2) Researchers can identify configurations with equivalent outcomes, which can help to understand the differential driving mechanism of outcomes in different case scenarios, and further discuss the matching relationship among condition variables. (3) Researchers can further compare the configurations that lead to the emergence or disappearance of outcomes, and broaden their theoretical interpretation dimension of specific research problems. Because under the logical premise of causal asymmetry, the condition variables that lead to the emergence of the outcome variables may be different from the condition variables that lead to the emergence of the "non-set" of the outcome variables.

QCA includes three basic categories: clear set qualitative comparative analysis (csQCA), fuzzy set qualitative comparative analysis (fsQCA) and multivalued set qualitative comparative analysis (mvQCA). csQCA and mvQCA are only suitable for dealing with category problems, while fsQCA can further deal with problems related to continuous changes and partial subordination [14][16]. Therefore, fsQCA has been widely used in empirical research in recent years.

2.2 TOE framework

TOE framework is mainly applied to the analysis of new technology adoption and its influencing factors, and a variety of analysis frameworks have been developed. Among existed analysis frameworks, the TOE framework proposed by Tornatzky and Fleischer in 1990 has the most extensive impact, which puts forward three factors affecting the adoption of organizational innovative technology [17–19]: Technology refers to the characteristics of various technical means and their relationship with the organization; Organization mainly includes organizational structure, organizational scale, system regulations, organizational resources, etc; Environment mainly refers to the industry structure, external pressure, institutional environment, etc. Recently, TOE framework is widely used in QCA analysis because it can help QCA users refine condition variables and construct configuration models from technology, organization and environment.

This study takes MEE as the outcome variable in the context of green technology application in coastal areas. The condition variables selected in this study include: (1) green technology innovation and green technology capability in the technology condition, (2) urbanization and environmental regulation intensity in the organization condition, and (3) peer competition pressure and openness in the environment condition. With the outcome variable and condition variables, we build a configuration model as shown in Fig. 1. The names and abbreviations the variables involved in the model are shown in Table 1.

Marine Eco-Efficiency. There is no unified definition of MEE in the present. Schaltegger and Sturm (1990) put forward the concept of ecological efficiency and defined it as the ratio of added value to increased environmental impact [20]. World Business Council for Sustainable Development (1995) defined ecological efficiency as meeting human needs and improving the quality of life by creating products and services with price competitive advantages, while controlling its environmental impact and resource utilization intensity within the earth's carrying capacity level [21]. The organization for economic cooperation and development (1998) extended this concept to governments, industrial enterprises and other organizations, and held that ecological efficiency refers to the efficiency of using ecological resources to meet human needs [22]. Although different institutions and scholars have made different definitions of ecological efficiency, they have reached a consensus on the basic connotation, that is, ecological efficiency is "creating more value with less impact" or "obtaining more benefits from less resources" [23–24]. Following this connotation, this study considers that MEE refers to the maximization of economic output and the minimization of environmental pollution with the least consumption of marine resources.

Technology Condition. Green Technology Innovation (GTI) and Green Technology Capability (GTC) are taken as two secondary conditions in the technology condition. (1) The sustainable configuration path of MEE is a concept related to the high-quality development of marine economy, and the driving force of high-quality economic development lies in technological innovation. Thus, from the perspective of sustainable development, to effectively improve MEE needs to rely on GTI [25–26]. (2) It is worth noting that possessing green technologies is different from the effective application of green technology. Technical capability is the ability to use technical knowledge effectively [27]. If technology innovation or technology progress is to have an effect on economic development, it must take the form of products [28]. From this point of view, if we want to make GTI work effectively on MEE, the GTC to transform GTI into products will be needed.

Organization Condition. Urbanization (URB) and Environmental Regulation Intensity (ERI) are taken as two secondary conditions in the organization condition. Most technology innovations occur and gather in cities [29]. (1) The advantages of cities in specialization and diversity, accumulation of human capital, formation of information exchange network and improvement of transaction efficiency are conducive to technological innovation [30]. Thus, cities have technological advantages and scale effects in improving resource utilization and pollution control. In this case, the higher the URB of coastal areas, the more likely it is to carry out green technology innovation and have higher green technology capacity. (2) Unexpected output is an important part of MEE accounting, which refers to marine environmental pollution. Due to the externality of marine environmental pollution, it is difficult to effectively solve this problem by relying solely on market mechanism, so that environment regulation has become an important means to make up for market failure and solve environmental problems [31]. Environment regulation not only increases the production cost of enterprises, but also forces enterprises to carry out green technology innovation [32–33]. Therefore, the ERI not only reflects the attention of governments in coastal area to the marine environment, but also is closely related to the innovation and application of green technology, which has an important impact on MEE.

Environment Condition. Peer Competition Pressure (PCP) and Openness (OPEN) are taken as two secondary conditions in the environment condition. (1) From the perspective of intergovernmental relations, the competitive pressure between governments at the same level constitutes an important environment condition [34]. Under the institutional background of China, the central government personnel appointment system based on the relative performance appraisal of officials profoundly shapes the behavioral logic of local governments [35]. When facing the pressure brought by superior task (such as the construction of marine ecological civilization), there will be competition between governments at the same level, and the winner will receive additional incentives. Thus, PCP drives the local government to improve the performance on the corresponding task [36]. (2) The openness of a region affects its MEE. Openness may bring technology diffusion effect, promote the flow and matching of green innovation elements, and promote product and technology innovation [37]. However, Openness may also lead to pollution transfer and the emergence of a "pollution paradise" [38].

2.3 Variables and Data Source

2.4 Data Calibration

In fsQCA method, calibration refers to the process of assigning collective membership to cases [44]. Specifically, researchers need to calibrate variables into sets according to existing theoretical knowledge and case scenarios. The membership degree of the calibrated set will be 0 ~ 1, and the calibration points include “fully in”, “crossover” and “fully out”. Due to the lack of clear theoretical concepts and external standards as the calibration basis for MEE and land-based conditions and, this study calibrates based on case descriptive statistics [14]. Referring to the existing research [45], calibration points in this study are set as the upper quartile, the lower quartile and the mean value of sum of the upper and lower quartiles of the sample data respectively. The calibration points of each variable are shown in Table 6. 

This study uses fsQCA3.0 to analyze the data. Since the crossover value of PCP in Guangxi is exactly 0.5 after calibration, according to the "partial subordination" of this crossover value, we adjust 0.5 to 0.501 [45]. Table 7 is the truth table or each variable after calibration.

3.1 Necessary Condition Analysis

Before analyzing the configuration, the necessity of each condition needs to be tested separately. Researchers need to analyze the sufficiency of each condition that cannot be used as a necessary condition alone, and identify the configuration with the greatest explanatory power to the target case through "Boolean algebra minimization". The results of necessary condition analysis are shown in Table 8. 

The necessary condition analysis results of high MEE show that the consistency of each condition is lower than the critical value of 0.9, which indicates that each condition alone cannot constitute the necessary condition to explain the result variable. The necessary condition analysis results of low MEE show that the consistency of each condition is lower than the critical value of 0.9, which indicates that each condition alone cannot constitute the necessary condition to explain their own result variables. The necessary condition analysis results show that the land-based impact on MEE is complex. The land-based technology, organization and environmental conditions need mutual linkage and matching to jointly affect MEE.

3.2 Configuration Analysis

Configuration analysis is the process of exploring whether the set composed of multiple conditions is a subset of the result set. During parameter setting, in order to distinguish whether the configuration passes the consistency of fuzzy set theory, the consistency threshold is set to 0.8 [14]. Considering that the setting of frequency threshold should include at least 75% of observed cases, the case frequency threshold is set to 1 for there are 11 cases in this study [16]. Proportional Reduction in Inconsistency (PRI) can effectively reflect the extent to which a particular truth table row is y rather than ~ y subset, which is suggested to be set to 0.75 [46]. Through the configuration analysis of high MEE and low MEE, their complex solution, parsimonious solution and intermediate solution are obtained respectively. Since the configuration analysis results of fsQCA mainly observe the parsimonious solution and intermediate solution, we use Table 9to show the results. 

In order to better present the results, we adopt the presentation form of QCA analysis results proposed by Ragin and Fiss (2008) [47], which can clearly show the relative importance of various conditions in each configuration. The results are shown in Table 10, in which  indicates the present of the core condition,  indicates the present of the supplementary condition,  indicates the absent of the core condition, indicates the absent of the supplementary condition, and “blank” indicates that the present or absent of the condition is insignificant for outcome. Since the parameters of each configuration are already shown in Table 9, they are not repeated in Table 10. Table 10. Configurations of High MEE and Low MEE

Table 10. Configurations of High MEE and Low MEE

3.3 Robustness Test

By increasing the PRI value from 0.75 to 0.8 [44], The robustness of the configuration model is tested (see Table 11). The single consistency and solution consistency of the new model are both higher than 0.9. The high MEE configuration is a perfect subset of the high MEE configuration of the original model (see Table 9). The low MEE configuration is consistent with the low MEE configuration of the original model (see Table 9). The results of new model meet the criteria for the robustness test of fsQCA [44].

4.1 Equivalent Configuration and Asymmetry of Causality

When we use the QCA method to analyze the configuration, the assumption that conditions are inter-dependent and causal relationship between configuration and outcome is complex has been adopted, which is different from the traditional regression analysis method which assumes that variables (conditions) act independently [48]. The configuration analysis based on QCA in this study shows that three configuration paths can achieve high MEE, while two configuration paths will lead to low MEE. This not only shows the equivalent configuration caused by the interdependence among land condition variables, but also reveals the asymmetry of the causal relationship between these condition variables and MEE (as shown in Fig. 6).

In the research from the traditional regression perspective, it is generally to control other conditions unchanged and analyze the unified and symmetrical relationship between a specific variable (such as GTI or ERI) and MEE. For example, high GTI promotes MEE, while low GTI reduces MEE. However, the configuration analysis here shows that different configurations consisted of high GTI (low GTI) and other conditions may produce high or low MEE. For example, the outcomes of configuration H2 and configuration H3 that contain low GTI are both high MEE, while the outcome of configuration L1 that contains high GTI is low MEE. For another example, high GTC in each of the high MEE configuration is a supplementary condition, but low GTC is a core condition in each of the low MEE configuration. The asymmetry of equivalent configuration and causality suggests that we must pay attention to the specific practice of improving MEE, because a unique best path may not exist.

4.2 Green Technology Capability, Urbanization and Marine Ecological Efficiency

This study holds that coastal areas must have sufficient GTC to make GTI GTI have a positive impact on MEE. This point of view has been confirmed in this study. The configuration path in table 10 shows that although high MEE does not necessarily require high GTC (see H2 and H3 in table 10), low MEE is accompanied by low GTC (see L1 and L2 in table 10). This finding is consistent with existing studies, that is, high-tech industrial agglomeration promotes green economic efficiency in corresponding regions [42–43].

This finding has also been empirically verified, which is reflected in the corresponding typical cases explained by different configuration paths. For example, Guangdong is a typical case explained by configuration H1, while Fujian is a typical case explained by configuration L1. In 2016, the GTI of Guangdong and Fujian were 14.25 pieces per 10000 persons and 12.72 pieces per 10000 persons respectively, but the location entropy of high-tech industry in Guangdong and Fujian were 2.20 and 0.79 respectively. There is little difference between Guangdong and Fujian in GTI. As to GTC, however, Guangdong has higher national advantages, while Fujian is at a disadvantage.

In addition, the configuration analysis of this study reflects the asymmetric dependence between GTC and URB. High GTC matching high URB can obtain high MEE (such as configuration H1), but high URB does not have to match high GTC to obtain high MEE (such as configuration H2 and H3). A possible explanation is that high URB is conducive to technological innovation and high-tech industrial agglomeration [49], but the former does not necessarily lead to the latter. For example, the typical case Liaoning explained by configuration H3 is mainly characterized by resource-based cities, of which the location entropy of high-tech industry is just 0.499.

4.3 Peer Competition Pressure and Marine Ecological Efficiency

The results of this study show that PCP is an important conditional variable affecting MEE under the institutional background of China. The concept of Political Tournament was put forward by Chinese scholars based on the institutional analysis of China's political centralization and economic decentralization. It is proved that there is an action logic of local officials competing for promotion in China's political system. For example, Zhou (2005) verified the positive correlation between local officials' promotion and local economic performance through provincial data [50]. Political tournament is the realistic political ecology of China's current local officials' promotion game. In operation, it evaluates officials' political performance during their tenure through explicit indicators (such as the annual evaluation of ecological civilization construction adopted in this study), and the winner will get the opportunity for promotion [51]. Therefore, political tournament is believed to be an important mechanism for local governance in China.

It can be seen from Table 10 that the high PCP generated by the evaluation of ecological civilization construction is the core condition of configuration H1 and H3. In November 2012, the 18th CPC National Congress made the strategic decision of “vigorously promoting the construction of ecological civilization” from a new historical starting point, and comprehensively and deeply discussed all aspects of the construction of ecological civilization. In May 2015, the opinions of the CPC Central Committee and the State Council on accelerating the construction of ecological civilization was issued. In October 2015, with the convening of the Fifth Plenary Session of the 18th CPC Central Committee, enhancing the construction of ecological civilization was written into the national five-year plan for the first time.

Marine ecology is an important part of China's ecological civilization construction. As China attaches importance to the construction of ecological civilization at the national level, ecological civilization (including marine ecology) has become a common task faced by local governments at all levels in China. In particular, the release and implementation of the evaluation indicators of ecological civilization construction means that the central government has launched a political tournament around ecological civilization. The winners will be rewarded, and the losers will be punished by losing the opportunity for promotion in this round. Therefore, when local governments (such as Tianjin, Guangdong and Liaoning) rank low in the ecological civilization evaluation in 2016, the officials of these local governments will have the motivation and pressure to improve the ecological civilization evaluation in the coming year. For instance, more attention and political resources will be invested in fields related to ecological civilization (such as marine ecology), which may lead to a better outcome (such as high MEE).

4.4 Other Condition Variables and Marine Ecological Efficiency

Firstly, in received literature, the relationship between ERI and MEE (including green economy and green innovation) has always been controversial. For example, some studies believe that the excessive use of environmental regulation policy tools may have a negative impact on marine environmental pollution control [52]. Some studies believe that strengthening the intensity of marine environmental regulation in coastal areas is conducive to promoting the green total factor productivity of marine economy [53]. Other studies have pointed out that there is an inverted “U” relationship between environmental regulation and green total factor productivity [54]. The results of this study show that the ERI in high MEE configuration is insignificant or low, while the ERI in low MEE configuration is high, which means that whether ERI is valid or not needs to consider its matching with other conditions. If the interdependence between ERI and other conditions is brought into study, the controversy in received literature may continue.

Secondly, by comparing the five configurations, we put forward such a view that if the coastal areas expect to obtain GTC through OPEN, they need to have a certain GTC. The diffusion and transfer of technical knowledge is costly [55]. Only when various elements reflecting technical competence are combined into a purposeful technical activity system that can effectively use technical knowledge in some way, can an industrial organization have technology capability [56]. For the coastal areas in this study, high-tech industry is their green technology activity system. The higher the degree of high-tech industrial agglomeration is, the higher GTC of the areas will be, and more likely can these areas absorb external green technology to improve their MEE (such as configuration H1) through OPEN. Otherwise, OPEN may lead to pollution transfer and “pollution paradise” as a consequence. Therefore, when a costal area does not have high-tech industrial agglomeration, low OPEN could actually lead to high MEE (such as configuration H2 and H3).

Finally, configuration H2 is a special path. It can be seen from Table 10 that each condition variable of configuration H2 is at a low level, but its outcome is high MEE. Why? MEE is not a concept divorced from the development of human society, but an economic concept related to costs and benefits. Although each conditional variable of configuration H2 is at a low level, it does not mean that there is no costs and benefits. Note that the typical case explained by configuration H2 is Hainan. With a tropical monsoon climate and surrounded by the sea (as shown in Fig. 7), Hainan Province is the southernmost provincial administrative region in China.

Compared with other coastal areas in China, the high-tech industry of Hainan Province is poor developed. According to the statistical data of 2016 used in this study, the total industrial output value of Hainan Province is RMB 48.25 billion, which is ranked last in 11 coastal areas (the penultimate place is Tianjin, with a total industrial output value of RMB 680.513 billion). However, Hainan's ecological agriculture and marine ecotourism are well developed. Hainan can use relatively less investment to obtain relatively large expected output and relatively small unexpected output, which leads to the outcome of higher MEE. The pity may be that because the high MEE of Hainan is highly related to its geographical location, this configuration path has little reference significance for other regions.