Analysis of Voluntary Environmental Behavior and Innovation Quality of Chinese Textile Enterprises Based on Grounded Theory and Propensity Score Matching

At a new stage of “carbon peak” and “carbon neutral,” we need to strengthen the internal impetus of institutions and policies to promote ecological civilization construction. The initiative implementation of voluntary environmental behavior (VEB) is an effective way for enterprises to achieve high quality sustainable development. The VEB of enterprises include but are not limited to compliance with voluntary environmental regulations (VER). Different from the existing literature on enterprise VER, most of which are just the verification of ISO14001, the study conducts a comprehensive exploratory research on the theoretical framework of what is driving the VEB of Chinese textile enterprises. First, based on grounded theory, the paper coded the hand-collected environmental text information of Chinese listed textile enterprises from 2004 to 2018 and constructed a complete theoretical analysis framework from motivation to performance evaluation of enterprise VEB. Second, PSM method is taken to verify the conceptual model. As a result, we find the following: (1) in addition to ISO14001, the textile enterprises involve many industry-specific VER. (2) The VEB and VER of Chinese textile enterprises promote each other. (3) Either following VER or implementing VEB can effectively increase the innovation input and then actively improve the innovation quality of Chinese textile enterprises. In this way, the conclusion not only provides analysis ideas for future empirical research, but also offers practical suggestions to Chinese textile enterprises and policy makers for environmental management.


Introduction
Environmental protection is China's basic state policy, and sustainable development strategy is a national strategy. After setting the goal of "carbon peak" and "carbon neutral," we need to strengthen the internal impetus of institutions and policies to promote ecological civilization to a new stage of more conscious and proactive construction. With global warming and dwindling resources, every member of the planet has a responsibility to contribute to a better tomorrow. Both individuals and social organizations should take an active part in environmental protection. e research on environmental behaviors of enterprises should not be limited to traditional command-based and market-based environmental regulations, but increase the research on voluntary environmental regulations (VER) [1], which can better describe the enterprise's willingness to protect the environment. Since it is different from the traditional command-based and market-based environmental regulations, the VER is not a mandatory regulation, and enterprises can choose according to their environmental intentions. However, the voluntary environmental behavior (VEB) of enterprises is not limited to following the VER, but including multilevels of enterprise activities related environment protection. e concept of enterprise VEB is broader and worthy of attention, becoming the theme of the study.
When we review the existing literature, most studies on VEB and VER are conducted from two perspectives. One is by obtaining data through questionnaire survey to study the VEB of enterprise employees [2][3][4]; the other is to study the relatively common VER ISO14001 [5,6]. Besides, by comparing domestic and foreign literature, it can be found that there is a lack of research on VEB and VER under China's socioeconomic background. It might mainly be because command-based environmental regulations have been dominated for a long time in China, and the practice of VER is relatively insufficient [7], which provide full space for the new study of VEB and VER. erefore, the innovation of this paper lies in the construction of a widely applicable theoretical framework of enterprise VEB based on grounded theory, which provides analysis ideas for future empirical research. In addition, under the background that China's economy is gradually changing from an export-oriented development mode to a domestic cycle, the living environment of traditional manufacturing industry is also changing. Among many traditional manufacturing enterprises, textile enterprises are representative. Today, China has become a textile power. Textile enterprise is vital in the national economy. e environmental problem is an important challenge that traditional textile enterprises have to be faced with [8]. In this way, it is of great practical significance to study whether traditional textile enterprises can make full use of intelligent method and voluntarily carry out environmentally friendly behaviors to support the sustainable development of themselves and the society.
But in practice, what kind of ways do enterprises manage to protect the environment? Is it through indirect behavior at the individual level or direct behavior at the enterprise level? Is it the compulsion of external policy, or the spontaneous behavior of the enterprise? Traditional commandbased policies mostly punish those who fail to meet the standards after the event, but do not encourage environmental pioneers [9]. However, sometimes enterprises may only follow in name but do not carry out substantive green management [10,11]. We can further classify the relevant literature into the following two categories.

VEB.
Most studies on VEB focus on employee behavior. Employees' cognition of corporate social responsibility can effectively promote their VEB, which can be promoted by substantive incentives but can be weakened by symbolic incentives. It also requires enterprises to take social responsibility into consideration when formulating sustainable development strategies, so as to truly promote employees' VEB [2]. At the same time, how employees participate in VEB is also affected by the organization's definition of the concept [3]. At the same time, individual consumption reduction is also one of the behaviors to achieve sustainable economy [12].
From the perspective of enterprises, data from Japanese manufacturing enterprises show that enterprises' spontaneous application of environmental management system can not only effectively reduce the negative impact on the environment, but also effectively improve the productivity of enterprises [13]. At the same time, corporate environmentalism can effectively promote enterprise innovation [14]. In terms of environmental information disclosure, it may only be reflected in quantity, but the quality is not satisfactory [15]. However, the disclosure of voluntary environmental information does affect the future value of enterprises [16].

VER.
Even VER cannot be implemented without the promotion of government departments [17], and the existence of government is conducive to the construction of green cooperation network for private enterprises [18]. e experience from India shows that the willingness of enterprises to comply with VER will be affected by enterprises' size, location, export-orientation, intangible asset value, and other factors [19]. Moreover, voluntary compliance with environmental regulations does help enterprises to improve the environmental performance [20].
As a VER tool, ISO14001 is a supplement to the traditional environmental regulations. Research shows that ISO14001 certification can significantly promote enterprise innovation [21], so managements should take ISO14001 certification as a core of the company's innovation development strategy, effectively carrying out environmental management activities and investing resources in a long-term way [5]. In addition, the implementation of voluntary environmental management systems (such as ISO14001 and United Nations Global Compact) by enterprises can effectively improve their environmental performance [22]. At the same time, green competitors and external environment are also important factors affecting whether an enterprise complies with VER [23]: the more stringent the formal environmental regulations and the more ISO14001 certified enterprises in the region, the more likely an enterprise is to carry out ISO14001. e enterprise with high degree of external and ISO9000 certification is more likely to carry out ISO14001 [6]. However, the communication of positive image may not be one of the motivations for enterprises to follow VER [24], but the management diversity may be one of the influencing factors [25].
In general, we find that the previous literature of the theme is centered on employees' voluntary behavior and ISO14001, the data source is mainly questionnaire and enterprise financial report data, and for the method most research applies statistical analysis [2][3][4] and some adopt social network [18]. However, most of the existing studies are only test static single data, which cannot fully reflect the whole VEB of enterprises. erefore, this research firstly applies the grounded theory to comprehensively and systematically characterize the VEB of Chinese textile enterprises through the three-level coding technology and then takes PSM method to further verify the conceptual model. To make up the existing research deficiencies from the theme, research methods and data are used.

Research Method.
In order to fully characterize the VEB of textile enterprises in China, the grounded theory is adopted in the paper. Grounded theory is a classical method widely used in qualitative research.
rough repeated reading and comparison of text information, effective information points are naturally presented, so as to conceptualize behavior patterns, extract core elements, and describe the logical relationship between core elements [26].
At the same time, in order to verify the rationality of the model, we take Propensity Score Matching (PSM) method after model construction. e variable included and research framework are introduced in Sections 2.4 and 2.5.

Data Sources.
In the paper, 39 enterprises with sufficient environmental text information were selected from 111 listed textile companies in China, and more than 165,395 words of environment-related text information were handextracted from their annual report and social responsibility report from 2004 to 2018. Among them, the sample size with valid text information sample reached 218.

Data Processing.
After collecting and sorting the environmental text of the sample enterprises, we mainly combed it manually and analyzed the data with the help of NVivo12. 30 enterprises coded to build models, and the remaining 9 were used for saturation test. Due to different data sources, sometimes text information and report information do not match, and repeated comparison and cross-data verification are required.

Dependent Variables.
Innovation input: the input of innovation activities mainly considers the enterprise's innovation capital input (calculated by the enterprise's R&D expenditure) [27,28].
Innovation quality: the paper mainly studies the impact of environmental regulation on innovation quality of textile enterprises, and greenness is an important aspect of innovation quality. erefore, we use green innovation to measure the innovation quality here, and green patent is a common indicator to measure the performance of green innovation [8].

Independent
Variable. VEB: the VEB data are obtained from the grounded theory research conclusion in the above part. Textile enterprises that have implemented VEB are marked as 1, while those that have not implemented VEB are marked as 0.
VER: the most widely practiced VER in China is ISO14001 [7]. e study measures the compliance of enterprises with VER based on whether enterprises have obtained ISO14001 certification. Textile enterprises that have completed ISO14001 will be recorded as 1, and those that have not completed ISO14001 will be recorded as 0.

Control Variables.
Considering the influence of enterprise heterogeneity, the subindustry, region, and property rights of the enterprise are put into the model as control variables [8]. In addition, the size and operating performance of an enterprise also affect the innovation behavior of an enterprise, the size (measured as total assets), revenue growth rate, and return on total assets (ROA) [28].

Counterfactual Research Framework.
We take the textile enterprises implemented VEB/VER as treatment group, and the other ones as control group. Since it is difficult to ensure that the probability distribution of the experimental group and the control group is consistent, the counterfactual research framework is constructed by referring to Rosenbaum and Rubin's method [29], to find a control group as similar as possible to the experimental group, so as to effectively reduce the biased estimation caused by the nonrandom distribution of samples. In this context, the average treatment effect (ATT) of the treatment experimental group (textile enterprises implementing VEB/VER) was defined as where Y1 is the input/output level of textile enterprises' innovation activities when they implement VEB/VER, and Y0 is the input/output level of textile enterprises' innovation activities when they do not implement VEB/VER. e premise hypothesis of ATT is that when the textile enterprise implements VEB/VER (M = 1), the difference (Y1-Y0) of the input/output level of textile enterprise innovation activities from those enterprises without implementing VEB/VER is calculated. In this way, the influence of implementing VEB/ VER on the input/output of innovation activities of textile enterprises can be obtained. However, in the model (1), the result of E (Y1|M = 1) can be observed, but E (Y0 |M = 1) cannot, called counterfactual result. erefore, we can use PSM method to construct the alternative indicator of E (Y0| M = 1). In other words, on the premise of a set of covariables, a virtual sample group (control group) matching with the textile enterprises implementing VEB/VER (treatment group) was constructed by PSM method, so as to establish a reasonable counterfactual framework.

Descriptive Statistics.
Based on the statistics of environmental disclosure information of sample enterprises, we analyze the distribution of samples (number of texts, how many texts of sample enterprise disclose environment protection information) and content (number of contents, how many words of the sample enterprise disclose environment protection information) from three dimensions: subindustry, property right, and region of enterprises. First of all, for the number of texts (Figure 1), the clothing industry accounts for the highest proportion, even higher than the textile and fiber industries which are traditionally considered to have more serious environmental problems.
Due to the small sample size of the leather industry, it is normal that it ranks last. However, digging into the content of disclosed information (Figure 2), it is true that the textile industry and fiber industry are more specific and in depth in green management. Because of relatively less pollution, the clothing industry has more green management from the overall strategy, specifically from the daily office behavior to implementing environmental protection.
Secondly, from the perspective of the property rights of enterprises, private enterprises have an absolute advantage in both the sample number ( Figure 3) and the content (Figure 4), which is inseparable from the fact that private enterprises account for 78% of listed textile enterprises in China.

Computational Intelligence and Neuroscience
In addition, from the regional point of view, the environmental information disclosure of Zhejiang province (ZJ) ranks first in both sample size ( Figure 5) and content ( Figure 6) and is far ahead of other provinces. On the one hand, Zhejiang province (ZJ) itself is a big textile province of China, and the listed textile enterprises in Zhejiang province (ZJ) account for 27.93% of all listed textile enterprises in China. On the other hand, in the 31 listed textile enterprises in Zhejiang province (ZJ), 15 have disclosed environmental text information, accounting for nearly half. Jiangsu Province (JS), which ranks second in terms of content and information disclosure and belongs to the Yangtze River Delta region with Zhejiang province (ZJ), is also a big textile province of China and also takes the lead in green management.

Open Coding.
Open coding is a dialogue between researcher and data and the process of continuous decomposition and analysis of the original data. rough repeated comparison of the similarities and differences of the original data word by word, objects with similar characteristics are given the same conceptual label [30]. By sorting out the obtained environmental text information of China's listed textile enterprises, we screened and obtained 5285 words of key coding statements from the 165,395 words of original data and further summarized the following categories (Table 1).    Computational Intelligence and Neuroscience

Axial Coding.
e axial coding is to construct the master-slave category of the concepts formed by the open coding, and the combining of the structure between concepts is another test of data [30]. We further analyze the 16 categories obtained by open coding to explore the causal and subordinate relationships between different categories. In this stage of coding, the 16 parallel categories formed by open coding are further classified into five main categories (external factor, internal factor, VEB, VER, performance). e relationship between categories is shown in Table 2.

Selective Coding.
Selective coding is to find the "core category" of the study and build a story line around it for description [30]. rough further comparative analysis of 16 concepts obtained by open coding and 5 main categories formed by axial coding, the core category "VEB" that can radiate the whole story is found out. And the category connecting its upstream and downstream is thus divergent to form a complete logical chain. e complete logical chain can help us form a theoretical framework for the analysis of enterprises' VEB, as shown in Table 3.

eoretical Saturation Test.
eoretical saturation test means that all categories are fully reflected in their attributes, dimensions, and forms of change, and further data collection and analysis will hardly make new contributions to the conceptualization. e paper coded the text information of the remaining 9 sample enterprises but failed to get the initial concept and relationship structure significantly different from the above 16 categories and 5 main categories, proving that the information of this theoretical research has reached saturation, and further information collection and sorting can be stopped.

eoretical Model Construction and
Interpretation. rough open coding, axial coding, and selective coding, the following theoretical framework (Figure 7) is formed for the motivation and influence path of the VEB of Chinese textile enterprises.

Motivation.
From the external perspective, first of all, environmental policy is a very important influencing factor. No matter the national environmental protection law, environmental protection tax law, or the local water pollution prevention and control regulations, river chief system will restrict the implementation of VEB of enterprises, which is also the macroenvironmental motivation in the process of enterprise development. Secondly, the position of enterprises in the industry will also have an impact on the VEB of Environmental policy e company earnestly practices the development concept that "lucid waters and lush mountains are golden mountains and silver mountains" NO.12 2 Position in industry e very few enterprises in the spandex industry can produce this kind of products NO.14 3 Honor Awarded as "regional green enterprise," "Henan green enterprise," and other honorary titles NO. 16 Circular economy We will continue to promote clean production and a circular economy and strive to improve the comprehensive utilization of resources NO.11 Computational Intelligence and Neuroscience  Enterprises' implementation of VEB and compliance with VER are mutually reinforcing. Enterprises with active implementation of VEB are more willing to comply with VER (such as voluntary environmental product certification). Companies that have followed VER (such as ISO14001) will be more active in implementing VEB VEB/VER ⟶ Performance e effect of implementing VEB and complying with VER will be reflected in the environmental performance of enterprises and the sustainability of their future development External Motivation⟶VEB/ VER⟶Performance e external motivation such as policy environment and the position in industry of enterprises will affect the implementation of VEB and compliance with VER, which will be reflected in the environmental performance and economic effect of enterprises Internal Motivation ⟶ VEB/ VER ⟶ Performance e internal motivation such as corporate honor, development strategy, and quality requirement will affect the implementation of VEB and compliance with VER and then be reflected in the environmental performance and economic effect of the enterprise Internal and external factors interact with each other to influence the implementation of VEB and compliance with VER, which are reflected in the environmental performance and economic effects of enterprises enterprises. If an enterprise is in a leading position in its industry, it is more motivated to lead the industry in environmental protection, such as becoming the first enterprise in the industry to conduct a voluntary environmental certification or being one of the few enterprises in the industry to produce a certain environmentally friendly product. From the internal point of view, the honorary titles such as "energy-saving advanced enterprise" and "regional green enterprise" help enterprises to further implement VEB. And if companies put environmental protection into their development strategies, it will also be implemented in the daily VEB. e pursuit of excellence of some advanced enterprises will also make them not satisfied with merely reaching the national environmental protection standards, but through increasing investment in capital and technology to achieve excellence in environmental protection.
External motivation and internal motivation do not affect the VEB of enterprises alone but also have a mixed effect. Because both external and internal factors jointly constitute the macro-, medium, and micro environment for Chinese textile enterprises to implement VEB and follow VER, there are cross-layer interactive influences.

VEB vs. VER.
VEB is a series of actions taken by enterprises to achieve the purpose of environmental protection, including but not limited to the formulation of internal environmental regulations, the establishment of the corresponding environmental management organization, increasing capital investment, research and development of environmental protection technology, and the implementation of supply chain management, etc.
VER is a series of environmental protection standards that enterprises voluntarily follow. e most representative one is ISO14001 as an environmental management system certification for the whole enterprise to be environment friendly. In addition, it also includes ISO50001 energy system certification for energy conservation, voluntary environmental protection product certification, and standard 100BYOEKo-TEX certification (one of the most well-known labels in the world stating that textiles and accessories do not pose a health risk), etc.
In order to analyze the relationship between Chinese textile enterprises' implementation of VEB and compliance with VER, software STATA was used to analyze the correlation between the sample size of Chinese text information (number of text) and the number of ISO14001 certifications. It can be seen from the results in Table 4 that there is a positive correlation between the two. In Figure 8, the comparative data of the number of text disclosure of sample enterprises and their ISO14001 certification can also be further supported. And theoretically, the two are complementary [31]. Textile enterprises that carry out substantive green management disclose more text information and are more inclined to carry out ISO14001 certification. In turn, textile enterprises with ISO14001 certification will be more standardized to implement VEB and information disclosure.

Performance.
Whether an enterprise implements VEB or complies with VER, the effectiveness will be reflected in its performance, which can be divided into direct performance and indirect performance. e direct performance of enterprises' VEB must be a breakthrough in environmental protection. Enterprises can achieve the cleaner production through investment in R&D and technological transformation. In this way, the direct purpose of environmental protection can be achieved from the production process and results. e indirect performance of enterprises' VEB is mainly reflected in the economic benefits of long-term development, such as enabling China's textile enterprises to achieve transformation and upgrading, sustainable development, and circular economy. e VEB of enterprises can help to get more competitive and long-term development in the future, which are also in line with the expectations of "Porter hypothesis."

Propensity Score Matching Method.
Since it is "voluntary environmental behavior" and "voluntary environmental regulation," the implementation of enterprises has a certain "self-choice." In order to further verify the rationality of the model proposed by the grounded theory, predicting the innovation quality impact of enterprise after "voluntary" implementation, the nonparametric estimation method, Propensity Score Matching (PSM) method, will be introduced. Under the premise that the covariables of the two sample firms are as similar or the same as possible, the propensity score of each sample firm to enter the treatment group is calculated. is method can accurately evaluate the effect of implementing VEB/VER on the input/output level of textile enterprises' innovation activities.

e Effect of VER on Innovation Input of Chinese Textile Enterprises.
Firstly, the samples of the treatment group and the control group were matched with k-neighborhood 1:4 within the caliper (default radius 0.05) to study the effect of VER on innovation input of Chinese textile enterprises.
(1) Common Support Area and Balance Check. According to the matching results in Table 5, only 1 sample was lost and 1664 samples were retained in the treatment group and the control group. Most of the observed values were within the common value range, as can be seen from Figures 9 and 10, indicating good matching effect.
In order to ensure the reliability of PSM, it is necessary to test the balance of covariables to ensure that there are no significant differences in other influencing factors except VER. e results in Table 6 show that, compared with the results before matching, the standardization bias (% bias) of most covariables after matching is basically controlled below 10%, which reduces the total matching bias to a large extent.

Computational Intelligence and Neuroscience
Although the standardized deviation of ROA is -12.4% after matching, it is still within the acceptable range. e results show that the matching of covariables is reliable and has passed the balance test. Figure 11 intuitively reflects the improvement of bias after sample matching.
(2) Measurement of Treatment Effects. According to Model 1, the impact of VER on innovation input of textile enterprises is measured. It can be seen from Table 7 that implementing VER can effectively promote innovation input of textile enterprises (the difference of ATT is positive and t-stat is 3.21).

e Effect of VER/VEB on Innovation Quality of Textile
Enterprises. en, the effect of VER and VEB on the innovation quality (green innovation) of textile enterprises is tested by using k-neighborhood 1:4 matching.
As mentioned above, the common support area and balance of matching samples were tested first. It can be seen from Tables 8 and 9 and Figures 12-15 that there are few lost samples and the matching effect is satisfied.
It can be seen from Tables 10 and 11 that both matching samples of VER and VEB have passed the balance test. Figures 16 and 17 visually show the improvement of sample bias after matching.   Computational Intelligence and Neuroscience According to the model 1, to measure the effect of VER/ VEB on innovation quality (green innovation) of textile enterprises, it can be seen from Tables 12 and 13 that implementing VER and VEB indeed can effectively promote the innovation quality (green innovation) of textile enterprises. Comparatively, VEB (the T-stat value of ATT is 5.52) has a more significant promoting effect on textile enterprises' innovation quality than VER (the T-stat value of ATT is 4.89). In other words, when Chinese textile enterprises voluntarily follow environmental regulations and implement environmental behaviors, the innovation quality of enterprises, especially green innovation, can be effectively promoted. It is also the original intention of environmental regulation.
Although the results of different matching methods are slightly different, if the conclusions drawn by different methods are similar, the matching results are robust [32].
e results of robustness test by different matching methods are consistent with the above results. Due to space limitation, the results of robustness test will not be presented in detail here.

Conclusion and Discussion
Based on the grounded theory, the paper establishes a comprehensive and specific theoretical framework for the motivation and influence path of enterprises' VEB; furthermore the PSM method is taken to verify the model, which can provide some references and enlightenment to the management practice of enterprises and governments, respectively.
To the enterprise, the implementation of VEB is not only to follow the external policies and regulations, but also to promote its industry position and pursue the excellence. Take the environment protection as enterprise strategy, and earnestly implement it in the daily operation, so as to improve enterprise competitiveness for the future sustainable development [33]. At the same time, to be specific, complying with VER such as ISO14001 can help enterprises to sort out their VEB and standardize it systematically. Voluntary environmental protection product certification is helpful to externalize the positive effect of VEB of enterprises to the market and consumers and form the overall green development of the supply chain system. Most importantly, either following VER or implementing VEB can effectively increase the innovation input and then actively improve the quality of innovation of textile enterprises.
To the government, first of all, it is better for the government to appropriately guide the VER, in addition to the traditional command-based and market-based environmental regulations. Although the enterprise voluntarily follows, if the government can provide some financial support and rewards [34], it can form a positive encouraging effect on the enterprises. Secondly, in terms of the environmental policy, in addition to the top national level policy, it can form more complete and targeted policy with industry level or regional level [35]. For example, we analyze the clothing industry and textile industry on the expression of VEB which will make a huge difference. In the aspect of green textiles, the textile industry has more practice of voluntary certification, relative to other industries. erefore, more targeted guidance can improve the policy effectiveness in segmented areas [36].
At last, as all data sources in this paper are collected manually, there may be some errors, which is the biggest limitation of the study.
Data Availability e data that support the findings of this study are available on request from the corresponding author.

Conflicts of Interest
e authors declare no conflicts of interest.

Authors' Contributions
Conceptualization was done by Yabin Yu; data curation was done by Yabin Yu and Qian Xu; formal analysis was done by Yabin Yu; methodology was done by Yabin Yu and Hua Cheng; validation was done by Qian Xu; visualization was done by Qian Xu; writing-original draft was done by Yabin Yu; Writing-review & editing was done by Yabin Yu and Hua Cheng.