Systemic Configurations of Functional Talent for Green Technological Innovation: A Fuzzy-Set QCA Study

Abstract

Achieving high-level green technological innovation in heavily polluting enterprises is critical for advancing sustainable development, particularly in the context of both organizational and regional digitalization. This study adopts a configurational perspective grounded in the Technology–Organization–Environment (TOE) framework and integrates theoretical insights from resource orchestration, resource dependence, and IT capability theories. It investigates how different types of skilled talent, such as production, technical, sales, and managerial employees, contribute to green innovation under varying digital conditions. By applying fuzzy-set qualitative comparative analysis (fsQCA) to a sample of 96 publicly listed firms from China’s heavily polluting industries, this study identifies four distinct talent-based configurations that can lead to high levels of green innovation: production-centric, management-led, technical talent driven, and regionally enabled models. Each configuration reflects a specific system state in which a core group of skilled employees plays a leading role, supported by complementary functions, and shaped by the interaction between internal digital transformation and the external digital environment. This study contributes to the systems literature by elucidating the combinational roles of digital resources and talent deployment within the systemic TOE framework, and offers practical guidance for enterprises aiming to strategically utilize human capital to enhance green innovation performance amid ongoing digital transformations.

1. Introduction

Accelerating the green transformation of traditional industries is critical for achieving global sustainable development. Recent research shows that digitalization can strengthen green innovation in manufacturing firms by optimizing resource allocation, reshaping governance structures, and generating positive spill-over effects [1]. For incumbent traditional enterprises, digital technologies therefore represent a critical means of overcoming resource and environmental constraints and advancing sustainable growth [2]. Yet heavily polluting firms—characterized by high resource consumption, elevated emissions, and comparatively weak digital capabilities—face acute institutional and market pressures in this transition [3]. Understanding how digitalization can catalyze green technological innovation within such firms has thus become essential to the broader green transformation of traditional industries.

Green technological innovation is a multifaceted, complex system-level endeavor, involving intricate interactions between technological capabilities, organizational resources, external environmental factors, and strategic and managerial dimensions that guide and sustain innovation processes [4]. A critical determinant of green technological innovation success is the extent to which employees can transform organizational strategies, resources, and technologies into greener products or processes [5]. Heavily polluting firms typically deploy human resources across functional units—production, technology, sales, and the like—based on specialized skills and business responsibilities. Such functional segmentation clarifies innovation objectives and helps manage risk throughout the green-innovation pipeline. To enhance innovation performance, some firms adopt flexible arrangements—matrix structures, cross-functional teams, or other agile configurations [6]. Yet whether organized along traditional lines or newer forms, these organizations confront perennial challenges in staffing, resource allocation, and conflict resolution. Addressing those challenges requires prioritizing different categories of skilled talent at distinct stages of the innovation process. Production and technical staff drive product and process improvements, whereas sales personnel convert green innovations into market share and financial returns. Consequently, firms must allocate scarce resources strategically and manage cross-functional relationships effectively. Although prior studies have examined how human-capital composition and green HRM practices influence green technological innovation [7,8], they have not assessed the relative importance of technical, sales, managerial, and production talent within a comprehensive systemic framework that accounts for technological and environmental contexts. This study therefore investigates whether the contributions of these talent categories are hierarchical, identifying which play core versus supporting roles. In doing so, we aim to inform more strategic and agile talent-management practices [9] grounded in a systemic understanding of innovation drivers.

Digital transformation reshapes job profiles, skill structures, and organizational innovation processes [10]. Accordingly, the deployment of different categories of skilled talent should correspond to digitalization levels at both the firm and regional scales, reflecting the multi-level nature of the innovation system. At the firm level, digitalization has two opposing effects on the green innovation efforts of heavily polluting enterprises: a data-driven effect that facilitates innovation and a capability-curse effect that can hinder progress [11]. Enterprises that combine substantial digital investment with high employee digital literacy can leverage data analytics to accelerate green technological innovation. By contrast, firms constrained by limited capital or digital expertise struggle to convert digital resources into green outcomes. This divergence raises an important question: Which category of skilled talent becomes the critical driver of green technological innovation under high versus low levels of firm-level digitalization? At the regional level, a developed digital economy supplies robust infrastructure and rich market-demand data for green transformation [12]. Digital platforms and consumer analytics not only support R&D activities but also help firms align green innovations with customer preferences, thereby enhancing commercialization success. Yet in areas where digital infrastructure is weak, these advantages are muted. Consequently, it remains unclear how firms should configure skilled talent when operating in regions with high versus low digital maturity.

Hence, this study asks the following research question: Under what configurations of skilled talent (production, technical, sales, managerial), enterprise-level digital transformation, and regional digital-economy development do heavily polluting firms achieve a high level of green technological innovation performance?

To address these research gaps, this study constructs a Technology–Organization–Environment (TOE) analytical framework, recognized for its utility in understanding complex systems where technological capabilities, organizational structures and resources, and external environmental conditions interact synergistically or antagonistically to influence outcomes [13]. This systemic TOE framework provides the necessary structure to analyze the configurational interplay of skilled talent, firm digitalization, and regional digitalization in driving green innovation. Drawing on a sample of 96 A-share firms in China’s heavily polluting industries, we apply fuzzy-set qualitative comparative analysis (fsQCA)—a methodology well-suited for uncovering causal complexity and identifying distinct system configurations—to examine the complex causal configurations linking skilled talent to green technological innovation in a digitalizing context.

2. Theoretical Foundation and Research Model

2.1. Literature Review

Green technological innovation refers to the innovation of technologies, processes, and products that follow ecological principles and reduce environmental pollution. It enhances energy efficiency and green total factor productivity [14] and serves as a strong guarantee for balancing both green and financial performance. External institutional and market environments, as well as internal green orientation and capabilities, are important drivers of green innovation [14]. However, due to high costs and insufficient investment in innovation, heavily polluting enterprises, in addition to relying on voluntary or involuntary green behaviors and environmental management measures [15], must also focus on the role of personnel in specific departments, such as R&D, in achieving green competitive advantages [16]. The core business and work resources in departments such as production and sales are heterogeneous, providing organizations with differentiated skilled talents. The human capital accumulated through these resources is a supporting condition for green innovation. Green human resource management theory introduces green elements into human resource practices, suggesting that methods such as green recruitment, green training, and green incentives can stimulate employee efficiency and motivation, optimizing the structure of green human capital to acquire the talent necessary for green technological innovation [7]. Strategic talent management theory further emphasizes the need to distinguish between skilled talents with or without strategic performance [17], which involves identifying a key talent pool that is beneficial for achieving organizational strategic goals and configuring innovation resources accordingly [9,17]. For example, during the early stages of a company’s development, technical personnel play a crucial role in obtaining core competitive advantages, while in the growth stage, sales talent becomes key to expanding market share. It is also worth noting that digital technologies, characterized by data homogenization and programmability, have changed the way green innovation elements are combined. The innovation-driving role of skilled talent is closely linked to the digitalization levels of both the region and the enterprise. Therefore, enterprises need to carefully adjust management behaviors and organizational structures based on the characteristics of digital technologies [18], enhancing the coupling between different types of talent and between talent and digital technologies, in order to maximize the green-driving effect of skilled talent based on existing digital resources.

However, although existing research has reviewed the driving factors of green technological innovation based on institutional theory, market theory, and the resource-based view of natural resources [19], and employed hierarchical regression models to analyze the mechanisms through which policy and market pressures, as well as innovation resources and capabilities, influence green innovation, there has been insufficient research on the heterogeneous innovation-driving effects of skilled talent in the context of dual digitalization. Specifically, there is a lack of understanding of which types of skilled talent are key drivers of green technological innovation and what roles other members play. Due to the inextricable link between enterprise and regional digitalization and skilled talent, traditional regression analysis is inadequate for clarifying their complex interactive effects on green technological innovation, making the need for configurational analysis evident. Ragin [20], from a configurational perspective, proposed the QCA method, which avoids issues such as bidirectional causality and multicollinearity. However, the use of QCA requires the selection of appropriate causal variables through inductive, deductive, or other methods [21]. Compared to other types of innovation, green technological innovation demands higher requirements and carries greater risks. The increasingly diverse digital world, in which technological, organizational, and environmental elements are interdependent, further complicates the innovation process [22]. Enterprises must be prepared for innovation in terms of technological advantages, organizational capabilities, and environmental changes to achieve green innovation development [23]. Therefore, this study introduces the TOE framework, which encompasses organizational characteristics, technological conditions, and environmental factors, to select the antecedent conditions for analysis [24].

2.2. Research Questions

Building on the integrated TOE framework and the existing literature, this study seeks to answer the following key research questions:

RQ1: How do different configurations of skilled talent—including production, technical, sales, and managerial personnel—influence the achievement of high-level green technological innovation in heavily polluting enterprises?

RQ2: In what ways do varying levels of enterprise digital transformation interact with skilled talent configurations to drive green technological innovation?

RQ3: How does regional digital economy development moderate the combined effects of enterprise digital transformation and skilled talent configurations on green technological innovation?

RQ4: How can an integrated TOE-based framework, incorporating resource orchestration, resource dependence, and IT capability theories, be used to systematically identify effective configurations of talent and digitalization that lead to superior green technological innovation outcomes?

2.3. Research Framework

The TOE framework not only provides theoretical support for selecting the causal variables in this study but also offers a valuable perspective for analyzing configurational multiplicity. According to this framework, organizational, technological, and environmental conditions collectively influence green technological innovation, with multiple configurations representing different combinations of these three factors. In addition to configurational multiplicity, theoretical multiplicity necessitates the use of multiple, competing, or complementary theories to fully explain the reasons behind the emergence of a specific configuration [21]. Given the multiplicity of configurations and theories, this study integrates resource orchestration, resource dependence, and IT capabilities theories to construct a TOE analysis framework that examines the impact of different types of skilled talent on green technological innovation under the context of dual digitalization, as illustrated in Figure 1.

2.3.1. Organization: Skilled Talent

Organizational conditions refer to the characteristics or resources that enterprises possess, such as company size and organizational structure [24]. Resource orchestration theory suggests that a firm’s competitive advantage arises not only from its unique, inimitable resources but also from the effective management of these resources [25]. Heavily polluting enterprises in China typically require four types of skilled talent—production, technical, sales, and management—to ensure the smooth operation of the organization and foster green technological innovation. Specifically, production staff often improve production processes through tacit knowledge gained from experience. Technical talent primarily drives green technological innovation by updating existing technologies and developing new ones. Sales personnel, being more attuned to the latest market demands, excel at generating creative ideas from consumers, thus guiding the direction of green innovation. Meanwhile, managers influence green innovation behaviors in other departments and impact green innovation performance through opportunity recognition, political connections, and green human resource management practices [7]. However, due to resource and time constraints, organizations struggle to fully leverage the innovative capabilities of all talent categories in a balanced way. If certain talent groups are disproportionately emphasized while others are neglected, it may increase coordination difficulties, ultimately hindering green innovation. Furthermore, the role of these four types of skilled talent in driving green technological innovation may vary depending on the internal and external environments in which the enterprise operates. Therefore, this study incorporates all four types of talent into the analysis framework to explore which talent configuration is most conducive to green technological innovation. In other words, it seeks to identify the relative importance of skilled talent in the green innovation process and, based on this, design the optimal talent structure, allocate scarce resources, and ultimately enhance human capital.

2.3.2. Technology: Digital Transformation of Enterprises

The application of various technologies by enterprises is a key enabler of green technological innovation, and technological conditions refer to the organization’s level of technological development. Based on the resource-based view, Bharadwaj [26] proposed the IT capabilities theory, which asserts that IT capabilities—derived from the integration of IT resources and other resources and capabilities—enhance enterprise performance. New-generation information and communication technologies, such as big data, blockchain, and artificial intelligence, emphasize the innovative value of data elements. These technologies not only expand the scope of IT capabilities but also encourage organizations to leverage digital technologies to promote information sharing and resource integration among skilled talent, thereby facilitating green technological innovation. However, the positive effects of digital technologies, such as improvements in work quality and efficiency, may vary depending on the enterprise’s level of digitalization and employees’ professional skills. Moreover, potential environmental rebound effects, including increased electricity demand and electronic waste generation, may offset some benefits of digital transformation [27]. To mitigate these negative impacts, enterprises can adopt ecological design and circular economy practices, which focus on resource efficiency and waste reduction, thereby enhancing the sustainability of their digital innovation efforts. Therefore, enterprises should integrate and orchestrate various types of skilled talent according to their level of digitalization in order to more effectively enhance the efficiency and quality of green technological innovation through digital technologies. Consequently, digital transformation is incorporated into the research model to identify which types of skilled talent are better positioned to drive green technological innovation, based on existing digital transformation initiatives.

2.3.3. Environment: Regional Development of the Digital Economy

The external technological environment refers to key environmental factors that influence the survival and development of enterprises, such as technological infrastructure and the availability of high-skilled labor. Resource dependence theory posits that organizations can acquire innovative elements—such as capital, technology, information, or talent—from the external environment and must also make adaptive adjustments based on dynamic environmental changes, thereby reducing the negative impact of uncertainty on core competitiveness [28]. The digital economy of a city, as the external environment for enterprise green innovation, not only provides digital infrastructure and innovation platforms for green technological innovation but also creates a market environment where pollution-intensive enterprises with weaker innovation capabilities can more easily access specialized talent and technologies. This environment compels these enterprises to accelerate their digital transformation efforts [29]. However, the development of the digital economy also makes the external innovation environment more volatile, complex, ambiguous, and uncertain, with digital economic resources being unevenly distributed and limited. As a result, there are significant disparities in the external digital resources available to enterprises. Consequently, organizations are forced to adjust their digital transformation processes and the configuration of skilled talent to ensure alignment with the digital economy environment. Therefore, it is essential to explore the interactive effects of regional digital environments, enterprise digital transformation, and various types of skilled talent on green technological innovation.

In summary, the system-oriented TOE framework integrates resource orchestration, resource dependence, and IT capability theories to not only explain the individual influence of each antecedent but also to reveal their interdependent and systemic effects on green technological innovation in the context of dual digitalization.

3. Research Design

3.1. Methodology

The qualitative comparative analysis (QCA) method, based on set theory, enables a comprehensive analysis of condition configurations at the case level. This approach not only differentiates between core and peripheral conditions that lead to outcomes but also accommodates larger sample sizes, enhancing the generalizability of the conclusions [20]. It is particularly suitable for addressing the complex, systemic issues explored in this study, as it allows for the identification of multiple, equifinal pathways (different system configurations) leading to the same outcome, reflecting the inherent complexity of socio-technical systems like green innovation under digitalization. QCA consists of three types: csQCA, mvQCA, and fsQCA. csQCA is applicable to binary variables, mvQCA handles multi-value nominal variables, and fsQCA analyzes interval and ratio variables. Since the antecedent conditions (including talent types, firm digitalization, regional digitalization—representing key dimensions of our systemic TOE framework) and outcome variables in this study are primarily ratio variables, fsQCA is employed for the configuration analysis. FsQCA’s ability to handle fine-grained differences and model causal complexity makes it ideal for uncovering how different combinations of TOE factors (sets of conditions) constitute sufficient configurations for high green technological innovation.

3.2. Data Collection and Variable Description

In accordance with the Guidelines for Industry Classification of Listed Companies (2012) published by the China Securities Regulatory Commission, heavily polluting firms listed on the A-share market in 2022 were selected as the case sample for this study. As the world’s largest manufacturing economy, China is currently in the early exploratory phase of digital transformation. It features a unique industrial scale, a continuously evolving regulatory framework emphasizing environmental sustainability, and distinctive labor market dynamics. Selecting heavily polluting Chinese enterprises as the research subjects helps to uncover the mechanisms by which digital transformation and talent configurations influence green technological innovation. The findings can provide valuable theoretical insights and practical implications for other manufacturing-intensive countries undergoing similar digital transformation processes. Due to the unsuitability of certain companies’ operational statuses for this research, the following data treatments were applied: companies with abnormal financial conditions, including ST, PT, and *ST firms, were excluded, and all samples with missing data for any variable were removed. Ultimately, a dataset consisting of 96 heavily polluting firms was constructed. The measurements and sources of the variables for each dimension are detailed below.

The outcome variable, corporate green technological innovation, is conceptualized as encompassing both green product innovation and green process innovation [14]. Three principal operationalizations are prevalent in the literature. First, from the product-energy perspective, scholars calculate the ratio of new-product sales revenue to energy consumption to capture the greenness of novel offerings. Although closely tied to financial performance and managerial practice, this indicator is rarely disclosed by heavily polluting firms, making it unsuitable for the present study. Second, from the R&D-input perspective, researchers use either the sum of R&D expenditure and technological-renovation outlays or the ratio of R&D expenditure to “three wastes” emissions to gauge green process innovation. While this approach links innovative investment to ecological performance, it does not adequately reflect the green innovation outputs of heavily polluting enterprises. Third, the patent-based approach counts green invention or utility-model patent applications or grants, capturing both product and process dimensions of green technological innovation and becoming the dominant measure in contemporary research [30]. Given that invention patents embody a higher inventive step than utility models and that granted patents more accurately represent realized innovation quality [31], we adopt the number of green invention patents granted in year t + 1 as a lagged proxy for green technological innovation in year t. Accordingly, the quantity of green invention patents authorized in 2023 represents firms’ green technological innovation performance for 2022. Patent data were compiled from the China National Intellectual Property Administration and cross-validated with green patent records available on the Chinese Research Data Services platform and the CSMAR database.

The organizational conditions encompass four types of talent: production, technical, sales, and management personnel. Talent generally refers to high-performance or high-potential employees. Inclusive talent management theory suggests that organizations should seek to activate the motivation and creativity of all employees [32]. Effectively utilizing existing human resources is crucial for cost reduction and efficiency improvement in heavily polluting firms. Therefore, this study considers all employees within the organization as potential talent. Functional departments bring together employees with specific skills, and the proportion of employees in each department reflects the relative number of skilled personnel within the organization. Thus, the proportion of employees in each category is used as a proxy variable for different types of skilled talent. The Wind database provides data on the proportions of employees in production, technical, sales, finance, and other departments based on the annual reports of listed companies. Since the number of employees in finance, human resources, administration, and general management is relatively small, listing all categories would result in significant data gaps and an excessive number of condition variables, which could hinder QCA analysis. Given that these personnel are generally non-frontline managers, human resources, administration, and other management-related personnel are consolidated to represent management talent.

The technological condition refers to the digital transformation of the enterprise. Digital transformation is a systematic process through which enterprises utilize digital technologies to digitize various elements and processes [33]. Common methods for measuring the extent of digital transformation in listed companies include text analysis and the total value of digital assets. Since digital transformation in enterprises is reflected through implemented digital projects, the total value of digital project assets not only reflects the level of organizational digitalization but also provides quantitative references for heavily polluting companies to make talent-related decisions based on the current value of digital assets. Therefore, this study uses the total value of digital assets related to digital projects as a measure of digital transformation. The data are sourced from the intangible assets section of the financial statement notes in the Guotai An database. Specifically, digital projects within the company’s intangible assets are manually collected, including production technologies such as HCCPR, human resource evaluation systems, and customer relationship management systems like SageCRM, covering various aspects such as production, human resources, and customer management. The total value of these digital assets is then calculated by summing the values of each digital project. Companies without any digital projects are recorded as having a digital asset value of 0.

The environmental antecedent condition is the level of regional digital-economy development. The digital economy—an emerging economic system in which data serve as the pivotal factor of production and digital technologies act as the principal engine of growth—is typically assessed using composite indicator systems. Consistent with this practice, we employ the China City Digitalization Development Index, compiled by the Digital China Research Institute of New H3C Group, to capture the digital maturity of each focal city. This index reflects both the robustness of local digital infrastructure and the richness of the digital service environment that supports firms’ green innovation processes.

3.3. fsQCA Procedure for Talent-Led Green Innovation

We employ the fsQCA approach to examine how multiple factors drive high green technological innovation under a dual digital context. As a set-theoretic method, fsQCA requires calibrating the raw data into fuzzy-set membership scores (ranging from 0 to 1) for all variables before analysis. Each case’s degree of membership in each relevant set (antecedent condition or outcome) must be determined through calibration. Calibration can be performed using either an indirect method or a direct method. The indirect calibration method assigns membership values between 0 and 1 to cases based on expert knowledge or practical experience. In contrast, the direct calibration method determines three qualitative anchor points—full membership, the crossover point, and full non-membership—based on both theoretical criteria and practical considerations, and then calibrates using algorithms provided by the fsQCA 3.0 software [34]. In this study, we prepare the data for fsQCA by applying the direct calibration method to convert the raw values of all variables (the talent and digitalization indicators as conditions, and green innovation as the outcome) into fuzzy-set membership scores. The following are the method steps for the direct calibration method:

(1)

Transformation of raw data into a single-case fuzzy-set matrix

We calibrate each variable using fsQCA’s direct method, which employs a logistic function to transform raw values into fuzzy-set scores between 0 and 1 [20]. The general form of the calibration function is

$$S(x)={\displaystyle \frac{1}{1+e^{-\alpha (x-c)}}}$$

(1)

where S(x) is the fuzzy set membership score, x is the raw value, c is the crossover (the point of maximum ambiguity, corresponding to a 0.5 membership), and α is a parameter that controls the rate of change around the crossover.

This logistic calibration model allows for a smooth and systematic assignment of membership scores, making the direct method more formalized and widely applied than the indirect method. Accordingly, we set the calibration anchor points for each condition variable based on the empirical distribution of the sample: the 95th percentile value as the threshold for full membership, the median (50th percentile) as the crossover point, and the 5th percentile value as the threshold for full non-membership. For the outcome variable—green technological innovation, measured by the number of granted green invention patents—we incorporate theoretical insight that obtaining at least one patent already signifies a moderate level of green innovation.

(2)

Construction of intuitionistic-fuzzy judgment matrices

For every condition set C_j and the outcome set Y, three indices are defined for case i:

$${\mu }_{ij}=S\left(C_{ij}\right)$$

(2)

$$v_{ij}=1-S\left(C_{ij}\right)$$

(3)

$${\pi }_{ij}=1-{\mu }_{ij}-v_{ij}$$

(4)

The intuitionistic-fuzzy judgment matrix for a given antecedent set C_j is more rigorously expressed as a three-column array that records, for every case i (i = 1, 2, …, n), its membership (${\mu }_{ij}$), non-membership ($v_{ij}$), and hesitation (${\pi }_{ij}$) degrees:

$$M_j=\left[\begin{array}{lll}{\mu }_{1j}\hfill & v_{1j}\hfill & {\pi }_{1j}\hfill \\ {\mu }_{2j}\hfill & v_{2j}\hfill & {\pi }_{2j}\hfill \\ ⋮\hfill & ⋮\hfill & ⋮\hfill \\ {\mu }_{nj}\hfill & v_{nj}\hfill & {\pi }_{nj}\hfill \end{array}\right]$$

(5)

This matrix M_j compactly summarizes the intuitionistic-fuzzy information required for subsequent necessity and sufficiency analyses in the fsQCA procedure.

(3)

Consistency testing and necessity analysis

Consistency calculation. We next conduct a necessity analysis for each individual antecedent condition to determine whether any single factor is indispensable for achieving the outcome. A condition is considered necessary for an outcome if its presence is required for the occurrence of that outcome. In fsQCA, the consistency of a necessary condition is calculated using the following formula:

$$Consistenc{y}_{necessity}={\displaystyle \frac{{\displaystyle \sum \min (X_i,Y_i)}}{{\displaystyle \sum Y_i}}}$$

(6)

In this equation, Xi represents the membership score of case i in the condition set and Yi represents the membership score of case i in the outcome set; the operator min(Xi,Yi); min selects the lower of the two scores for each case.

Threshold adjustment. This consistency measure ranges from 0 to 1, with higher values indicating that the outcome Y is almost always a subset of the condition X. In practice, a condition is typically considered almost always necessary if its consistency exceeds a high threshold (e.g., 0.90). We calculate the necessity consistency for each talent and digitalization condition (as well as for each condition’s absence) with respect to achieving high green technological innovation, and similarly for the absence of the high innovation outcome.

(4)

Identification of sufficient configurations and evaluation

Configuration analysis identifies which combinations of condition variables lead to the outcome, revealing the sufficient conditions for the outcome variable. In QCA, the consistency of sufficiency and coverage are calculated using the following formulas:

$$Consistenc{y}_{sufficiency}={\displaystyle \frac{{\displaystyle \sum \min (X_i,Y_i)}}{{\displaystyle \sum X_i}}}$$

(7)

$$Coverage={\displaystyle \frac{{\displaystyle \sum \min (X_i,Y_i)}}{{\displaystyle \sum Y_i}}}$$

(8)

where Xi is the membership score in the configuration and Yi is the membership score in the outcome for case i. The min(Xi,Yi) operator selects the lower of the two scores for each case.

Before conducting the analysis, the consistency threshold, frequency threshold, and PRI (Proportional Reduction in Inconsistency) must be set. The consistency threshold should be chosen based on the specific problem at hand, typically set between 0.80 and 0.85, although it should not be lower than 0.75. In this study, the consistency threshold is set at 0.85 [20]. The frequency threshold should be determined based on the number of cases, with the number of cases included not being lower than 75%. When the sample size is small, it may be set to 1, while a larger sample size may warrant a value greater than 1. Given that this study includes 96 cases, a relatively large sample size, the frequency threshold is set to 3.

4. Results and Analysis

4.1. Data Calibration

Thus, we set the crossover at one patent, while retaining the full membership and full non-membership thresholds at the sample’s 95th and 5th percentile values of patent counts. Using these anchors, all raw values for the skilled talent and digitalization indicators (antecedent conditions) and for the outcome were calibrated into fuzzy-set membership scores by the fsQCA 3.0 software. The calibration results and descriptive statistics for each variable are summarized in

Table 1. Variable calibration and descriptive statistics.

Variables	Calibration			Descriptive Statistics
	Full Membership	Crossover	Full Non-Membership	Mean	SD	Min	Max
Production personnel	83.35	68.54	28.27	65.09	17.59	0.00	85.98
Sales personnel	17.07	3.16	0.62	5.45	7.84	0.00	57.09
Technical personnel	28.96	12.53	5.16	15.02	10.67	0.00	81.54
Managerial personnel	29.57	11.60	3.98	14.45	11.98	1.81	90.00
Corporate digital transformation	52.33	1.92	0.00	12.12	33.85	0.00	296.52
Urban digital-economy development	89.40	64.60	40.15	65.17	14.24	36.80	90.50
Corporate green technological innovation	3.25	1.00	0.00	0.70	2.42	0.00	18.00

.

4.2. Necessity Analysis of Individual Conditions

We then interpret the necessity analysis results to identify any individual drivers of the outcome.

Table 2. Necessity analysis.

Antecedent Conditions	High		Non-High
	Consistency	Coverage	Consistency	Coverage
Production Personnel	0.7599	0.7755	0.7913	0.6209
Non-Production Personnel	0.6285	0.7966	0.7139	0.6957
Sales Personnel	0.6527	0.8283	0.7414	0.7236
Non-Sales Personnel	0.7822	0.7973	0.8241	0.6459
Technical Personnel	0.6505	0.7600	0.7572	0.6803
Non-Technical Personnel	0.7264	0.7956	0.7328	0.6172
Managerial Personnel	0.6619	0.7836	0.7050	0.6418
Non-Managerial Personnel	0.6974	0.7546	0.7623	0.6342
Corporate Digital Transformation	0.5725	0.8153	0.6468	0.7082
Non-Corporate Digital Transformation	0.7951	0.7454	0.8313	0.5992
Urban Digital Economy Development	0.7024	0.7845	0.7771	0.6674
Non-Urban Digital Economy Development	0.7022	0.8038	0.7491	0.6594

presents the consistency and coverage values for each antecedent condition as a potential necessary factor for achieving high green technological innovation (and for avoiding it). All conditions exhibit necessity consistency scores below 0.90 for the high-innovation outcome (and similarly all below 0.90 for the non-high outcome). Since the consistency of every condition variable is well under the 0.90 benchmark, it can be concluded that no single condition qualifies as a necessary condition driving high (or non-high) green technological innovation in the sampled firms. In other words, the presence of any given type of skilled talent or digital context factor by itself is not always observed in all high-performing cases. This finding implies that the influence of skilled talent on green innovation is realized through specific combinations of conditions rather than through any universally essential factor. Having found no individual condition to be necessary (see Table 2), we proceed to examine how different configurations of multiple conditions jointly produce the outcome in the sufficiency analysis (Section 4.3).

In summary, to examine whether there are necessary conditions for achieving high (or non-high) green technological innovation, a necessity analysis was conducted for each condition variable. The results are presented in Table 2. Since the consistency of all condition variables is below 0.9, it can be concluded that no necessary conditions are driving high (or non-high) green technological innovation in firms.

4.3. Sufficiency Analysis of Condition Configurations

To avoid contradictory configurations, the PRI consistency threshold is set to 0.75 [21]. The results of the configuration analysis are presented in

Table 3. Condition configuration of high-level green technological innovation.

Antecedent Conditions	Production-Worker Type		Managerial Type	Urban-Environment Type		Technical-Talent Type
	C1	C2	C3	C4	C5	C6
SC	●	●	●	${\displaystyle \otimes }$	${\displaystyle \otimes }$	●
XS	${\displaystyle \otimes }$	${\displaystyle \otimes }$	⊗	●	●	●
JS		${\displaystyle \otimes }$	${\displaystyle \otimes }$		●	${\displaystyle ●}$
GL	${\displaystyle \otimes }$	${\displaystyle \otimes }$	${\displaystyle ●}$	●	●	⊗
DIG	⊗	⊗		●		●
CITY	${\displaystyle \otimes }$		${\displaystyle \otimes }$	${\displaystyle ●}$	${\displaystyle ●}$	${\displaystyle \otimes }$
Consistency	0.9725	0.9608	0.9878	0.9927	0.9878	0.9662
Raw Coverage	0.4309	0.4244	0.3125	0.2261	0.3125	0.3368
Unique Coverage	0.0472	0.0486	0.0597	0.0293	0.0597	0.0389
Solution Consistency	0.9436
Solution Coverage	0.7619

Note: ${\displaystyle \otimes }$ indicates non-occurrence, ● indicates occurrence. Larger symbols represent core conditions (simple solution), smaller symbols represent peripheral conditions (intermediate solution). Empty cells indicate optional conditions.

. Table 3 reveals that both the overall solution and each of the six individual configurations achieve consistency scores greater than 0.90, confirming that every configuration represents a sufficient pathway to high green technological innovation. The overall solution coverage of 0.76 indicates that these six causal recipes collectively explain a substantial share of the firms exhibiting superior green technological innovation performance. Each configuration’s unique coverage is low, indicating minimal explanatory overlap; thus, the configurations address largely distinct sets of cases and function more as complements than as rivals. Based on their core conditions, the six configurations can be distilled into four broader archetypes.

Production Worker-Led Configuration (Configurations 1 and 2). This causal recipe suggests that when digital transformation fails to serve as an effective catalyst for green technological innovation, breakthrough advancements are primarily achieved through the efforts of production-line employees. In China, heavy-polluting firms are typically concentrated in capital-intensive sectors such as metal smelting, power generation, and coal mining. These firms often lack both the financial flexibility and risk tolerance required to undertake large-scale digital initiatives, resulting in their digital transformation capabilities being significantly weaker compared to those in more knowledge-intensive industries (e.g., internet services). Moreover, the long payback period for digital investments further discourages their use as immediate drivers of green innovation—hence the absence of the digital-transformation condition in this configuration. Confronted with these constraints, companies focus on incremental improvements in processes and products along existing production lines, relying on frontline production workers to innovate greener operations and reduce emissions. Sales personnel and managers provide basic functional support but play only a marginal role in the innovation process, as evidenced by their absence from the sufficiency solution. Therefore, we label this archetype “production worker-led,” reflecting the reality that some heavily polluting listed firms in China, constrained by limited resources, industry characteristics, and product properties, must primarily rely on shop-floor ingenuity to achieve green technological breakthroughs, thus failing to fully leverage the innovation-driving potential of digital transformation.

Proposition 1.

When the innovation-enabling effect of digital transformation is weak, production workers constitute the pivotal force driving green technological innovation.

Manager-Led Configuration (Configuration 3). This causal pathway suggests that, when an organization’s market-sensing capability is constrained, green technological innovation is driven by managerial personnel. Many heavy-polluting firms retain core customers and secure key contracts primarily through the relational capital of their senior managers. These managers then orchestrate product and process greening initiatives, directing production workers to implement cleaner operations. In the configurational terminology of QCA, managerial talent emerges as a core condition, whereas production staff constitute a peripheral condition. Sales talent is absent from the solution, signaling its limited contribution to the green innovation process. We therefore label this archetype manager-led. It reflects the reality that some pollution-intensive enterprises in China discover and exploit green-innovation opportunities primarily through managerial networks and leadership rather than sales-driven market intelligence, utilizing a talent-deployment pattern in which managers lead and production workers support.

Proposition 2.

When the market-scanning capability of sales talent is weak, managerial personnel become the central driver of green technological innovation, with production workers playing a complementary, supportive role.

Urban-Environment-Leveraging Configuration (Configurations 4 and 5). This causal pattern shows that, when frontline production staff lack the capability to generate green innovations, firms can still achieve high green technological innovation by capitalizing on the advantages offered by a digitally advanced urban environment. In these configurations, production talent is absent, indicating insufficient shop-floor innovation capacity. In contrast, a highly developed municipal digital economy emerges as a core condition: robust digital infrastructure and a competitive, data-rich market environment provide a fertile external context for green innovation. By exploiting city-level digital platforms, managers identify promising green business opportunities, while sales personnel dynamically track customers’ green preferences and communicate the firm’s eco-oriented value proposition to the market. Accordingly, managerial and sales talent are peripheral conditions that convert urban digital opportunities into firm-specific innovation inputs. A comparison of Configurations 4 and 5 reveals a substitution relationship between technical talent and firm-level digital transformation: when digital investment is limited, skilled technical staff can compensate for low organizational digitalization, whereas difficulties in recruiting or retaining technical talent can be offset by greater adoption of digital technologies.

Proposition 3.

When production workers’ innovation effectiveness is weak, capturing digital-economy opportunities becomes the core driver of green technological innovation. Managerial and sales personnel function as peripheral enablers of urban digital advantages, while technical talent and organizational digitalization act as substitutes in facilitating green innovation.

Technological-Talent-Driven Configuration (Configuration 6). This pathway indicates that, when managerial personnel lack the domain-specific expertise required to lead green technological innovation, superior performance is primarily achieved through technical talent. In heavy-polluting industries, the knowledge intensity of green innovation often exceeds the competencies of managers with non-technical backgrounds, making outsider-led innovation unfeasible; hence, managerial talent is absent from the sufficiency solution. By contrast, technically trained employees—who possess the requisite specialized knowledge—emerge as the core condition enabling green technological breakthroughs. Organizational digital-transformation projects equip these experts with big data, cloud computing, and other digital tools; production personnel implement and scale the technical staff’s innovations; and sales personnel both promote eco-oriented offerings and relay real-time market feedback, providing the latest customer insights. Accordingly, production talent, sales talent, and digital transformation are peripheral conditions that collectively reinforce the innovation efforts of technical personnel. We therefore label this archetype technological-talent-driven. It underscores that, when managerial leadership in green innovation is limited, firms can stimulate green technological progress through a talent-deployment pattern in which technical experts lead, while production and sales personnel, as well as digital infrastructure, provide complementary support.

Proposition 4.

When managerial personnel are unable to guide the organization toward green innovation, technical talent becomes the pivotal driver of green technological innovation, while sales talent, production workers, and organizational digital transformation serve as auxiliary conditions in the innovation process.

To intuitively compare the differentiated talent arrangement models of “different paths leading to the same goal,” the configurations outlined above are briefly illustrated in Figure 2. None of the four configurations require all types of skilled talent to be core conditions; instead, the presence of any specific type of talent is sufficient. This suggests that organizations can enhance green innovation by leveraging key talent breakthroughs and collaborative efforts from other employees, based on the digital resources available to the company and the region. Specifically, (1) the Production Worker Type indicates that when digitalization within the enterprise cannot drive innovation, production personnel play a key role in achieving high green technological innovation, representing a one-size-fits-all development model. This suggests that a talent advantage in a specific area is a significant driver of green technological innovation and highlights the importance of investing in and cultivating core skill talent in specialized firms. (2) The Managerial Type suggests that when both the enterprise and the region’s digitalization have minimal impact on green innovation, innovation can still be driven through the leadership of managers guiding production workers. (3) The Urban Environment Type suggests that when the enterprise’s digitalization capability is weak, but the city’s digital economy is developing rapidly, regional digital platforms should be fully utilized to promote green technological innovation, with an emphasis on the positive role of management and sales talent in leveraging the benefits of the digital economy. (4) The Technical Talent Type indicates that when the city’s digitalization is low, but the enterprise’s digital projects are operating well, the focus should be on developing the green innovation capacity of technical personnel, guiding production and sales personnel to cooperate actively. In conclusion, companies should strategically arrange various skilled talents based on their own and the region’s digitalization levels to drive green technological innovation.

We also explored configurational explanations for non-high green technological innovation; however, the analysis yielded PRI-consistency scores that fell not only below the 0.75 benchmark adopted earlier but also beneath the minimum acceptable threshold of 0.50 [35]. These results suggest that the candidate configurations simultaneously constitute subsets of both the outcome and its negation, reflecting the “same antecedents, different outcomes” (causal asymmetry) problem commonly encountered in QCA. Consequently, the antecedent conditions driving firms toward low levels of green technological innovation remain unclear and cannot be conclusively identified within the current analytical framework.

4.4. Theoretical Analysis of Configurations: The Diversity of Theory and Configurations

Based on the TOE framework, this study selects condition variables and conducts fsQCA analysis using resource orchestration, resource dependence, and IT capability theories. Subsequently, counterfactual reasoning is employed to propose four causal recipes for the theoretical conceptualization of the configurations [21]. These four configurations reflect the theoretical diversity and configurational variations.

First, configuration diversification refers to the idea that various configurations reflect the same theoretical framework. The four causal configurations involve three types of conditions: organization, technology, and environment. Specifically, different combinations of antecedent variables within the TOE framework can lead to high levels of green technological innovation. More specifically, Configurations 1, 2, and 6 lack managerial involvement, indicating that when managers struggle to promote green innovation, it can still be achieved through the deployment of production, technical, or sales personnel. In Configurations 1, 2, 3, and 6, when digital transformation or urban digital economy development is absent, appropriate organizational resource arrangements can compensate for disadvantages arising from insufficient IT capabilities or an unfavorable external environment. For Configuration 4, both IT capabilities and the external environment are present, but they work in tandem with management and sales personnel to drive green technological innovation. Additionally, Configurations 1, 2, and 3 suggest that although digitalization is absent in the firm or region, the strategic use of production or management personnel can still foster green technological innovation, reflecting resource orchestration theory. In Configurations 4 and 5, the external digital environment plays a central role in driving green technological innovation, which aligns with resource dependence theory. Finally, Configurations 4 and 6 demonstrate that digital transformation within firms is conducive to green technological innovation, supporting IT capability theory.

Second, theoretical diversity refers to the idea that a single configuration can simultaneously instantiate multiple theoretical logics. In Configuration 4, the city’s digital-economy maturity appears as the core condition, while sales talent, managerial talent, and firm-level digital transformation function as peripheral conditions. This constellation jointly reflects resource dependence theory (relying on external digital infrastructure), resource orchestration theory (mobilizing and coordinating internal human resources), and the IT capability perspective (leveraging proprietary digital assets). In Configuration 6, technical talent emerges as the core driver, with production workers, sales personnel, and digital transformation serving as peripheral enablers—a pattern that embodies both resource orchestration and IT capability logics.

4.5. Robustness Test

Altering the consistency cutoff or the calibration scheme reshapes the truth table and can therefore change the configurational solutions; consequently, such sensitivity analyses are widely regarded as standard procedures for testing the robustness of QCA results [21]. (1) Higher consistency threshold: To isolate only the most stringent subset relations, we increased the consistency benchmark from 0.85 to 0.90. The recomputed solution set (details available upon request) shows no substantive change in overall consistency or coverage, and the number and composition of configurations remain virtually identical, indicating strong robustness. (2) Alternative calibration anchors: We replaced the 95th- and 5th-percentile anchors used in the baseline calibration with the 90th and 10th percentiles. Although the overall solution consistency declined marginally, it still exceeded 0.90, and the resulting causal recipes were essentially the same as in the baseline model. Together, these checks confirm that the core findings are insensitive to reasonable variations in consistency thresholds and calibration parameters.

5. Discussion

5.1. Conclusions

This study reveals that in the context of dual digitalization, different configurations of skilled talent, interacting with firm-level and regional digitalization, can effectively drive green technological innovation in heavily polluting enterprises. Grounded in the TOE framework and informed by resource orchestration, resource dependence, and IT capabilities theories, the study applies fsQCA to identify four distinct and effective talent configurations, in which skilled talent in production, sales, technology, and management (organizational factors), interacting with firm-level digitalization (technology factor) and regional digitalization (environmental factor), contributes to promoting green technological innovation.

First, the study identifies four distinct talent configuration patterns (representing different effective system states within the TOE framework), characterized by “key skilled talent breakthroughs and the collaborative efforts of other employees,” that are associated with high levels of substantive green technological innovation. These findings underscore the configurational and systemic nature of green innovation success. Second, when both enterprise and regional digitalization have a relatively low innovation-driving effect, production and management talent become the core drivers of green technological innovation, particularly in the production worker support-oriented and management leadership-oriented models. Third, when the enterprise’s digitalization level is low, but the region’s digital economy is well-developed, regional digitalization becomes a critical driver (leveraging the environmental context) of green technological innovation, with management and sales talent playing supportive roles, as seen in the urban environment utilization-oriented model. Finally, when the city’s digitalization level is moderate, but the enterprise’s digitalization level is high, firm-level technological capability enables technical talent to lead the innovation process, with production and sales personnel actively cooperating, as in the technical talent-driven creation-oriented model.

These findings highlight the systemic, contingent, and configuration-dependent nature of talent–digital interactions in shaping green innovation success. Future research may explore how firms dynamically transition between these configurations as their internal capabilities and external environments evolve. Moreover, policy interventions—such as investments in regional digital infrastructure or targeted subsidies for digital talent development—may play a crucial role in helping firms align talent structures with digital contexts to sustain long-term green innovation performance.

5.2. Theoretical Implications

First, this study explores the configurations of skilled talent that drive high levels of green technological innovation within the dual digitalization context, thereby expanding research on the driving factors of green innovation. The positive impact of both enterprise and regional digitalization on green innovation has been widely acknowledged [1,2], making it crucial for firms to adopt appropriate strategies to capitalize on digitalization opportunities. This study examines the differentiated impact of various skilled talents on green technological innovation within both organizational and external digital contexts, offering a novel theoretical explanation for how firms can advance green technological innovation through talent configurations aligned with the dual digitalization context. Additionally, it enriches the theoretical framework of talent management and strategic human resource management from a digitalization perspective. It also responds to Zhang et al.’s [23] call for enterprises to promote green innovation development by enhancing technological advantages, organizational capabilities, and environmental adaptation.

Second, this study extends digital innovation theory through a configurational lens. Digital innovation refers to the transformative process of reshaping business processes and organizational structures through digital technologies [10]. Adopting a configurational perspective, this study explores the asymmetric effects of skilled talent in heavily polluting enterprises on green technological innovation within both enterprise and city digitalization contexts. This contributes to the academic call for a deeper investigation into digital organizational innovation and the multi-level support provided by digital technologies [36], thereby enhancing the explanatory power of digital innovation theory in the context of green innovation within pollution-intensive enterprises.

Third, this study expands the specific implications of the TOE framework and the three explanatory theories. It supports the view that organizational, technological, and environmental factors all play a significant role in green technological innovation [23], while also revealing the complementary or substitutive relationships between these factors. Specifically, when the innovation-driving effect of one factor is weak, other factors can be leveraged to achieve innovation objectives; when a factor exhibits strong innovation-driving capabilities, it still needs to be aligned with other factors to better enhance innovation performance. This not only refines the TOE analysis framework but also extends the resource dependence and IT capabilities theories in two ways: First, external environment or IT capabilities are not the sole determinants of innovation performance. If external resources are insufficient or IT capabilities are weak, a flexible talent configuration model can be employed to capitalize on strengths and minimize weaknesses. Second, neither external resources nor IT capabilities alone can explain the final outcomes of innovation activities. Even with favorable external resources and IT capabilities, selecting the appropriate talent configuration remains crucial to driving green technological innovation. Additionally, while resource orchestration theory primarily examines resource utilization from a managerial perspective [25], this study incorporates management personnel as a distinct variable in the analysis framework, discussing their role and position in green innovation. In doing so, it recognizes managers as an important resource that requires orchestration, thus broadening the analytical scope of resource orchestration theory.

5.3. Managerial Implications

This study provides configuration-specific managerial implications for promoting green technological innovation in heavily polluting enterprises under conditions of dual digitalization.

First, when both firm-level digital transformation and regional digital economy development are weak, firms should adopt a bottom-up innovation strategy led by production talent. Managers are advised to empower experienced frontline workers through mechanisms such as autonomous “green cells” focused on micro-level process improvements, performance incentives linked to carbon reduction, and routine codification of best practices for broader diffusion. In this context, production employees serve as the primary drivers of green innovation.

Second, under similarly low digital transformation and low regional digital economy conditions, but where internal coordination capacity is relatively strong, senior managers can assume a leadership role in innovation orchestration. This involves setting clear green technological innovation mandates at the strategic level, mobilizing cross-departmental task forces, and leveraging managerial networks to access external support (e.g., government subsidies, third-party digital solutions). Here, management talent plays a central role in compensating for weak technological environments.

Third, when firm-level digitalization is limited but the surrounding region has a well-developed digital economy, external digital resources become the key enabler. In such cases, management and sales personnel should act as boundary spanners who integrate regional digital infrastructure—such as open data platforms and smart industrial park services—into the firm’s green innovation processes. Practical steps include establishing partnerships with local digital platforms, training sales teams to leverage digital tools for green value communication, and embedding firms into regional innovation ecosystems.

Finally, when internal digital capabilities are strong but regional digital conditions are moderate, technical talent should be positioned at the core of the innovation process. Firms should establish internal digital-innovation platforms, such as AI-assisted green design labs or digital twin systems, and form agile cross-functional teams that include production and sales staff to accelerate commercialization and market feedback. In this configuration, firm-level digital assets and technical expertise jointly drive high-impact green innovation.

Across all configurations, managers are encouraged to adopt a structured “diagnose–configure–enable–evaluate” approach: (1) diagnose the firm’s digital–environmental context, (2) configure talent leadership and collaboration accordingly, (3) enable implementation through targeted tools, incentives, and training, and (4) continuously monitor and refine the configuration as internal and external conditions evolve. Policymakers can also support these efforts by aligning infrastructure investments and digital talent subsidies with the specific needs of firms operating under different digital–talent configurations, thereby enhancing the systemic capacity for sustainable innovation across regions.

5.4. Limitations and Future Research Directions

This study has certain limitations and areas for future improvement. First, this research analyzes four talent configuration patterns that drive green technological innovation in the context of dual digitalization from a configurational perspective, addressing the “what” and “how” questions. Future studies could explore the underlying mechanisms behind these talent configuration models using case studies, causal mediation, and other methods to answer the “why” question. For example, it can be examined whether factors such as the congruence between employees’ knowledge structures and organizational digitalization initiatives, as well as the firm’s stage in its lifecycle, influence the pivotal role of skilled personnel in green technological innovation. Second, while this study highlights the critical role of certain skilled talents, it may lead to negative perceptions, such as unfairness or power relinquishment, among other employees, potentially hindering the flow of tacit knowledge. Future research could explore how to encourage non-key skilled talents to actively engage in green innovation through approaches such as job rotation, joint compensation systems, and leader supervision. Third, due to data availability constraints, this study uses the proportion of employees in production, sales, and other departments as a proxy for different skilled talents. However, these departments also include managers within the skilled talent category. Future research could use surveys or other methods to separate those in skilled talent categories who are more involved in managerial tasks, thereby further examining the differences in the impact of frontline employees versus lower-level managers on green technological innovation. Finally, the measurement of digital transformation by the total value of digital assets constitutes another limitation. Future work should adopt more comprehensive frameworks that incorporate assessments of digital capabilities across technological, organizational, and cultural dimensions, or utilize qualitative case-based analyses. Such multidimensional approaches would provide a richer understanding of digital transformation and its actual effects.