The Impact of Supply Chain Co-Innovation on the Total Factor Productivity of SRDI Enterprises: The Mediating Mechanism of Corporate ESG Performance

Abstract

This study investigates how supply chain co-innovation affects the high-quality development of SRDI enterprises, operationalized as total factor productivity (TFP) measured by the LP method. The mechanism remains unclear. Drawing on learning-by-doing theory, together with the resource-based view and stakeholder theory, we propose a sequential pathway: through repeated interactions and knowledge accumulation in collaborative innovation, SRDI enterprises improve their ESG performance, which in turn enhances TFP. Using a sample of listed “little giant” SRDI enterprises from 2018 to 2023, we find that supply chain co-innovation is significantly positively associated with TFP (coefficient 0.003), and the pattern is consistent with ESG performance playing a partial mediating role. Meanwhile, mechanistic analysis also reveals that this correlation is more pronounced in high-profitability enterprises and manufacturing enterprises. This research provides theoretical guidance for SRDI enterprises in choosing innovation models and managing supply chains, offering practical insights for improving total factor productivity.

1. Introduction

SRDI enterprises seek to establish a commanding presence in specific niche markets or technological domains [1,2]. Strategically, these enterprises aim to consistently enhance their specialization to obtain a leading industry position by utilizing differentiation advantages within a lean production framework [3]. A key function of SRDI enterprises is to bolster the robustness and completeness of industrial chains, which in turn stabilizes them and contributes to higher TFP improvement [4,5]. To bolster their innovative capacity and specialization, the government introduced central fiscal funding in June 2024 to support the TFP improvement of SRDI enterprises.

However, the success of SRDI enterprises depends not only on their internal R&D capabilities but also on how effectively they integrate external resources. As crucial entities addressing “choke point” technologies in critical core areas and reinforcing industrial chains, SRDI enterprises have established a virtuous interactive relationship with upstream and downstream supply chains through technological innovation-driven approaches [6]. Yet many SRDI enterprises, especially in their early stages, face resource constraints, limited innovation pathways, and insufficient R&D capacity [7]. Relying solely on internal capabilities is not sufficient to sustain growth. By building external collaboration networks, firms can access critical resources such as technology, knowledge, and information [8,9].

With the high integration of supply chains in today’s globalized economy, enterprises now function not as isolated entities but as tightly interwoven clusters within a supply chain network [10,11]. Within the framework of advancing industrial chain modernization under the 14th Five-Year Plan period, policy directives emphasize the need to “promote integrated innovation across upstream, midstream, and downstream industrial chains, as well as between large and small-medium enterprises.” This collaborative model, in which enterprises integrate and complement resources, share knowledge, and create value with external partners, is defined as supply chain co-innovation [12]. Unlike general technological innovation or closed innovation, supply chain co-innovation emphasizes repeated interactions, mutual learning, and knowledge accumulation among chain partners over time.

Despite the crucial role of supply chain collaboration, most existing research on SRDI enterprises has focused on internal drivers such as technological innovation [13], digital intelligence [3], and digital transformation [14]. A growing literature has explored pathways for promoting enterprise TFP improvement from other perspectives, including tax incentives [15], green supply chain [16], digital financial inclusion [17], environmental regulations [18], and stock market information efficiency [19]. However, far less attention has been given to the factors influencing SRDI enterprises‘ development from a supply chain co-innovation standpoint. Specifically, three gaps remain. First, current research on supply chain co-innovation remains limited, and systematic studies on the specific mechanisms through which it affects enterprise TFP improvement are still lacking [12,20]. Second, the potential mediating role of corporate ESG performance, which links collaborative innovation to sustainable development, has not been theoretically or empirically examined in the context of SRDI enterprises [21,22]. Third, the concept of “high-quality development” is often loosely tied to policy rhetoric without a clear operational definition, making it difficult to measure and compare across studies [13].

In this paper, we address these gaps by investigating how supply chain co-innovation affects the TFP improvement of SRDI enterprises. We operationalize TFP improvement as total factor productivity (TFP) measured by the LP method, which reflects resource allocation efficiency and core competitiveness. Drawing on learning-by-doing theory, which posits that productivity gains arise from repeated interactions and knowledge accumulation through practice, together with the resource-based view [23] and stakeholder theory [24], we propose a sequential pathway. Through iterative collaboration and knowledge accumulation in supply chain co-innovation, SRDI enterprises improve their ESG performance, which in turn is associated with higher TFP. In other words, corporate ESG performance serves as a key mediating channel linking supply chain co-innovation to TFP improvement. Furthermore, we examine how firm profitability and manufacturing sector heterogeneity moderate this pathway.

To test these hypotheses, we construct a novel dataset of listed “little giant” SRDI enterprises from 2018 to 2023. We measure supply chain co-innovation using joint patent applications and new product sales ratios, ESG performance using the Huazheng ESG rating, and TFP using the LP method. We employ panel fixed-effects regressions, propensity score matching, and lagged explanatory variables to address endogeneity concerns.

First, we contribute to the literature on SRDI enterprises and TFP improvement. While existing studies have examined technological innovation and digital transformation as drivers [13,14], we shift the focus to supply chain co-innovation as an external resource integration mechanism. By showing that co-innovation with supply chain partners compensates for SRDI enterprises’ internal resource constraints, we provide a new theoretical lens, rooted in the resource-based view, for understanding how these specialized firms achieve productivity gains [25,26].

Second, we expand the literature on the real effects of ESG performance. Prior research has documented that ESG practices contribute to corporate sustainability [24], but little is known about how supply chain collaboration shapes ESG performance and whether this channel translates into productivity. Drawing on stakeholder theory, we identify ESG performance as a mediating mechanism through which supply chain co-innovation drives TFP [27,28]. This finding bridges the supply chain management and ESG studies, showing that collaborative innovation not only yields direct efficiency gains but also enhances environmental, social, and governance outcomes that further boost productivity [29].

Third, we introduce learning-by-doing theory into supply chain co-innovation research [30,31]. Unlike static resource-based views that emphasize resource endowments, learning-by-doing highlights that repeated interactions and knowledge accumulation over time are the engine of productivity growth [32]. Our theoretical framework, which moves from co-innovation (iterative collaboration) to ESG improvement (cumulative learning outcomes) and then to TFP (productivity gains), provides a dynamic explanation for how supply chain partnerships generate sustainable competitive advantages [33]. This extends the learning-by-doing literature from manufacturing productivity to inter-organizational collaboration and ESG performance, offering a unified micro-foundation for understanding co-innovation’s long-run benefits [34,35].

2. Theoretical Analysis and Research Hypotheses

2.1. Supply Chain Co-Innovation, Learning-by-Doing, and TFP Improvement of SRDI Enterprises

With the global shifts and China’s rising position in the industrial division of labor, vulnerabilities in industrial and supply chains have emerged, which pose latent risks to industrial security. SRDI enterprises address bottlenecks through technological innovation, optimizing industrial chains and stabilizing supply chains, thereby serving as critical supports to overcome these weaknesses. As China moves toward innovation-driven high-quality growth, it is crucial to create a group of specialized “little giant” enterprises that are focused on niche sectors. These enterprises should be guided to play a crucial role in strengthening industrial chains, bridging gaps, securing supply chains, and enhancing economic efficiency is crucial [5]. Industrial transformation and upgrading in China have been strategically advanced in recent years by leveraging SRDI enterprises as a key driver [36].

In their early stages, SRDI enterprises often face challenges such as limited innovation pathways, insufficient R&D capacity, and resource constraints, leading to “bottleneck” risks that increase operational costs and instability [12]. Relying solely on internal capabilities is not sufficient to sustain growth. By building external collaboration networks, firms can access critical resources such as technology, knowledge, and information [7], thereby supporting their sustainable development.

We anchor our theoretical framework in learning-by-doing theory, originally formalized by Arrow [30]. Learning-by-doing posits that productivity gains arise from the accumulation of experience through repeated production activities. Arrow attributed increased productivity to learning that occurs as a byproduct of production: increases in cumulative experience lead to increased productivity [31]. This insight has since been extended from manufacturing productivity to inter-organizational contexts, where repeated interactions and accumulated experience generate knowledge that improves performance over time [32].

In the context of supply chain co-innovation, learning-by-doing operates through iterative collaboration between SRDI enterprises and their supply chain partners. Each cycle of joint problem-solving, resource integration, and knowledge exchange generates experience-based knowledge that accumulates over time. As partners engage repeatedly in co-innovation activities, they improve their ability to identify collaborative opportunities, coordinate actions, and adapt to shared goals. This learning process reduces transaction costs, accelerates information flow, and enhances the efficiency of resource allocation [33,34].

Unlike static resource-based views that emphasize resource endowments, learning-by-doing highlights that repeated interactions and knowledge accumulation over time are the engine of productivity growth. Supply chain co-innovation enables SRDI enterprises to access heterogeneous complementary resources and, through cumulative learning, convert these resources into higher total factor productivity (TFP) [23,36,37]. TFP reflects resource allocation efficiency and core competitiveness, and is widely used as a measure of enterprise productivity improvement.

Moreover, supply chain co-innovation transforms supply chain actors into tightly knit communities of shared interest. Guided by value co-creation, market entities optimize resource allocation and build information sharing platforms, effectively reducing institutional transaction costs and establishing long-term trust assurance mechanisms [9]. When SRDI enterprises face external risks, the stable cooperative network built through co-innovation generates a significant multiplier effect, providing effective crisis response support and fostering a steady rise in core competitiveness and production efficiency. Enhanced supply chain coordination yields synergistic benefits: mutual trust is strengthened, production factors and commercial information flow faster, and informational barriers are dismantled, resulting in lower vertical transaction costs [10,20,22]. Innovation is crucial for overcoming technological challenges and is the primary driving force behind TFP improvement. We thus propose:

H1. 

Supply chain co-innovation positively impacts the TFP improvement of SRDI enterprises.

2.2. The Mediating Role of Corporate ESG Performance: A Stakeholder Theory Perspective

Corporate ESG performance encompasses three dimensions: environmental (E), social (S), and governance (G) [38]. The environmental component reflects achievements in pollution control, energy management, and ecological conservation; the social dimension covers employee welfare, product and service quality, shareholder rights, and community relations; governance coordinates stakeholder interests through institutional design to achieve pluralistic value [39].

We draw on stakeholder theory as the unifying theoretical lens for explaining why ESG performance serves as a mediating mechanism. According to stakeholder theory, the long-term success of a firm depends on its ability to balance and respond to the demands of multiple stakeholders, including shareholders, creditors, employees, suppliers, customers, governments, environmental groups, and the public [40]. Firms that effectively address these diverse claims are more likely to secure critical resources, build legitimacy, and achieve sustainable performance. Supply chain co-innovation, through the lens of learning-by-doing, enables SRDI enterprises to improve their ESG performance along all three dimensions. The iterative, experience-accumulating nature of co-innovation generates knowledge and capabilities that directly respond to stakeholder expectations [41].

Environmental dimension (E). Green innovation has a higher risk premium and volatility than conventional innovation. Through repeated collaboration with supply chain partners, SRDI enterprises accumulate knowledge about upstream environmental technologies and downstream green market trends. Decision making coordination and information exchange at various stages of the supply chain significantly enhance the precision and relevance of green innovation [42]. This learning-by-doing process spurs green patents, improves environmental performance, and boosts ESG metrics, thereby addressing the demands of environmental stakeholders [43].

Social dimension (S). Supply chain co-innovation integrates the strengths of multiple parties, allowing efficient resource allocation and rapid technology transfer. Through cumulative learning, SRDI enterprises develop better products, improve service quality, and enhance worker safety and community relations. These outcomes respond to the claims of customers, employees, and local communities. Moreover, ESG disclosure eases information asymmetries between firms and stakeholders [44], helping companies accumulate social capital and promote TFP improvement.

Governance dimension (G). Companies need efficient innovation outputs to respond promptly to market demands. Innovation is fundamental to ensuring the long-term viability of SRDI enterprises [45]. From the stakeholder perspective, long-term-oriented stakeholders prioritize sustainable outcomes. Through repeated co-innovation, SRDI enterprises develop transparent governance structures, improve accountability, and reduce opportunistic behavior. The learning-by-doing process helps firms internalize governance best practices from their supply chain partners, thereby enhancing governance performance [46].

Thus, the learning-by-doing inherent in supply chain co-innovation enables SRDI enterprises to progressively improve their ESG performance across all three dimensions [47]. Improved ESG performance, in turn, enhances TFP by reducing financing constraints, lowering risk, attracting long-term investors, and improving operational efficiency. We therefore propose:

H2. 

Corporate ESG performance mediates the relationship between supply chain co-innovation and TFP improvement of SRDI enterprises.

2.3. Moderating Roles: Profitability and Manufacturing Heterogeneity

2.3.1. Moderating Role of Corporate Profitability

The learning-by-doing effect in supply chain co-innovation does not occur uniformly across firms; rather, it hinges on the resources that firms are able to allocate to collaborative learning. Drawing on the resource-based view, we argue that profitability, which reflects the capacity to generate earnings, supplies the slack resources required to intensify and deepen the learning process [48].

Firms with strong profitability have more abundant financial and organizational slack [49,50]. These surplus resources can be deployed in two ways. First, profitable firms can invest more in co-innovation activities, such as joint R&D projects, information sharing systems, and partner relationship management. Greater investment increases the frequency and depth of interactions with supply chain partners, thereby accelerating the accumulation of experience-based knowledge [49]. Second, profitable firms are better positioned to translate the knowledge gained from co-innovation into tangible TFP improvements [51].

Therefore, profitability positively moderates the direct effect of supply chain co-innovation on TFP performance. We propose:

H3. 

Corporate profitability positively moderates the direct relationship between supply chain co-innovation and TFP improvement, such that the positive effect is stronger for firms with higher profitability.

2.3.2. Moderating Role of Manufacturing Industry

The manufacturing sector differs from non-manufacturing sectors in several ways that amplify the learning-by-doing effect of supply chain co-innovation. Manufacturing typically involves long and complex production chains, requiring coordination across multiple stages. This complexity creates greater demand for supply chain integration and collaborative innovation [52]. The repeated interactions necessary for learning-by-doing are more frequent and more critical in manufacturing because any disruption or inefficiency in one stage propagates through the entire chain [53].

Moreover, manufacturing enterprises face stricter environmental regulations and greater social responsibility pressures. These pressures make TFP performance particularly salient for manufacturers. In non-manufacturing sectors, supply chain structures are often simpler, regulatory pressures on environmental and social dimensions are lower, and the marginal benefit of co-innovation for TFP may be smaller [54].

Empirically, we expect that the positive impact of supply chain co-innovation on TFP improvement will be more pronounced in manufacturing enterprises [45]. We thus propose:

H4. 

The positive effect of supply chain co-innovation on TFP improvement is stronger for SRDI enterprises in the manufacturing sector than for those in non-manufacturing sectors.

3. Research Design

3.1. Model Specification

This paper employs a baseline regression model to analyze the effect of supply chain co-innovation on the TFP of SRDI enterprises:

$$TF{P\_LP}_{it}={\beta }_0+{\beta }_1{Sci}_{it}+{\beta }_2{ESG}_{it}+{\sum }_{k=1}^K{\gamma }_{k}Control{s}_{k,it}+{\mu }_{year}+{\mu }_{ind}+{\mu }_{region}+{\epsilon }_{it}$$

(1)

where $Control{s}_{k,it}$ denotes control variables, ${\epsilon }_{it}$ denotes the error term, and ${\mu }_{year}$, ${\mu }_{ind}$ and ${\mu }_{region}$ represent year, industry, and region fixed effects, respectively. Our model does not include firm fixed effects; therefore, identification relies primarily on cross-sectional variation across firms. The findings should be interpreted as a strong correlation between supply chain collaborative innovation and TFP, rather than a strict causal inference. All regressions control for year, industry, and region fixed effects.

3.2. Variable Definitions and Measurement

3.2.1. Total Factor Productivity (TFP_LP)

Total factor productivity (TFP) reflects a company’s core competitiveness and resource allocation efficiency. Methods like DEA, SFA, and OLS are commonly used in existing studies to compute TFP. Given the dynamic nature of TFP changes and the econometric challenges such as endogeneity bias and sample selection bias, this study employs the LP method for measuring total factor productivity, as referenced in the research by Levinsohn and Petrin (2003) [55]. This method uses intermediate inputs as a proxy variable for unobserved productivity shocks, which can effectively mitigate the bias arising from endogenous input selection in the estimation of production functions. We employ the OP method for variable substitution to verify the robustness of the results.

3.2.2. Supply Chain Co-Innovation (Sci)

This study draws on the research of Zhang and Li and Huang et al. to measure supply chain co-innovation from two dimensions [56,57]. First, from the perspective of integration degree, the indicator is the fusion of innovation outputs, measured as the natural logarithm of one plus the number of patents jointly filed with enterprises along the supply chain. Second, from the perspective of connectivity degree, the indicator is the transformation of innovation outcomes, measured as the ratio of new product sales revenue to the number of patent applications in the firm’s location. Given that integration degree and connectivity degree are measured in different units and scales, this study adopts the entropy weight method to calculate their objective weights and further constructs a comprehensive supply chain co-innovation index. Additionally, adopting the methodology of Zhang and Li [56], the supply chain co-innovation indicators processed by the entropy method are expanded by 100 times.

3.2.3. Corporate ESG Performance (ESG)

At present, there are significant differences in the evaluation indicators, standards, and backgrounds of ESG among various domestic and international rating agencies. The academic community has also not yet reached a consensus on the selection of ESG evaluation indicators. This paper, referring to the research by Ren et al. [58], uses the Huazheng ESG rating to measure corporate ESG performance. The Huazheng ESG rating focuses on the three core dimensions of environment, social, and governance, selecting 14 themes and 26 key indicators to construct a localized ESG evaluation system. Its nine-tier rating scale, from “C” to “AAA”, is coded numerically from 1 to 9 (e.g., C = 1, CC = 2). A higher ESG score indicates better ESG performance. In addition, this study uses the average quarterly score to measure annual ESG performance1.

3.2.4. Moderating Variables

Enterprise Profitability (Prof). The profitability of a company refers to its ability to generate earnings. When a company’s profitability is strong, it usually indicates that it has more financial resources and occupies a more advantageous market position [59]. This enables firms to allocate more resources to innovation activities and seek external partnerships, thereby fostering their sustainable development. In accordance with Nanda and Panda [60], we use the average of the return on assets and net profit margin to measure enterprise profitability.

Manufacturing Enterprise Dummy Variable (Manu). The manufacturing industry often involves complex production processes that require collaboration across multiple stages, efficient resource allocation, strict control over production, and adherence to high-quality standards and regulations. As a result, the need for co-innovation within the supply chain and its impact are particularly significant. To further explore how supply chain co-innovation relates to total factor productivity across these two types of enterprises, we construct a dummy variable, where manufacturing firms are coded as 1 and non-manufacturing firms are coded as 0.

3.2.5. Control Variables

We included firm age (Age), company size (Size), CEO duality (Dual), board size (Board), Tobin’s Q (Qa), leverage ratio (Lev), and ownership concentration (Ownhold) as control variables. Additionally, we employ industry fixed effects, regional fixed effects, and time fixed effects to control for other unobservable factors in the model that do not vary with industry, region, and time.

Table 1. Variable definitions and measurement methods.

Variable Type	Variable Symbol	Variable Name	Measurement Method
Dependent Variable	TFP_LP	Total factor productivity	LP method
Independent Variable	Sci	Supply Chain Co-innovation	The fusion of innovation outputs and the transformation of innovation outcomes
Mediating Variable	ESG	Corporate ESG Performance	Huazheng ESG rating (C–AAA) is coded 1–9, with annual values calculated as quarterly averages.
Moderating Variables	Prof	Enterprise Profitability	The average of return on total assets and net operating margin
	Manu	Manufacturing Enterprise Dummy Variable	Manufacturing firms are coded as 1, and non-manufacturing firms are coded as 0
Control Variables	Age	Firm Age	Natural logarithm of the number of years since the company was listed
	Size	Firm Size	Natural logarithm of total assets
	Dual	Dual Role	1 if the chairman and CEO positions are held by the same person, otherwise 0
	Board	Board Size	Natural logarithm of the number of board directors
	Qa	Tobin’s Q	Market value/Total assets
	Lev	Debt Ratio	Total debt/Total assets
	Ownhold	Ownership Concentration	Shares held by the top three shareholders/Total shares

contains detailed measurement information for these variables.

3.3. Data Sources

“Little Giant” enterprises, recognized as champions among SRDI enterprises, are leading players with exceptional and sustained innovative capabilities, mastering key technologies within their respective fields. Accordingly, we focus on “Little Giant” companies listed from 2018 to 2023 as our research sample. After excluding samples with incomplete data, the final dataset consists of 190 companies and 1087 observations. This study primarily draws on data from CNRDS, CSMAR, and Wind.

4. Results

4.1. Descriptive Statistics

As presented in

Table 2. Descriptive statistics.

Variable	Observed Value	Mean	Standard Deviation	Minimum	Maximum
TFP_LP	1087	7.856	0.673	6.139	11.173
Sci	1087	7.088	11.096	0.000	65.138
ESG	1087	4.288	0.762	1.000	6.750
Prof	1087	0.064	0.119	−0.772	0.575
Manu	1087	0.928	0.258	0.000	1.000
Size	1087	21.508	0.709	19.714	24.578
Age	1087	1.681	0.685	0.000	3.258
Dual	1087	0.443	0.497	0.000	1.000
Board	1087	2.024	0.197	1.386	2.398
Qa	1087	2.364	1.793	0.884	22.557
Lev	1087	0.294	0.157	0.006	0.773
Ownhold	1087	46.123	12.630	11.473	75.000

, the variable for total factor productivity (TFP_LP) exhibits a mean of 7.856 and a standard deviation of 0.673. The observed values fall between 6.139 and 11.173, reflecting pronounced heterogeneity in the development levels across the sample enterprises. The minimum and maximum values of supply chain co-innovation are 0 and 65.138, respectively, indicating significant differences in how various SRDI enterprises engage in co-innovation with companies within the chain.

4.2. Regression Analysis

4.2.1. Direct Effect Analysis

With the regression results provided in

Table 3. Results of regression analysis.

	m1	m2	m3	m4	m5	m6
	TFP	ESG	TFP(High Profitability)	TFP(Low Profitability)	TFP(Manufacturing Enterprises)	TFP(Non-Manufacturing Enterprises)
Sci	0.003 ***	0.006 ***	0.005 ***	0.001	0.004 ***	0.002
	(0.001)	(0.002)	(0.002)	(0.002)	(0.001)	(0.006)
Size	0.559 ***	0.012	0.604 ***	0.549 ***	0.570 ***	0.119
	(0.026)	(0.042)	(0.030)	(0.043)	(0.025)	(0.144)
Age	0.055 *	−0.125 **	−0.108 ***	0.113 **	0.053 *	−0.196
	(0.029)	(0.048)	(0.034)	(0.050)	(0.029)	(0.251)
Dual	−0.025	0.041	−0.039	−0.014	−0.005	−0.554 ***
	(0.028)	(0.047)	(0.032)	(0.044)	(0.028)	(0.176)
Board	−0.102	−0.350 ***	−0.015	0.036	−0.050	1.018
	(0.071)	(0.118)	(0.080)	(0.112)	(0.070)	(0.649)
Qa	0.034 ***	−0.011	0.043 ***	−0.027	0.032 ***	0.073
	(0.008)	(0.014)	(0.008)	(0.025)	(0.008)	(0.060)
Lev	0.749 ***	−0.577 ***	0.359 ***	0.661 ***	0.743 ***	2.523 ***
	(0.097)	(0.161)	(0.126)	(0.155)	(0.095)	(0.842)
Ownhold	0.005 ***	0.002	0.002	0.007 ***	0.007 ***	−0.058 ***
	(0.001)	(0.002)	(0.001)	(0.002)	(0.001)	(0.010)
Industry	Yes	Yes	Yes	Yes	Yes	Yes
Region	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
Constant	−4.431 ***	5.008 ***	−5.240 ***	−4.367 ***	−4.945 ***	6.894 **
	(0.552)	(0.914)	(0.660)	(0.919)	(0.542)	(3.157)
Observations	1087	1087	604	483	1009	78
R-squared	0.602	0.148	0.637	0.703	0.628	0.743
r2_a	0.587	0.116	0.614	0.678	0.614	0.653

Note: *** p < 0.01, ** p < 0.05, * p < 0.1, double-tail test.

, we can find that the majority of control variables exert statistically significant influences on TFP, confirming the rationality and effectiveness of the control variables selected in this study. With respect to the productivity implications of supply chain collaboration, Alfaro Urena [61] documented that domestic firms joining multinational supply chains in Costa Rica achieve a 4% to 9% higher total factor productivity (TFP) four years after entry. Similarly, in the context of China‘s renewable energy sector, participation in supply chain alliances has been found to increase firm-level productivity by approximately 9.6% [62]. Our coefficient of 0.003 for supply chain co-innovation on TFP reflects the marginal change in TFP associated with a one unit increase in the co-innovation measure. Although the magnitude appears modest in absolute terms, the estimate is statistically significant at the 1% level, reinforcing the robustness of the positive relationship between supply chain co-innovation and TFP, thus providing evidence consistent with H1.

4.2.2. Mechanisms Analysis

The Mediating Role of ESG

Building on the baseline regression, a mediation effect test was performed. Model 2 is the regression model including control variables and supply chain co-innovation on corporate ESG performance. The analysis reveals a strong positive association between supply chain co-innovation and corporate ESG performance, meaning that SRDI enterprises collaborating with chain enterprises on co-innovation improves corporate ESG performance, which in turn partly explains the promotion of total factor productivity. This finding is consistent with a mediating pathway and provides empirical support for hypothesis H2.

The Moderate Effects of Corporate Profitability

To evaluate whether the relationship between supply chain co-innovation and total factor productivity varies among firms with different profitability levels, we categorize the sample into high and low profitability groups and conduct separate regressions. Based on the estimates in Table 3 (m3 and m4), supply chain co-innovation serves as a positive driver for the total factor productivity of highly profitable SRDI firms (α1 = 0.005, p < 0.01), whereas this effect attenuates and loses statistical significance at lower levels of profitability. This finding supports the theoretical expectation that financial slack and market advantages associated with higher profitability amplify the returns to collaborative innovation activities.

The Moderate Effects of the Manufacturing Industry

To explore how supply chain co-innovation relates to total factor productivity across these two types of enterprises, we divided the sample accordingly and conducted separate regression analyses. As illustrated in m5 and m6 of Table 3, for SRDI enterprises in the manufacturing sector, supply chain co-innovation significantly promotes their total factor productivity (α1 = 0.004, p < 0.01), whereas this effect is attenuated and fails to reach statistical significance for firms in the non-manufacturing sector. This finding supports the theoretical expectation that the inherent complexity and regulatory intensity of manufacturing supply chains heighten the marginal returns to collaborative innovation activities.

4.3. Endogeneity Analysis

A key empirical challenge in identifying the causal effect of supply chain co-innovation on the high-quality development of SRDI enterprises is the presence of endogeneity. First, reverse causality is a salient concern. As noted in our theoretical discussion, firms that are more advanced or already experiencing high-quality development may be more inclined and better positioned to engage in collaborative innovation with supply chain partners to acquire resources and further enhance their competitiveness. This creates a bidirectional relationship wherein TFP improvement could simultaneously drive co-innovation activities. Second, omitted variable bias may arise from unobserved firm-level characteristics that influence both co-innovation engagement and productivity outcomes. For instance, superior managerial ability or an inherently innovative corporate culture could lead an SRDI enterprise to pursue both supply chain collaboration and internal efficiency improvements, generating a spurious correlation between the two. Third, sample selection bias is a potential issue, as the decision to engage in supply chain co-innovation is unlikely to be random; firms with specific resource endowments or strategic orientations may self-select into collaborative partnerships. To address these concerns, we employ propensity score matching, one-period lagged core explanatory variables and the instrumental variable method for further verification.

4.3.1. Propensity Score Matching (PSM)

With the aim of minimizing the impact of sample selection bias on our findings, this paper refers to the study by Rosenbaum and Rubin [63] and employs the propensity score matching (PSM) for testing [64]. Treat control variables as covariates and divide the samples into treatment and control groups based on the average value of supply chain integration innovation. To assess the effect of supply chain co-innovation on high-quality development of enterprises, regression analyses were conducted employing 1:4 and 1:1 nearest neighbor matching, respectively. In

Table 4. Results of endogeneity analysis.

	(1)	(2)	(3)
	m1	m2	m3
	TFP_LP	TFP_LP	TFP_LP
Sci	0.003 **	0.004 **
	(0.001)	(0.002)
L.Sci			0.003 **
			(0.001)
Size	0.535 ***	0.568 ***	0.559 ***
	(0.029)	(0.041)	(0.026)
Age	0.092 ***	0.024	0.060 **
	(0.034)	(0.047)	(0.029)
Dual	−0.017	−0.030	−0.025
	(0.032)	(0.044)	(0.028)
Board	−0.040	−0.048	−0.107
	(0.082)	(0.118)	(0.071)
Qa	0.018	0.039 **	0.034 ***
	(0.012)	(0.016)	(0.008)
Lev	0.703 ***	0.791 ***	0.744 ***
	(0.114)	(0.158)	(0.097)
Ownhold	0.005 ***	0.005 ***	0.005 ***
	(0.001)	(0.002)	(0.001)
Industry	Yes	Yes	Yes
Region	Yes	Yes	Yes
Year	Yes	Yes	Yes
Constant	−4.071 ***	−4.603 ***	−4.425 ***
	(0.622)	(0.871)	(0.553)
Observations	848	441	1086
R-squared	0.608	0.640	0.599
r2_a	0.590	0.605	0.584

Note: *** p < 0.01, ** p < 0.05, double-tail test.

, m1 represents the regression results of 1:4 nearest neighbor matching, and m2 represents the regression results of 1:1 nearest neighbor matching. The significantly positive coefficient for supply chain co-innovation corroborates the robustness of this study’s main findings.

4.3.2. Addressing Lagged Effects

Supply chain co-innovation influences the high-quality enterprise development of SRDI firms through factors like corporate ESG performance. In contrast, firms that are more advanced may be more inclined to work with supply chain partners to acquire resources, improve innovation performance, and increase competitiveness. This could suggest a bidirectional causality, where the high-quality development of SRDI enterprises may, in turn, affect supply chain co-innovation. Following Autry and Golicic [65], we mitigate reverse causality endogeneity through a one-period lag of the core explanatory variable. As shown in m3 of Table 4, the result also documents that supply chain co-innovation significantly promotes TFP.

4.3.3. Instrumental Variable Method

To address endogeneity concerns arising from reverse causality and omitted variables, we employ the one-period lagged value of the firm’s own supply chain co-innovation as the instrumental variable. The lagged instrument leverages temporal precedence, as past co-innovation is correlated with current co-innovation but predetermined relative to current productivity shocks [66]. The instrument provides within-firm temporal variation that may help to mitigate potential biases, offering some support for relationship interpretation. The two-stage statistical results in

Table 5. Results of instrumental variable method.

	(1) First Stage	(2) Second Stage
	(Dependent Variable: Sci)	(Dependent Variable: TFP_LP)
Sci		0.004 * (0.002)
L.Sci	−0.454 * (0.037)
Size	0.992 (2.985)	0.635 *** (0.057)
Age	11.706 ** (5.192)	−0.216 ** (0.101)
Dual	−2.644 * (1.564)	0.006 (0.030)
Board	−2.182 (5.217)	−0.010 (0.099)
Qa	−0.557 (0.341)	0.038 *** (0.007)
Lev	−12.598 * (6.405)	0.141 (0.124)
Ownhold	0.318 ** (0.144)	−0.004 (0.003)
Industry	Yes	Yes
Region	Yes	Yes
Year	Yes	Yes
Constant	−1.136 (1.008)	0.014 (0.019)
Observations	713	713
R-squared	0.192	0.326
Kleibergen-Paap rk Wald F	68.799
Kleibergen-Paap rk LM	124.78 ***
Stock-Yogo critical value(10%)	16.38

Note: *** p < 0.01, ** p < 0.05, * p < 0.1, double-tail test.

suggest that the instrumental variable appears to be valid, lending some credibility to the main conclusion as a supplementary robustness check.

In the first stage of Table 5, the coefficient of the one-period lagged collaborative innovation (L.Sci) is negative (−0.454). This may be due to the mean reversion of Sci over the sample period or specific sample characteristics. Nevertheless, the Kleibergen-Paap rk Wald F statistic is 68.799, well above the Stock-Yogo 10% critical value (16.38), indicating no weak identification problem. It should be noted that when using a lagged variable as an instrumental variable, the exclusion restriction requires that L.Sci affects TFP only through the contemporaneous Sci. This assumption may not be fully satisfied in the presence of productivity persistence. Therefore, we interpret the IV results as a supplementary robustness check for the main regressions rather than as a strict causal identification. Future research could employ external instrumental variables to further strengthen causal inference.

4.4. Robustness Test

4.4.1. Alternative Dependent Variable

Following the approach of Sun et al. [32], this study employs the OP method to measure the TFP of SRDI enterprises. The study finds that supply chain co-innovation shows a significant positive correlation with the TFP of SRDI enterprises (β₁ = 0.004, p < 0.01). This indicates that supply chain co-innovation actively promotes TFP. Secondly, the results are consistent with ESG performance playing a partially mediating role in the link between supply chain co-innovation and TFP (β₂ = 0.006, p < 0.01). This attests to the robustness of the findings.

4.4.2. Addition of Control Variables

Extending the approach of Zhang et al. (2025) [26], we also incorporate firm asset scale as a new control variable, and the empirical results consistently indicate that supply chain co-innovation enhances the TFP (β₁ = 0.003, p < 0.01). The partially mediating effect of corporate ESG performance is empirically supported (β₂ = 0.004, p < 0.05). These conclusions are strengthened, and the study’s hypotheses are validated by these findings.

4.4.3. Sample Period Adjustment

We also adjust the sample period from 2018–2023 to 2018–2022 and conduct the regression analysis again. The results show that supply chain co-innovation has a driving effect on the TFP (β₁ = 0.004, p < 0.01), and supply chain co-innovation influences TFP through corporate ESG performance as a partially mediating channel (β₂ = 0.006, p < 0.01). This provides validation for the study’s proposed hypotheses.

4.5. Further Analysis

According to Arrow’s learning-by-doing theory, productivity gains originate from the accumulation of experience during production and interaction processes [30]. In the context of supply chain co-innovation, however, this tacit inter-organizational learning process is difficult to measure directly. Joint patent applications, as the codified product of knowledge co-creation between collaborating parties, provide a quantifiable proxy for this learning effect. Unlike independent patent applications, which reflect a firm’s internal R&D capabilities, the process of generating a joint patent requires repeated technical exchanges, alignment of solutions, and conflict resolution between partners. This process itself constitutes a concrete manifestation of the learning-by-doing mechanism: through iterative collaboration with supply chain partners, enterprises progressively accumulate experiential knowledge in identifying complementary resources, aligning R&D objectives, and reducing coordination costs. An increase in the number of joint patents thus signals an upward shift in the collaboration experience curve, and the temporal lag structure, wherein joint patent accumulation follows collaborative inputs, aligns precisely with the dynamic trajectory posited by Arrow that experience accumulation leads to productivity improvement [30,33]. Consequently, joint patent applications allow for the effective capture of the learning effect pathway through which supply chain co-innovation enhances total factor productivity (TFP) via the accumulation of interactive experience.

As a form of innovation output, joint patents may carry the organizational efficiency gains derived from collaborative learning, but they may also merely represent an increase in technological achievements that improve TFP through the direct application of patented technologies to production processes or product features. The latter constitutes a competing explanatory pathway distinct from the learning-by-doing effect, namely the direct transformation of innovation outcomes. To disentangle these two effects, it is essential to incorporate the two dimensions in the empirical model, respectively, thereby clarifying the contributions of both to TFP. Under this specification, the effect captured by joint patents can be attributed to the experiential benefits arising from the collaborative interaction itself, rather than the direct economic returns from patented technological outputs. Furthermore, the ESG performance mediation pathway hypothesized earlier based on stakeholder theory also indicates to some extent that the roles of the two dimensions may be different [48,49]. If supply chain co-innovation enhanced TFP solely through the transformation of technological achievements, it would be difficult to account for the significant mediating role of ESG performance across the environmental, social, and governance dimensions. Conversely, the process of generating joint patents is accompanied by improvements in supply chain transparency, the co-development of green technologies, and the internalization of governance norms, all of which align closely with the organizational capability improvements emphasized by learning-by-doing theory [38,39]. This study aims to identify the distinctive learning premium inherent in supply chain co-innovation, rather than the mere accumulation of innovation outputs.

Table 6. Results of further analysis.

	(1)	(2)	(3)	(4)	(5)	(6)
	TFP	TFP	TFP	TFP	ESG	ESG
Size	0.557 ***	0.563 ***			0.005	−0.004
	(0.032)	(0.033)			(0.043)	(0.043)
Age	0.056 *	0.060 *			−0.129 **	−0.109 **
	(0.032)	(0.032)			(0.055)	(0.055)
Dual	−0.027	−0.030			0.045	0.026
	(0.029)	(0.029)			(0.046)	(0.046)
Board	−0.106	−0.106			−0.349 ***	−0.382 ***
	(0.083)	(0.083)			(0.116)	(0.119)
Qa	0.034 ***	0.035 ***			−0.012	−0.011
	(0.007)	(0.007)			(0.015)	(0.015)
Lev	0.744 ***	0.748 ***			−0.582 ***	−0.620 ***
	(0.103)	(0.104)			(0.154)	(0.155)
Ownhold	0.005 ***	0.005 ***			0.002	0.001
	(0.001)	(0.001)			(0.002)	(0.002)
Sci-1	0.060 **		0.114 ***		0.188 ***
	(0.027)		(0.036)		(0.044)
Sci-2		0.000		−0.000		−0.000
		(0.000)		(0.000)		(0.000)
Industry	Yes	Yes	Yes	Yes	Yes	Yes
Region	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
_cons	−4.524 ***	−4.656 ***	7.829 ***	7.871 ***	5.164 ***	5.509 ***
	(0.722)	(0.735)	(0.019)	(0.021)	(0.879)	(0.895)
N	1087	1087	1087	1087	1087	1087

Note: *** p < 0.01, ** p < 0.05, * p < 0.1, double-tail test.

reports the results of the test for the learning-by-doing mechanism as well as the estimates after ruling out the competing mechanism. The estimation results in columns (1), (3), and (5) demonstrate that the joint patent application index (Sci-1), serving as a proxy for the learning-by-doing mechanism, exerts a statistically significant and economically meaningful positive effect on both corporate ESG performance and total factor productivity. In contrast, the results in columns (2), (4), and (6) indicate that the patent technology conversion rate (Sci-2) produces no substantial enhancement effect on either corporate ESG performance or total factor productivity in terms of statistical significance or economic magnitude. These findings suggest that the learning-by-doing mechanism (proxied by joint patents) plays a more prominent role in this context, while the direct transformation pathway (proxied by Sci-2) does not receive significant empirical support in our specification. This does not rule out the possibility of other commercialization channels but indicates that they are less salient relative to the learning effect captured by joint patents.

5. Discussion

5.1. Research Conclusions

As a critical micro-economic foundation and core entity of socioeconomic development, SRDI firms play a significant role in upgrading industrial technology upgrades and strengthening the overall competitiveness of the supply chain. Fostering co-innovation between these SRDI firms and their supply chain partners through the establishment of mutually beneficial and trust-based strategic alliances exerts a profound influence on their development trajectory. Focusing on the distinctive interactive relationship inherent in supply chain co-innovation, this study utilizes data from publicly listed “little giant” enterprises spanning the period from 2018 to 2023 to systematically evaluate the effects of supply chain co-innovation on total factor productivity (TFP) improvement and the underlying mechanisms at work. The empirical results support that supply chain co-innovation is significantly positively associated with TFP improvement among SRDI enterprises, and corporate ESG performance serves as a partial mediator in this relationship. Further analysis reveals that SRDI firms can effectively enhance their ESG performance through joint patent applications, thereby establishing a key transmission channel that drives productivity gains consistent with the learning-by-doing perspective [30,31,32].

The analysis of moderating effects demonstrates that the positive impact of supply chain co-innovation is substantially stronger for enterprises operating within the manufacturing sector and for those exhibiting higher levels of profitability. Compared with firms in other industries, manufacturing enterprises face more intense demands for continuous innovation. Confronted with global market competition, rapidly evolving customer preferences, and an accelerating technological environment, manufacturing firms must persistently improve product quality and production efficiency to secure a competitive advantage and achieve sustainable development. The inherent complexity of manufacturing supply chains, characterized by multi-stage coordination requirements and stringent regulatory oversight, amplifies the marginal returns to collaborative innovation activities in this sector. Enterprises with stronger profitability are better positioned to cultivate deep cooperative relationships with a diverse array of suppliers and partners and typically possess more robust risk resilience capabilities. Such financial slack facilitates real-time information sharing and seamless technological integration across organizational boundaries [49,50]. Moreover, profitable enterprises are able to sustain investments in supply chain co-innovation even amid shifting market conditions, thereby concentrating resources on the development of novel technologies, reinforcing their competitive positioning, and promoting sustained TFP improvement.

5.2. Theoretical Implications

This study offers several theoretical contributions that advance the literature on SRDI enterprises, supply chain management, and the mechanisms underpinning productivity growth. First, we sharpen the focus on the distinctive characteristics and developmental constraints of SRDI firms. While the extant literature extensively discusses pathways to high-quality enterprise development, considerably less attention has been devoted to the specific application of these pathways within resource-constrained SRDI enterprises. This study redirects attention toward how SRDI enterprises engage in co-innovation with supply chain partners to consolidate external resources, thereby compensating for inherent resource deficiencies, generating new value, and achieving TFP improvement. By doing so, we address a notable theoretical gap concerning the unique needs and challenges confronting SRDI enterprises and respond directly to the call by Zeng et al. (2024) [67] to “further explore how SRDI enterprises can achieve leapfrog development through the co-innovation among large, medium, and small enterprises.”

Second, by integrating co-innovation into supply chain management research, this study enriches the theoretical discourse on both supply chain dynamics and innovation processes. The analysis of the impact mechanisms of supply chain co-innovation reveals how efficient resource allocation and information sharing, cultivated through repeated interactions and cumulative experience, enhance an enterprise’s responsiveness and innovation velocity. This finding provides a novel perspective for existing supply chain management theories and extends the learning-by-doing framework from intra-firm manufacturing contexts to inter-organizational collaborative settings [32,33,34]. It thus contributes theoretical support and practical guidance for constructing a modern supply chain system that is simultaneously efficient and sustainable.

Third, this study develops a unified theoretical framework that explicates the connection between supply chain co-innovation and TFP improvement. Drawing on learning-by-doing theory and stakeholder theory, we posit and empirically validate a sequential pathway wherein iterative collaboration and knowledge accumulation within supply chains enhance corporate ESG performance, which in turn translates into higher total factor productivity. By identifying corporate ESG performance as a crucial mediating mechanism, we bridge the supply chain management and ESG literatures, demonstrating that collaborative innovation yields not only direct efficiency gains but also environmental, social, and governance improvements that further amplify productivity [28,29]. This framework moves beyond static resource-based views that emphasize resource endowments and instead offers a dynamic, process-oriented explanation for how supply chain partnerships generate sustainable competitive advantages. Elucidating the intrinsic mechanisms through which supply chain innovation shapes enterprise development deepens our theoretical understanding of the micro-foundations of productivity growth among specialized small and medium-sized enterprises.

5.3. Practical Implications

SRDI enterprises should actively seek opportunities for co-innovation with supply chain partners. They should fully recognize the strategic value of supply chain co-innovation for achieving high-quality development. By engaging in resource integration, information exchange, and technological collaboration with supply chain partners, these enterprises can significantly shorten market response times, reduce costs, enhance efficiency, and upgrade quality. This not only strengthens the market competitive advantage of SRDI enterprises themselves but also promotes the coordinated evolution of the industry and supply chain.

SRDI enterprises should focus on enhancing their sustainable development capabilities. As a crucial cornerstone for strengthening the resilience of China’s industry and supply chains and ensuring their security, these enterprises should prioritize sustainability in their development. By engaging in co-innovation with supply chain companies, they can make progress in environmental, social, and governance (ESG) dimensions, contributing directly to stronger overall ESG performance. It is essential to formulate and implement ESG strategies, integrate sustainable development goals with business development strategies, and ensure long-term competitiveness and social value.

Adopt targeted implementation strategies for different SRDI enterprises. Research indicates that SRDI enterprises in the manufacturing sector and those with high profitability should prioritize supply chain co-innovation. Therefore, in promoting supply chain co-innovation, the government needs to pay attention to the differentiated characteristics of various SRDI enterprises and adopt targeted implementation strategies to enhance innovation effectiveness. These enterprises also need to develop strategies tailored to their own characteristics and market environments. For example, highly profitable, specialized and innovative enterprises can fully leverage their resource advantages to develop stronger partnerships within the supply chain, which facilitates information sharing and collaborative innovation.

5.4. Limitations and Future Prospects

Several limitations of this study should be addressed in future work to enhance its robustness. First, the study focuses only on “little giant” enterprises. Future research should expand the sample scope to encompass the broader population of SRDI enterprises. Second, this paper uses empirical research methods to find the mechanism of supply chain co-innovation on the high-quality development of SRDI enterprises. In the future, case study methods can be used to further explore the underlying mechanisms linking them.