Low-Carbon Restructuring, R&D Investment, and Supply Chain Resilience: A U-Shaped Relationship

Abstract

Low-carbon restructuring serves as a critical strategy for enterprises to achieve the “dual-carbon” target and foster sustainable development, whereas supply chain resilience is essential for maintaining competitiveness in complex environments. Based on the data of Chinese A-share listed companies in the manufacturing industry from 2011 to 2023, this paper empirically examines the relationship between low-carbon restructuring, R&D investment, and supply chain resilience. This study reveals a U-shaped relationship between low-carbon restructuring and supply chain resilience, with an inflection point at approximately 2.34. R&D investment significantly strengthens supply chain resilience and positively moderates the relationship by accelerating technological synergies and optimizing resource allocation. Further analysis shows that heavily polluted industries face more pressure in the early stage of low-carbon restructuring compared to non-heavily polluted industries, but R&D investment has a more significant moderating effect on heavily polluted industries. The prediction results based on the Holt–Winters model show that the level of low-carbon restructuring in China’s manufacturing industry will increase steadily in the next seven years, with an average annual growth rate of about 0.021. These new findings are important for managers and researchers to improve supply chain resilience during the low-carbon transition process.

1. Introduction

Sustainable development is indispensable for securing the long-term stability of the global economy and the environment, and it has emerged as a central priority for governments and businesses alike [1]. Following the adoption of the 2030 Agenda for Sustainable Development by the United Nations in 2015, 17 overarching development goals and 169 targets were established to address global challenges such as environmental pollution, climate change, and resource waste [2]. The Chinese government has placed significant emphasis on bolstering the resilience and security of industrial and supply chains. Between 2020 and 2024, it has repeatedly underscored the objective of building “autonomous, controllable, safe, and reliable industrial and supply chains.” Since the introduction of the dual-carbon target [3], an increasing number of enterprises have implemented carbon reduction initiatives. Prominent companies, including Apple and Contemporary Amperex Technology Co. Ltd. (CATL), have adopted comprehensive low-carbon strategies [4] to mitigate carbon emissions at the source of their supply chains. Today, there is a growing consensus on the necessity of promoting low-carbon restructuring and decarbonizing supply chains [5]. Low-carbon restructuring (LCR) refers to the process of reducing carbon emissions within the supply chain through technological innovation, process optimization, and resource allocation adjustments to achieve the dual-carbon goal [6]. However, while LCR offers environmental benefits, it also poses significant challenges to supply chain resilience. Supply chain resilience (SCR) refers to the ability of a supply chain to recover to its original or an improved state after experiencing external disruptions [7]. Enterprises may struggle to meet the emission reduction requirements set by leading firms in the supply chain, heightening the risk of disruptions and eroding resilience. The low-carbon transition highlights the critical importance of SCR. LCR has instigated substantial changes in production methods, technologies, and standards, complicating coordination and synergy among enterprises and thereby diminishing SCR. In this context, LCR demands that enterprises comprehensively decarbonize their operations while responding to shifts in market demand and technological challenges. Firms that fail to adapt promptly risk competitive disadvantages, jeopardizing the stability and resilience of their supply chains.

In recent years, recurrent epidemics, geopolitical tensions, and international dynamics have led to frequent supply chain disruptions, significantly increasing the risk of interruptions [8]. This dynamic has pushed the relationship between LCR and SCR to the forefront of strategic research—how can companies maintain or even enhance the risk resilience of the supply chain system while advancing the whole-chain low-carbon reconfiguration in the carbon-neutral transition process? Currently, the related literature mainly explores the direct impact of LCR on operational performance (Shi et al., 2022) [9] and regional economic transition effects [10]. In order to enhance the risk resistance of supply chains, scholars have mainly studied the proactive strategies for SCR enhancement from different perspectives; for example, the design of the supply chain network structure [11], proper supplier selection [12], redundancy design of the supply chain [13], flexibility of the supply chain [14], and diversification of supply chain members [15], among others. Few have systematically explored the relationship between both LCR and SCR. Both innovation theory and firm growth theory posit that R&D innovation is a pivotal driver for manufacturing firms to maintain competitiveness in complex environments [16,17]. Research has demonstrated that R&D not only enhances technological innovation capabilities but also fortifies supply chain flexibility and risk resistance by fostering technological synergies and optimizing resource allocation [18,19]. Nevertheless, the existing literature has yet to fully elucidate the intrinsic mechanisms linking LCR, R&D, and SCR. Additionally, there is a dearth of predictions regarding the development trajectory of LCR. These research gaps impede theoretical advancements and create uncertainties for enterprises undergoing low-carbon transformation. This study focuses on the period from the 12th Five-Year Plan to the 14th Five-Year Plan, and selects the manufacturing industry as the research object. This industry contributes 72% of the total industrial carbon emissions (China Environmental Statistics Yearbook, 2023), which is representative of the policy response and the industry. Therefore, this paper empirically examines the relationship between LCR and supply chain elasticity using the panel data of China’s A-share listed companies in the manufacturing industry (2011–2023), with R&D as a moderating variable. Furthermore, the study predicts the extent of LCR for the next seven years based on the seasonality-free Holt–Winters additive exponential model. The seasonality-free Holt–Winters additive exponential model (SF–Holt–Winters Additive) is a time-series forecasting method that explicitly excludes seasonal components, decomposing only the level and trend terms of the data. The study addresses the following core questions: First, what is the mechanism through which LCR influences SCR? Does this relationship exhibit a curvilinear pattern, such as a U-shaped or inverted U-shaped effect? Second, what role does R&D play in the relationship between LCR and SCR? Specifically, can it mitigate the decline in resilience during the initial stages of LCR? Third, do significant differences exist between industries (e.g., heavily polluting versus non-polluting industries) in the process of LCR? These questions hold substantial theoretical and practical significance, offering valuable guidance for enterprises aiming to enhance SCR during low-carbon transformation.

The innovations of this paper are threefold: First, while the existing literature has predominantly focused on the economic benefits of LCR or isolated dimensions of SCR, this paper systematically explores the relationship between the two, revealing a U-shaped mechanism. Second, it introduces R&D as a moderating variable and analyzes its role in mitigating the decline in SCR during the early stages of LCR, providing new empirical evidence for the “sunk cost-innovation gain balancing mechanism.” Third, A time-series model based on the SF–Holt–Winters Additive model predicts the trend of the LCR over the next seven years (see Chapter 4 for details), providing a forward-looking reference for policymakers. This study aims to provide theoretical insights and practical pathways for achieving sustainable development under the dual-carbon target while enhancing the global competitiveness of Chinese enterprises. The second part reviews low-carbon reconfiguration, SCR, and related theories, on the basis of which the research hypotheses are formulated; the third part elaborates on the data sources and model construction methods; the fourth part presents the results of the empirical analyses and conducts a robustness test; the fifth part synthesizes and discusses the research findings and their theoretical significance; and, finally, the sixth part summarizes the main conclusions and looks ahead to the future directions of the research.

2. Literature Review and Research Hypothesis

2.1. Low-Carbon Restructuring and Supply Chain Resilience

With the intensification of global climate change and the advancement of the United Nations’ Sustainable Development Goals (SDGs), LCR has become the core of corporate strategies to cope with environmental regulations and market competition [20,21]. This process not only requires enterprises to decarbonize their production processes through technological innovation and process optimization [22], but also to embed decarbonization mechanisms throughout the supply chain life cycle to cope with policy pressures such as carbon caps, carbon trading, and carbon taxes [23,24,25]. Empirical studies have shown that leading companies such as Apple and Ningde Times have already penetrated carbon emission standards upstream through supply chain dominance, forming a “chain-driven” LCR paradigm, but this has exacerbated the risk of supply chain disruptions.

SCR refers to the ability of a supply chain system to resist, adapt, and quickly recover from external shocks, and its core connotation can be summarized into two dimensions: dynamic responsiveness and risk resilience [26]. The former emphasizes the flexibility of the supply chain, i.e., to maintain operational efficiency by efficiently coordinating various types of resources and quickly adjusting operational plans when risks occur, while the latter is reflected in the stable collaborative relationship among supply chain members, which enables enterprises to enhance their overall risk-resistant capability through in-depth collaboration in the face of market fluctuations. Scholars have interpreted SCR in multiple dimensions and explored SCR, including competitiveness, ability to sustain operations, and resilience to external shocks [27]. Scholars have identified SCR through core competencies such as flexibility, visibility, and agility [28,29], as well as quantitative metrics such as recovery time, degree of recovery, and post-disruption financial performance [30,31]. Together, these studies reveal the realization path of SCR: enhancing environmental awareness through intelligent monitoring and information sharing, reducing the risk of supply–demand mismatch with the help of optimal resource allocation, and relying on flexible collaborative networks to achieve rapid reconfiguration after disruptions [32,33]. Based on this, this paper constructs indicators of SCR from two perspectives of supply chain efficiency (SCA) and supply chain stability (SCE) in the empirical test.

However, SCR research faces new challenges in the context of low-carbon transition. The technological and process changes brought about by LCR may increase the risk of supply chain disruptions [34]. On the one hand, initial sunk costs and organizational restructuring associated with LCR may weaken LCR by reducing operational efficiency and leading to “transition pains” [35]. On the other hand, LCR can enhance SCR by improving efficiency and adaptability through innovation and resource optimization [22]. For example, the adoption of clean energy and low-carbon technologies can reduce dependence on traditional energy sources and lower the risks associated with energy price volatility [36]. This dual effect suggests that the relationship between LCR and SCR may be characterized by nonlinearity.

Based on the above analysis and according to the dynamic capability theory [37], this paper suggests that the relationship between LCR and SCR follows a U-shaped pattern with initial inhibition followed by enhancement. During the early stages of LCR, technological transformation and resource allocation adjustments may negatively impact SCR. However, as low-carbon technologies mature and synergies emerge, SCR is gradually strengthened. Therefore, the following hypothesis is proposed:

H1: 

Low-carbon restructuring and supply chain resilience exhibit a U-shaped relationship.

2.2. Relationship Between R&D Investment and Supply Chain Resilience

The relationship between R&D and SCR has yielded mixed conclusions, which can be categorized into two perspectives: first, R&D is positively correlated with SCR [29,38]; second, R&D is negatively correlated with SCR [39,40]. In the short term, if firms fail to achieve the expected outcomes of R&D activities, it may result in resource wastage and reduced supply chain efficiency, thereby weakening resilience.

In the context of the dual-carbon goal, enterprises’ R&D serves three primary purposes: first, to enhance the green innovation capacity of the supply chain. Supply chain sustainability is enhanced through the dual path of green technology development and digital technology application. The former directly reduces carbon emissions and enhances adaptability to environmental policy fluctuations by developing low-carbon technologies, optimizing energy structures and promoting circular economy models [41,42]; the latter indirectly optimizes the resource cycle by enhancing transparency and collaboration efficiency [43]. Digital technologies represented by the Internet of Things (IoT) and blockchain [44,45] significantly enhance supply chain visibility and responsiveness while improving sustainability through real-time data sharing and smart contracting mechanisms, while increasing SCR [46]. Second, it builds technological barriers to improve supply chain competitiveness. By achieving technological specialization through R&D innovation, firms can raise barriers for potential competitors, ensuring supply chain stability [47]. Third, it enables rapid response capabilities within the supply chain. R&D enhances firms’ technological reserves and resource integration capabilities, equipping them with stronger adaptability to market demand changes and external shocks [48].

Ultimately, the R&D investments described above are aimed at improving SCR. through technological innovation, process optimization, and resource integration, R&D investments can significantly improve supply chain agility, stability, and resilience to risk, thereby enhancing firms’ resilience in a complex environment [49,50]. Based on this analysis, the following hypothesis is proposed:

H2: 

R&D investment is positively related to supply chain resilience.

2.3. The Moderating Role of R&D Investment in the Relationship Between LCR and Supply Chain Resilience

Resource-based theory posits that unique and hard-to-replicate resources and capabilities can build a firm’s differentiated competitive advantage, although single types of resources make limited contributions to firm performance [51]. In the context of LCR, the factors influencing SCR include not only green technology innovation and process optimization, but also key resources such as R&D [33]. R&D serves as a critical driver for enhancing supply chain flexibility and risk resilience by advancing technological development, knowledge accumulation, and resource integration during LCR [52,53]. During LCR implementation, R&D acts as a significant moderating factor, manifesting in the following aspects:

First, R&D accelerates the development and application of low-carbon technologies, thereby enhancing SCR. In the early stages of LCR, firms often face challenges such as technological lag and insufficient resource allocation, which may reduce supply chain efficiency [54]. In contrast, R&D significantly improves sustainability and risk resilience by promoting the adoption of green technologies, such as renewable energy and circular economy solutions [55]. For example, digital technologies (e.g., Internet of Things, blockchain) enhance traceability and transparency, strengthening SCR [56]. As an example, Ningde Times invested more than 10% of its annual revenue in R&D for the development of module-less battery technology to increase energy density by 20% while reducing production costs by 30% (Ningde Times Annual Report, 2023).

Second, R&D enhances firms’ resource integration capabilities, optimizing the synergies of LCR. LCR requires firms to achieve resource coordination and technological innovation across the supply chain. Through knowledge sharing and collaborative innovation, R&D significantly improves supply chain synergy efficiency [57]. Additionally, it helps firms address uncertainties in LCR by developing new low-carbon technologies or optimizing existing ones, thereby improving supply chain adaptability and recovery capabilities [58].

Finally, R&D strengthens the innovation gains of LCR, further enhancing SCR. As LCR progresses, the technological advantages and market competitiveness gained through R&D translate into stability and flexibility within supply chains [59]. For instance, by developing green products and services, R&D not only meets market demand but also enhances supply chain value creation [60].

Based on the above analysis, this paper posits that R&D positively moderates the relationship between LCR and SCR, leading to the following hypothesis:

H3: 

R&D investment positively moderates the relationship between LCR and SCR.

The relationships between the three hypotheses are illustrated in Figure 1.

3. Methodology

3.1. Sample Selection and Data Sources

The data used in this study are from Chinese A-share listed companies, spanning from 2011 to 2023. The financial data are mainly collected from CSMAR database and company annual report supplements. To ensure the representativeness and accuracy of the data, the following processing steps are used in this study: (1) samples from companies with persistent bankruptcies and serious missing data are excluded; (2) the relevant continuous variables are winsorized at the 1% level to control for the effects of outliers; and (3) the statistical analysis is performed using Stata 17.0. After these treatments, the final sample includes 1296 listed companies with a total of 8967 observations. In the heterogeneity analysis, the division of heavy polluting and non-heavy polluting industries is based on the 16 categories of heavy polluting industries (specifically thermal power, iron and steel, cement, etc.) delineated by the Ministry of Ecology and Environment’s “Guidelines for Disclosure of Environmental Information of Listed Companies” (EIAO Consolidated [2021] No. 27) for the screening of the sample. This process ensures the representativeness of the sample and also conforms to the data processing norms of similar studies.

3.2. Variable Definitions

3.2.1. Dependent Variable

This paper measures SCR from two perspectives: supply chain efficiency (SCA) and supply chain stability (SCE). The SCA index is derived from inventory turnover days, following Zhang’s (2022) methodology [61]. Specifically, SCA is calculated as the natural logarithm of 360 divided by the ratio of the cost of goods sold (COGS) to the average inventory balance:

$$SCA=In\left({\displaystyle \frac{360}{COGS/average inventory balance}}\right)$$

A higher SCA indicates a longer inventory turnover period, reflecting lower supply chain efficiency.

Supply chain stability (SCE) is measured by supply chain volatility, calculated as the sum of the absolute values of the percentage changes in procurement/sales of the top five suppliers/customers compared to the previous year, divided by the sum of the proportions of the top five suppliers/customers in the current year. A higher SCE indicates greater volatility in relationships with upstream and downstream supply chain partners, reflecting lower supply chain stability. The SCR index is constructed by assigning equal weights (0.5) to SCA and SCE, i.e., SCR = 0.5 × SCA + 0.5 × SCE.

3.2.2. Independent Variable

Low Carbon Reconstruction. Since enterprises rarely disclose their carbon dioxide emissions, this paper collects data on industry carbon emissions from the CSMAR database based on the measurement method of Zhao et al. (2021) [62]. Therefore, this paper takes the revenue per unit of carbon emission as a proxy variable for the degree of LCR, and the larger the value of this indicator, the higher the degree of LCR of the enterprise. Its specific calculation method is shown in Equation (1).

$$LCR={\displaystyle \frac{\mathrm{Enterprise} \mathrm{business} \mathrm{income}}{\left(\frac{\mathrm{Industry} \mathrm{Carbon} \mathrm{Emissions}}{\mathrm{Industry} \mathrm{main} \mathrm{operating} \mathrm{costs}}+1\right)\times \mathrm{Enterprise} \mathrm{operating} \mathrm{costs}}}$$

(1)

R&D investment: To account for differences among enterprises, this paper adopts a relative measure of R&D intensity, defined as R&D as a percentage of operating revenue, following Hirshleifer et al. (2012) [63].

3.2.3. Controlled Variables

Building on existing research [64,65,66,67], the following control variables are included: return on assets (ROA), return on equity (ROE), gross profit margin (Gross Profit), cash flow ratio (Cashflow), accounts receivable ratio (REC), inventory ratio (INV), fixed assets ratio (FIXED), intangible assets ratio (Intangible), and revenue growth rate (Growth). Detailed definitions and measurements of these variables are provided in

Table 1. Main variables.

	Variable Name	Variable Indicator	Variable Definition
Dependent variable	Supply chain resilience	SCR	The average of supply chain efficiency and supply chain stability indicators
Independent variable	Low-Carbon Restructuring	LCR	Revenue per unit of carbon emissions; the larger the value, the higher the degree of low-carbon restructuring
	R&D investment	R&D	R&D investment as a percentage of revenue; the larger the value, the higher the R&D investment
Controlled variables	Return on Assets	ROA	Operating profit/Total assets
	Return on Equity	ROE	Operating profit/Net assets
	Gross Profit Margin	Gross Profit	Gross profit/Revenue
	Cash Flow Ratio	Cashflow	Net cash flow from operating activities/Total assets
	Accounts Receivable Ratio	REC	Accounts receivable/Total assets
	Inventory Ratio	INV	Inventory/Total assets
	Fixed Assets Ratio	FIXED	Fixed assets/Total assets
	Intangible Assets Ratio	Intangible	Intangible assets/Total assets
	Revenue Growth Rate	Growth	Company revenue growth rate
Other Variables	Seasonality-Free Holt–Winters Additive exponential model	SF–Holt–Winters Additive	The seasonality-free Holt–Winters additive exponential model is a time-series forecasting method that explicitly excludes seasonal components, de-composing only the level and trend terms of the data.
	Sample Size	N	The sample size is the total number of sample elements drawn from the total population.
	R-squared	R2	The proportion of variance in the dependent variable explained by the independent variables in the linear regression model.
	λ, α, β, φ, γ, ω, θ	λ, α, β, φ, γ, ω, θ	Model parameters to be estimated.

.

3.3. Model Construction

To empirically examine the impact of LCR and R&D on SCR, as well as the moderating role of R&D in the relationship between LCR and SCR, the following three models are constructed:

$$SCR=\lambda +\alpha LCR+\beta LC{R}^2+\theta {\displaystyle \sum Control+\epsilon }$$

(2)

$$SCR=\lambda +\phi R\&D+\theta {\displaystyle \sum Control+\epsilon }$$

(3)

$$SCR=\lambda +\alpha LCR+\beta LC{R}^2+\phi R\&D+\gamma LCR\times R\&D+\omega LC{R}^2\times R\&D+\theta {\displaystyle \sum Control+\epsilon }$$

(4)

In this context, ∑Control represents the set of control variables, λ, α, β, φ, γ, ω and θ denote the parameters to be estimated, and ε is the error term. Please see Table 1 for the selection of control variables. We then control for industry-fixed effects and time-fixed effects.

4. Empirical Results and Analysis

4.1. Descriptive Statistic and Correlation Analysis

Stata17 software was used to conduct descriptive statistics and correlation analyses of the relevant variables involved, and the results are shown in

Table 2. Descriptive statistics and correlation analysis of variables.

Variable	Mean	SD	SCR	LCR	LCR2	R&D	ROA	ROE	Gross Profit	Cashflow	REC	INV	FIXED	Intangible	Growth
SCR	1.0163	0.1631	1.000
LCR	2.4404	0.17	0.288 ***	1.000
LCR2	5.9846	0.8118	0.283 ***	0.999 ***	1.000
R&D	1.1663	0.8006	0.261 ***	0.486 ***	0.490 ***	1.000
ROA	0.0357	0.1677	−0.048 ***	0.045 ***	0.044 ***	0.012	1.000
ROE	0.0382	1.9521	−0.022 **	0.005	0.005	0.013	0.030 ***	1.000
Gross Profit	0.2626	0.1749	0.323 ***	0.251 ***	0.249 ***	0.244 ***	0.322 ***	0.039 ***	1.000
Cash flow	0.0496	0.0778	−0.140 ***	−0.050 ***	−0.049 ***	−0.024 **	0.155 ***	0.066 ***	0.215 ***	1.000
REC	0.126	0.0973	0.080 ***	0.291 ***	0.295 ***	0.227 ***	−0.026 **	0.012	−0.036 ***	−0.200 ***	1.000
INV	0.1344	0.0884	0.496 ***	0.102 ***	0.096 ***	−0.035 ***	−0.020 *	0.012	−0.095 ***	−0.145 ***	0.089 ***	1.000
FIXED	0.2573	0.1546	−0.284 ***	−0.472 ***	−0.469 ***	−0.268 ***	−0.069 ***	0.007	−0.249 ***	0.159 ***	−0.354 ***	−0.207 ***	1.000
Intangible	0.0506	0.0528	0.008	−0.015	−0.021 **	−0.067 ***	−0.043 ***	0.006	0.043 ***	−0.006	−0.140 ***	−0.111 ***	−0.008	1.000
Growth	0.2892	3.142	−0.088 ***	0.024 **	0.024 **	−0.016	0.051 ***	0.013	0.010	0.019 *	0.010	−0.017	−0.015	−0.001	1.000

Standard errors in brackets. * p < 0.1 ** p < 0.05 *** p < 0.01.

. In the statistical sample, the mean value of SCR is 1.0163 with a standard deviation of 0.1631, indicating that the overall SCR level of the sample enterprises is high and the differences among enterprises are small. The average value of LCR is 2.4404, with a standard deviation of 0.17, indicating that the degree of low carbon reconstruction of the current manufacturing enterprises is at a medium level, and the implementation level is relatively close. The mean value of research and development (R&D) investment is 1.1663 with a standard deviation of 0.8006, indicating that there is a large difference in the level of R&D among enterprises. In addition, the correlation coefficients between variables are all less than the critical value of 0.7. Furthermore, the variance inflation factors (VIFs) are all less than the critical value of 10, with an average VIF of 1.05. This excludes the problem of multicollinearity between variables and allows for further empirical analysis.

4.2. Benchmark Regression

The Hausman test was used to determine whether the fixed effect model or the random effect model should be selected for regression analysis. As can be seen from

Table 3. Benchmark regression.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	SCR	SCR	SCR	SCR	SCR	SCR	SCR
LCR		−0.0291 *		−0.0426 **	−1.823 ***	−0.984 ***	−1.142 ***
		(0.0175)		(0.0186)	(0.314)	(0.353)	(0.357)
LCR2					0.390 ***	0.193 **	0.227 ***
					(0.0661)	(0.0755)	(0.0763)
R&D			0.0205 ***				−0.456 ***
			(0.00200)				(0.152)
LCR*R&D				0.00919 ***		−0.0297 **	0.346 ***
				(0.000859)		(0.0136)	(0.126)
LCR2 *R&D						0.0156 ***	−0.0617 **
						(0.00552)	(0.0263)
ROA	−0.0486 ***	−0.0481 ***	−0.0445 ***	−0.0444 ***	−0.0470 ***	−0.0432 ***	−0.0427 ***
	(0.00670)	(0.00671)	(0.00667)	(0.00667)	(0.00669)	(0.00666)	(0.00666)
ROE	−0.00162 ***	−0.00162 ***	−0.00163 ***	−0.00163 ***	−0.00162 ***	−0.00162 ***	−0.00162 ***
	(0.000496)	(0.000496)	(0.000493)	(0.000493)	(0.000495)	(0.000492)	(0.000492)
Gross	0.134 ***	0.131 ***	0.123 ***	0.126 ***	0.128 ***	0.123 ***	0.123 ***
Profit	(0.0114)	(0.0116)	(0.0114)	(0.0115)	(0.0116)	(0.0115)	(0.0115)
Cash	−0.0731 ***	−0.0726 ***	−0.0701 ***	−0.0704 ***	−0.0736 ***	−0.0713 ***	−0.0702 ***
flow	(0.0151)	(0.0151)	(0.0150)	(0.0150)	(0.0151)	(0.0150)	(0.0150)
REC	−0.0407 **	−0.0384 **	−0.0389 **	−0.0424 **	−0.0525 ***	−0.0525 ***	−0.0547 ***
	(0.0194)	(0.0195)	(0.0193)	(0.0193)	(0.0196)	(0.0194)	(0.0194)
INV	0.676 ***	0.678 ***	0.699 ***	0.699 ***	0.687 ***	0.703 ***	0.702 ***
	(0.0196)	(0.0196)	(0.0196)	(0.0196)	(0.0196)	(0.0196)	(0.0196)
FIXED	0.0166	0.0188	0.0199 *	0.0176	0.0219 *	0.0194 *	0.0199 *
	(0.0116)	(0.0117)	(0.0116)	(0.0116)	(0.0117)	(0.0116)	(0.0116)
Intangible	0.0932 ***	0.0920 ***	0.0950 ***	0.0978 ***	0.108 ***	0.111 ***	0.109 ***
	(0.0272)	(0.0272)	(0.0270)	(0.0270)	(0.0273)	(0.0271)	(0.0271)
Growth	−0.00500 ***	−0.00502 ***	−0.00496 ***	−0.00493 ***	−0.00508 ***	−0.00497 ***	−0.00493 ***
	(0.000392)	(0.000392)	(0.000389)	(0.000389)	(0.000391)	(0.000389)	(0.000389)
_cons	0.893 ***	0.822 ***	0.868 ***	0.969 ***	3.006 ***	2.108 ***	2.294 ***
	(0.00627)	(0.0434)	(0.00671)	(0.0452)	(0.372)	(0.413)	(0.417)
Hausman test p-value	0.000	0.000	0.000	0.000	0.000	0.000	0.000
N	8967	8967	8967	8967	8967	8967	8967
R2	0.164	0.164	0.175	0.176	0.168	0.178	0.180
adj. R2	0.021	0.022	0.035	0.036	0.166	0.177	0.178

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

, the p-value of the Hausman test for all models is less than 0.1, so the fixed effect model is selected for regression analysis.

4.2.1. Analysis of Model Regression Results of the Relationship Between Low-Carbon Restructuring and SCR

Based on the results of Model 2 and Model 5 in Table 3, it can be seen that the coefficient of the primary term of LCR α is −1.823, which is significantly negative at the 1% confidence level; the coefficient of the squared term of LCR β is 0.390, which is significantly positive at the 1% level, indicating that the relationship between LCR and SCR is a significant U-shape, and the assumption of H1 is verified. The possible reasons for this are that at the initial stage of LCR, the increase in enterprise resource input and the higher cost of technological transformation may have a certain negative impact on the supply chain toughness; with the in-depth promotion of LCR, optimization of resource allocation, and improvement in technological efficiency, the positive effect of LCR on the supply chain toughness gradually appears.

4.2.2. Analysis of Model Regression Results of the Relationship Between R&D and Supply Chain Resilience

According to the results of Model 3 in Table 3, the coefficient φ of R&D is 0.0205, which is significantly positive at the 1% confidence level, indicating that R&D is positively correlated with enterprise performance, and hypothesis H2 has been verified. The possible reason for this is that R&D helps enterprises enhance their ability to cope with supply chain disruption risk through technological upgrading and innovative resource allocation, thus improving SCR.

4.2.3. Regression Analysis of the Moderating Effect of R&D Investment on the Relationship Between Low-Carbon Restructuring and Supply Chain Resilience

Based on the results from Model 4 and Model 6 in Table 3, it is evident that the interaction coefficient γ between the first-order term of LCR and R&D is 0.00919, which is significantly positive at the 1% confidence level. Additionally, the interaction coefficient ω between the squared term of LCR and R&D is 0.0156, which is significantly positive at the 5% confidence level. These results indicate that R&D positively moderates the relationship between LCR and SCR, thereby supporting hypothesis H3. The possible reason for this is that R&D enhances the positive impact of LCR on SCR through technological innovation and optimization of resource allocation. Moreover, it mitigates the negative effects on SCR that may occur during the initial stages of LCR. Model 7, which represents the full model of this study, aligns with the aforementioned analysis, and thus, further elaboration is unnecessary.

For the control variables, based on the results of Model 1 in Table 3, return on assets is significantly negative at the 1% confidence level, indicating that profitability has less direct impact on SCR; gross profit margin is significantly positive at the 1% confidence level, indicating that the improvement of corporate profitability contributes to the enhancement of SCR; cash flow is significantly negative at the 1% confidence level, possibly reflecting the complex relationship between liquidity management and SCR; tangible assets are significantly positive at the 1% confidence level, indicating that asset size has a positive impact on SCR; intangible assets are significantly positive at the 1% confidence level, indicating that technological innovation and knowledge accumulation play an important role in the enhancement of SCR; and enterprise growth is significantly negative at the 1% confidence level, indicating that high-growth enterprises may face greater supply chain risks, thus negatively affecting SCR.

4.3. U-Turn Analysis

Based on the U-shaped results of Model 5 in Table 3, it can be judged that there is a first inhibition and then promotion effect of LCR on supply chain toughness, i.e., the initial period of LCR may have a certain negative impact on supply chain toughness, but when the level of LCR reaches a certain degree, its positive effect gradually appears. In order to identify the inflection point of the U-shaped curve, this study calculates the derivative of Model 5, resulting in ∂SCR/∂LCR = −1.823 + 2 × 0.390 × LCR. For the supply chain toughness of the LCR of the first-order derivatives of “0”, to get the inflection point of the U-shaped curve, that is, the quadratic function of the threshold for the LCR = 2.34, the LCR is recorded as an LCR inflection point, that is, the quadratic function of the threshold for the LCR = 2.34. It is recorded as the LCR inflection point C, as shown in Figure 2. When the degree of LCR of enterprises is lower than 2.34, LCR of SCR is inhibitory, and when the degree of LCR is higher than 2.34, the promotion effect will replace the inhibitory effect.

After the introduction of R&D innovation input, based on the results of Model 6 in Table 3, the inflection point of the U-shaped curve is calculated to be LCR = 1.96, which is recorded as the inflection point D of LCR, as shown in Figure 2. Compared with the original curve and the original inflection point, the new curve and the new inflection point are shifted to the upper left as a whole, which fully indicates that the improvement in R&D can not only alleviate the negative impact on SCR at the early stage of LCR, but also accelerate the positive promotion effect on SCR at the later stage of LCR.

4.4. Further Analysis

Manufacturing enterprises in different industries face pressures from technological transformation and differences in environmental regulations in the process of LCR. In order to further reveal the heterogeneity of the relationship between “LCR-SCR” in heavily polluting industries and non-polluting industries, and the location of the U-shaped inflection point, the regression results are shown in

Table 4. Comparison based on heavily polluting and non-heavily polluting industries (1).

	Heavily Polluting Industries	Non-Heavily Polluting Industries	Heavily Polluting Industries	Non-Heavily Polluting Industries	Heavily Polluting Industries	Non-Heavily Polluting Industries
	SCR	SCR	SCR	SCR	SCR	SCR
LCR			0.0704	−0.0948 *	−3.001 **	−4.423 ***
			(0.0540)	(0.0536)	(1.174)	(1.135)
LCR2					0.668 ***	0.884 ***
					(0.253)	(0.227)
ROA	−0.121 ***	−0.0379 *	−0.121 **	−0.0378 *	−0.115 **	−0.0344 *
	(0.0222)	(0.0201)	(0.0508)	(0.0200)	(0.0503)	(0.0176)
ROE	−0.00130 **	−0.0217 **	−0.00129 ***	−0.0215 **	−0.00125 ***	−0.0209 **
	(0.000542)	(0.00859)	(0.000454)	(0.00851)	(0.000448)	(0.00842)
Gross	0.107 ***	0.162 ***	0.0967	0.167 ***	0.0777	0.164 ***
Profit	(0.0179)	(0.0364)	(0.0628)	(0.0357)	(0.0589)	(0.0344)
Cash	−0.0264	−0.0780 ***	−0.0230	−0.0784 ***	−0.0229	−0.0797 ***
flow	(0.0263)	(0.0235)	(0.0350)	(0.0233)	(0.0347)	(0.0233)
REC	−0.0189	−0.0690	−0.00593	−0.0716	-0.0252	−0.0868 *
	(0.0343)	(0.0512)	(0.0604)	(0.0499)	(0.0598)	(0.0447)
INV	0.702 ***	0.596 ***	0.700 ***	0.579 ***	0.699 ***	0.584 ***
	(0.0295)	(0.0460)	(0.0628)	(0.0440)	(0.0631)	(0.0432)
FIXED	−0.0348 **	0.0850 ***	−0.0328	0.0792 ***	−0.0310	0.0808 ***
	(0.0174)	(0.0271)	(0.0402)	(0.0271)	(0.0402)	(0.0259)
Intangible	0.155 ***	−0.0312	0.147 *	−0.0302	0.160 *	−0.000529
	(0.0414)	(0.0648)	(0.0842)	(0.0641)	(0.0841)	(0.0646)
Growth	−0.00319 ***	−0.00741 ***	−0.00320	−0.00738 ***	−0.00328	−0.00743 ***
	(0.000567)	(0.00153)	(0.00197)	(0.00153)	(0.00202)	(0.00153)
_cons	0.889 ***	0.910 ***	0.727 ***	1.152 ***	4.246 ***	6.433 ***
	(0.0102)	(0.0128)	(0.123)	(0.136)	(1.360)	(1.416)
N	3941	5026	3941	5026	3941	5026
R2	0.162	0.181	0.163	0.183	0.170	0.199
adj. R2	0.026	0.179	0.161	0.182	0.168	0.197

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

and

Table 5. Comparison based on heavily polluting and non-heavily polluting industries (2).

	Heavily Polluting Industries	Non-Heavily Polluting Industries	Heavily Polluting Industries	Non-Heavily Polluting Industries	Heavily Polluting Industries	Non-Heavily Polluting Industries
	SCR	SCR	SCR	SCR	SCR	SCR
LCR			0.0191	−0.196 ***	−3.127 **	−3.732 ***
			(0.0546)	(0.0579)	(1.402)	(1.222)
LCR2					0.687 **	0.718 ***
					(0.307)	(0.248)
R&D	0.0182 ***	0.0191 ***			−1.257	−0.362 **
	(0.00343)	(0.00343)			(1.124)	(0.161)
LCR*R&D			0.00768 ***	0.0102 ***	1.082	0.263 **
			(0.00222)	(0.00153)	(0.966)	(0.130)
LCR2 *R&D					−0.229	−0.0437
					(0.207)	(0.0277)
ROA	−0.117 ***	−0.0361 *	−0.116 **	−0.0349 *	−0.116 **	−0.0313 *
	(0.0221)	(0.0199)	(0.0508)	(0.0193)	(0.0509)	(0.0167)
ROE	−0.00131 **	−0.0205 **	−0.00131 ***	−0.0198 **	−0.00125 ***	−0.0194 **
	(0.000540)	(0.00818)	(0.000436)	(0.00786)	(0.000433)	(0.00779)
Gross	0.0956 ***	0.153 ***	0.0916	0.160 ***	0.0780	0.157 ***
Profit	(0.0179)	(0.0359)	(0.0617)	(0.0348)	(0.0592)	(0.0330)
Cash	−0.0234	−0.0772 ***	−0.0222	−0.0776 ***	−0.0251	−0.0769 ***
flow	(0.0262)	(0.0232)	(0.0345)	(0.0227)	(0.0343)	(0.0227)
REC	−0.00748	−0.0700	−0.00407	−0.0756	−0.0249	−0.0880 **
	(0.0343)	(0.0505)	(0.0612)	(0.0477)	(0.0606)	(0.0433)
INV	0.714 ***	0.625 ***	0.713 ***	0.602 ***	0.711 ***	0.605 ***
	(0.0295)	(0.0475)	(0.0632)	(0.0450)	(0.0638)	(0.0444)
FIXED	−0.0357 **	0.0857 ***	−0.0350	0.0743 ***	−0.0316	0.0758 ***
	(0.0174)	(0.0268)	(0.0400)	(0.0266)	(0.0402)	(0.0255)
Intangible	0.153 ***	−0.0343	0.152 *	−0.0319	0.162 *	−0.00568
	(0.0413)	(0.0616)	(0.0839)	(0.0593)	(0.0835)	(0.0589)
Growth	−0.00313 ***	−0.00730 ***	−0.00313	−0.00720 ***	−0.00320	−0.00719 ***
	(0.000564)	(0.00146)	(0.00194)	(0.00142)	(0.00199)	(0.00143)
_cons	0.873 ***	0.882 ***	0.830 ***	1.373 ***	4.424 ***	5.713 ***
	(0.0106)	(0.0140)	(0.125)	(0.144)	(1.598)	(1.503)
N	3941	5026	3941	5026	3941	5026
R2	0.169	0.193	0.170	0.203	0.178	0.217
adj. R2	0.033	0.191	0.167	0.201	0.175	0.214

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

.

According to Table 4 and Table 5, the “LCR-SCR” of enterprises in heavy polluting industries and non-polluting industries shows a U-shaped relationship. Based on the above U-shaped inflection point formula, it can be seen that before joining the R&D innovation input, the LCR inflection point of the heavy pollution industry occurs at LCR = 2.28, and the LCR inflection point of the non-pollution industry occurs at LCR = 2.60; after joining the R&D innovation input, the LCR inflection point of the heavy pollution industry shifts left to LCR = 2.25, and the LCR inflection point of the non-pollution industry shifts left to LCR = 2.55. This phenomenon indicates that (1) the time node of the LCR inflection point of the heavy pollution industry is earlier than that of the non-heavy pollution industry, indicating that the heavy pollution industry faces more transformation pressure at the early stage of LCR and needs to realize technological synergies earlier; (2) compared with the non-heavy pollution industry, the moderating effect of the investment in R&D on the heavy pollution industry in mitigating the negative effects of the early stage of LCR is more significant, which provides an opportunity for the heavy pollution industry to accelerate the low-carbon transformation through technological innovation. This provides empirical support for the acceleration of low-carbon transformation through technological innovation in heavy polluting industries.

4.5. Projections of the Extent of Low-Carbon Reconfiguration

Time-series analysis of the current level of LCR shows a significant linear trend in LCR (ADF test p < 0.01; Supplementary Materials). Combined with the feature that no structural breakpoints were detected during the policy smoothing period, this study adopts the deterministic demand hypothesis and proposes to use the SF–Holt–Winters Additive (Winters, 1960; Hyndman & Athanasopoulos, 2018) to predict the level of low-carbon remodeling of China’s manufacturing firms in the period of 2024–2030 [68,69], with the expressions of Equations (5) and (6):

$$A_t=\varphi x_t+(1-\varphi )(A_{t-1}+B_{t-1})$$

(5)

$$B_t=\pi (A_t-A_{t-1})+(1-\pi )B_{t-1}$$

(6)

where $x_t$ is the actual number of LCR degree in period t, $A_t$ is the adjusted smoothed value in period t, $B_t$ is the slope in period t, and ∅ and π are the smoothing parameters, which take values ranging from 0 to 1. The prediction model is Equation (6), where T denotes the number of periods in the prediction period away from the tth period.

$$F_{t+T}=A_t+B_{t}T$$

(7)

The optimal values of the smoothing parameters ∅ and π are calculated by MATLAB R2020b to be 0.312 and 0.768, respectively, and the predicted level of LCR of Chinese manufacturing enterprises in 2024–2030 is obtained. The results are shown in

Table 6. Future low-carbon reconfiguration projection data.

Year	2024	2025	2026	2027	2028	2029	2030
Degree of low-carbon reconfiguration	2.572	2.593	2.614	2.635	2.656	2.678	2.699

.

As can be seen from Table 6 that, although the sample size is limited, the trend analysis of the data shows that the degree of LCR of China’s manufacturing enterprises will show a steady increase in the coming period, with an average annual growth rate of 0.021. This result shows that China’s manufacturing industry has made significant progress in low-carbon transformation, laying an important foundation for realizing the “dual-carbon” target. According to a previous study, the EU manufacturing sector is in a leading position in terms of low-carbon transformation, with the average LCR level of its enterprises significantly higher than that of China, which further suggests that China still needs to make more efforts in low-carbon technology R&D and policy support (Busch, Foxon, & Taylor, 2018) [70]. This suggests that China still needs to further intensify its efforts in technology R&D, policy support, and international cooperation to accelerate the low-carbon transition process. In addition, with the rapid development of global low-carbon technologies and the looming goal of carbon neutrality, China’s manufacturing enterprises have great potential in the field of LCR, but also face multiple challenges in technological innovation, cost control, and policy guidance.

4.6. Robustness Tests

In order to ensure the reliability of the research results, the measure of R&D is replaced with the ratio of the number of R&D personnel to the total number of enterprises as the new measure [71]. Meanwhile, the reasonable choice of instrumental variables for dynamic panel generalized moments’ estimation can effectively control the endogeneity problem; therefore, this paper introduces the first-order lagged term of the explanatory variables as an instrumental variable, and chooses the two-step systematic generalized moments estimation method to conduct the robustness test for the above regression analysis. By searching the annual reports of the original sample enterprises and eliminating the enterprises that do not list the number of R&D personnel, 8967 new samples are finally obtained, and the results of the robustness test are shown in

Table 7. Robustness test.

	SCR	SCR	SCR
	(3)	(4)	(6)
YFRY	0.00402 ***	0.00220 **	0.00164 *
	(0.000838)	(0.000917)	(0.000929)
LCR		−0.0482 **	−0.826 **
		(0.0190)	(0.362)
LCR2			0.158 **
			(0.0774)
R&D	0.00196 ***
	(0.000250)
LCR*R&D		0.00870 ***	−0.0385 ***
		(0.000902)	(0.0140)
LCR2 *R&D			0.0190 ***
			(0.00565)
ROE	−0.00163 ***	−0.00164 ***	−0.00163 ***
	(0.000495)	(0.000494)	(0.000493)
Gross Profit	0.0949 ***	0.0937 ***	0.0922 ***
	(0.0103)	(0.0105)	(0.0104)
Cashflow	−0.0645 ***	−0.0711 ***	−0.0721 ***
	(0.0151)	(0.0151)	(0.0150)
REC	−0.0356 *	−0.0389 **	−0.0483 **
	(0.0193)	(0.0194)	(0.0195)
INV	0.697 ***	0.702 ***	0.703 ***
	(0.0199)	(0.0198)	(0.0198)
FIXED	0.0220 *	0.0190	0.0202 *
	(0.0116)	(0.0117)	(0.0117)
Intangible	0.0830 ***	0.102 ***	0.113 ***
	(0.0272)	(0.0271)	(0.0271)
Growth	−0.00513 ***	−0.00508 ***	−0.00511 ***
	(0.000390)	(0.000389)	(0.000389)
_cons	0.885 ***	0.987 ***	1.940 ***
	(0.00643)	(0.0459)	(0.421)
N	8967	8967	8967
R2	0.169	0.172	0.175
adj. R2	0.028	0.032	0.035

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

. The direction of the coefficients of the core variables and the significance level are consistent with the previous regression results, with no significant changes, indicating that the research conclusions have good robustness.

5. Discussion

5.1. Research Finding

This study examines the U-shaped relationship between LCR and SCR in the Chinese manufacturing industry and verifies the moderating role of R&D in the “LCR-SCR” relationship. Although the direct impact of low-carbon transition on firm performance has been widely discussed [9,10], we turn to the micro level to analyze how R&D activities affect supply chain system resilience during the LCR process, and further, we find that R&D inputs are able to mitigate the impact of organizational restructuring and technological change during the early stage of transition by accelerating the application of low-carbon technologies and optimizing the efficiency of resource integration. Further, it is found that R&D can effectively mitigate the loss of supply chain efficiency due to organizational restructuring and technological change at the early stage of transformation. It is found that LCR and SCR show a U-shaped relationship of inhibition followed by facilitation. This finding echoes the viewpoint of dynamic capability theory [37], which argues that firms need to acquire long-term competitive advantages through short-term adaptation costs. Specifically, the sunk costs of technology transition and resource reorganization reduce supply chain efficiency in the early stages of LCR [35], but when the degree of LCR exceeds the inflection point, the synergistic effects of low-carbon technologies begin to emerge and significantly enhance SCR. This result partially corroborates the findings of Zhao et al. (2021) [22] on the nonlinear effects of low-carbon transitions, but we further reveal the specific mechanisms by which they affect SCR. Notably, R&D inputs play a key positive moderating role in this process. Low-carbon oriented R&D effectively shortens the downward cycle of the U-curve by accelerating technology iteration [55]. This phenomenon is more significant in heavily polluting industries, consistent with the “environmental regulation-R&D compensation” mechanism proposed by Qiao-Xin (2016) [71]: strict low-carbon policies force firms to realize technological leapfrogging through R&D [54]. Moreover, the enabling role of digital technologies on R&D effectiveness further strengthens this regulatory pathway [57], complementing Guo & Li’s findings on R&D-supply chain synergies [39]. Finally, a prediction of the degree of low-carbon reconfiguration shows that the level of LCR in China’s manufacturing sector will continue to steadily increase, a trend that coincides with the path of achieving the global Sustainable Development Goals (SDGs) [2], with a view to providing a basis for the government to formulate differentiated policies and for firms to optimize the allocation of resources [3,22].

5.2. Theoretical Significance

The theoretical significance of this study is as follows: first, it reveals the mechanism of the U-shaped relationship between LCR and SCR and compensates for the limitations of the traditional linear assumption. Existing studies have generally explored the impact of LCR on supply chains from a linear perspective [20,22], or attributed SCR to static network structure design [11] and redundancy strategies [13]. This study found through dynamic capability theory and innovation theory that there is a significant U-shaped relationship between LCR and SCR [16,37]. This finding reveals a time-lag effect of resource restructuring and technology synergies: resilience is inhibited in the early stages by the cost of technology adaptation and the difficulty of supply chain coordination, and jumps in resilience in the later stages through the maturation of low-carbon technologies and the optimization of resource allocation. This mechanism provides empirical support for the theory of “dynamic construction of resilience” proposed by Christopher & Peck (2004) [7], and makes up for the neglect of nonlinear transition paths in existing studies. Secondly, it clarifies the regulating role of R&D inputs. Although the literature focuses on the direct impact of R&D on green innovation [41] or the enhancement of SCR [46], it does not systematically explain its moderating mechanism in the low-carbon transition. This study confirms that R&D not only directly improves SCR, but also mitigates the loss of resilience at the early stage of transition by accelerating technological synergy and resource integration. This finding deepens the understanding of the dynamic relationship between R&D and resilience, and provides a theoretical basis for enterprises to formulate stage-by-stage innovation strategies in the low-carbon transition.

5.3. Practical Inspiration

The findings of this paper have important practical guidance significance for manufacturing enterprises to realize low-carbon transformation and toughness enhancement, which are specifically manifested as follows:

(1)

Manufacturing enterprises should strategically coordinate LCR with technological innovation to achieve synergistic benefits between resource optimization and resilience enhancement. Specifically, for enterprises with limited low-carbon technological capabilities but strong innovation potential, priority should be given to advancing R&D efforts toward low-carbon technological breakthroughs. This approach can mitigate resource constraints during the initial transition phase through rapid technological upgrading. For firms with mature low-carbon technologies but weaker innovation systems, the focus should shift toward optimizing resource allocation and deeply embedding these technologies within existing production processes. This integration pathway accelerates the realization of resilience benefits.

(2)

Manufacturing enterprises should take advantage of the trend of digital transformation to realize the efficient integration of the low-carbon technology chain and supply chain. With the gradual enhancement of the enabling effect of digital technology on low-carbon transformation, enterprises should break the traditional resource allocation mode, realize the efficient collection and accurate allocation of resource data, and avoid the waste of resources and cost redundancy in LCR. By building a synergistic mechanism between the low-carbon technology chain and the supply chain, enterprises can identify and resolve the key bottlenecks in resilience enhancement, further enhancing the sustainability of low-carbon transformation.

(3)

Manufacturing enterprises should classify and manage the demand for low-carbon transformation, and develop differentiated resilience enhancement strategies. The core objective of LCR is to achieve long-term steady improvement of the enterprise resilience level through technological synergy and resource allocation optimization. To this end, enterprises should, according to their own resource endowment and low-carbon technology base, clarify the key links and priorities in the transformation process, and form a virtuous cycle of efficient resource allocation and continuous enhancement of resilience, oriented by dynamic capabilities.

6. Conclusions

6.1. Main Conclusions

Using a sample of Chinese A-share listed manufacturing companies (2011–2023), this study empirically analyzes the relationship between LCR, R&D, and SCR, and explores the moderating role of R&D in the LCR-SCR path. The study draws the following core conclusions:

(1)

LCR and SCR have a U-shaped relationship, with significant industry heterogeneity. The effect of LCR on SCR shows a nonlinear U-shaped relationship of inhibition followed by promotion, with an inflection point value of about 2.34. Initially, supply chain efficiency declines due to technological transformation and resource reorganization, but as low-carbon technology matures, synergistic effects gradually emerge, and long-term resilience is significantly improved. The U-shaped inflection point for heavily polluting industries (2.28) is earlier than that for non-heavily polluting industries (2.60), indicating that high environmental regulatory pressures force them to overcome the transition pains faster, but also to face more severe technological and resource challenges.

(2)

R&D directly enhances SCR. R&D improves flexibility and risk resistance through low-carbon technology breakthroughs and supply chain digitization. R&D optimizes the efficiency of resource integration and strengthens the synergy between supply chain nodes.

(3)

R&D significantly moderates the U-shaped relationship between LCR and SCR through technological innovation and resource allocation optimization: it not only mitigates the negative impacts in the early stage of transformation, but also strengthens the positive effects in the later stage, and shifts the inflection point to the left, accelerating the resilience improvement.

6.2. Limitations and Future Research

While this study sheds light on the relationship between LCR, R&D, and SCR in manufacturing enterprises, several limitations remain:

(1)

This paper does not explore the segmentation of R&D. R&D inputs during LCR may encompass multiple dimensions, such as technology R&D, equipment upgrades, and low-carbon product development. Future research could investigate the varying impacts of different R&D types on the “LCR–SCR” relationship, providing more precise guidance for firms to formulate fine-grained innovation strategies.

(2)

The sample primarily focuses on listed companies in the Chinese manufacturing sector. Due to differences in resource endowments, technological capabilities, and transformation goals between listed and unlisted firms, the generalizability of the findings may be limited. Future studies could expand the sample to include unlisted manufacturing firms, enhancing the broader applicability and relevance of the results.

(3)

The deterministic demand assumption adopted in this study, while simplifying the modeling process, cannot fully reflect the stochastic fluctuation characteristics of demand in the real market, thus bringing limitations. In the future, we can combine the institutional theory to quantify the impact of SCR demand fluctuations through stochastic process modeling, explore the heterogeneous impact of LCR on SCR under different policy environments, or use complex system simulation to simulate the dynamics of the supply chain disruption scenarios under the adaptive mechanism to improve the LCR prediction framework under uncertainty.