A study on the impact of new quality productivity on total factor productivity

1 Introduction

Manufacturing is the cornerstone of a nation’s economy and its overall strength. As a manufacturing powerhouse, China has developed a comprehensive industrial system that encompasses all 41 major industrial categories, 207 medium-level categories, and 666 minor categories as defined by the United Nations Industrial Classification. It is the only country globally to cover all industrial sectors listed by the UN (Zhao et al., 2021). Since 2010, China’s manufacturing industry has been the largest in the world for 11 consecutive years. However, this sheer size and extensive coverage do not necessarily indicate a strong foundation. China’s manufacturing sector grapples with the challenge of being “large but not strong” (Ji, 2022). Specifically, within the global four-tier manufacturing hierarchy, China remains in the third tier, significantly trailing behind manufacturing leaders such as the United States, Europe, and Japan (Brondoni, 2020).

The report to the 20th CPC National Congress clearly articulates that promoting high-quality development must remain a central focus, with improving total factor productivity serving as a key benchmark for development. As the engine of the national economy and the lifeblood of the real economy, the transformation and upgrading of the manufacturing sector have a profound impact on the overall socio-economic landscape. Enhancing TFP is therefore a crucial way to build a modern industrial system and drive high-quality development. However, China’s manufacturing sector currently faces numerous challenges: domestic factor costs, including labor, continue to rise, while resource and environmental constraints become increasingly stringent. Internationally, developed nations are advancing toward “re-industrialization,” while some developing countries leverage low-cost advantages, applying “low-end substitution” pressure. These factors collectively limit the growth of China’s manufacturing TFP (Wang et al., 2021). As the manufacturing sector transitions from rapid growth to high-quality development, persistently low productivity has become a significant barrier to transforming the nation from a manufacturing giant into a true manufacturing powerhouse (Pan and Zhao, 2020). Amid the rapid advancement of a new wave of technological revolution and industrial transformation, coupled with increasing competition among global powers, China’s economic development now stands at a pivotal point that calls for a shift from factor-based growth to innovation-driven growth. These converging historical circumstances demand higher standards for manufacturing productivity. Improving TFP is critical for achieving sustainable, high-quality economic growth and is essential for building a modern socialist powerhouse.

In this macroeconomic context, the emergence of new quality productivity (NQP) is essential. This concept was first introduced by General Secretary Xi Jinping during his inspection tour of Heilongjiang Province in September 2023. In his speech, the General Secretary emphasized the importance of actively developing strategic emerging industries and future industries—including new energy, new materials, advanced manufacturing, and electronic information—to expedite the development of new quality productivity and strengthen new growth drivers (Xu et al., 2024).

New quality productivity transcends traditional models of inefficient and resource-intensive growth. It represents an advanced state of productive forces catalyzed by technological innovation and industrial upgrading, occurring within the context of deep convergence between informatization and intelligentization. This transformation is achieved by enhancing the quality of production factors, reshaping platform-based organizational structures, steering the industrial system toward strategic emerging and future industries, and empowering the innovation ecosystem (Du et al., 2023). In contrast to traditional views of productive forces, NQP arises from breakthroughs in key and disruptive technologies, manifesting as new forms of productive forces characterized by innovative technologies, economies, and business models (Zhou and Xu, 2023).

Scholars have explored the essence of NQP by analyzing its two main dimensions: “novelty” and ‘quality. (Zhang and Tang, 2024) Novelty is represented by emerging production factors and their innovative combinations, while quality reflects robust industrial foundations and strong developmental momentum. The synergy of these two dimensions drives transformative changes in production models and facilitates significant qualitative shifts in developmental outcomes. Furthermore, novelty highlights the characteristics of the current era driven by emerging technologies, while quality emphasizes the fundamental transformation of traditional productive forces achieved through the deep integration of scientific innovation and the real economy (Su and Sun, 2024). Jiang and Qiao (2024) identified the profound essence of NQP as encompassing three key dimensions: ‘novelty,’ ‘quality,’ and ‘force.’ The ‘novelty’ dimension includes new factors, technologies, and industries, marking a departure from conventional factor combinations. The “quality” dimension signifies high-quality development and multi-attribute quality, indicating a qualitative leap in productive forces. Lastly, the “force” dimension reflects specific manifestations such as digital, collaborative, green, blue, and open productivity, underscoring its role as a core endogenous driver of economic progress.

Developing new quality productivity can enhance the efficient allocation of innovative factors, including technology, data, and talent. This improvement fosters better energy and resource utilization within the manufacturing sector, supports the clean transformation of production processes, and thereby increases overall productivity and sustainable development levels. Additionally, it can empower research and development, production, and management processes through the deep integration of new-generation information technologies—such as data components and large-model algorithms—with manufacturing. This integration can stimulate disruptive innovation and efficiency breakthroughs, providing crucial momentum for the transformation and upgrading of the manufacturing sector (Zhao and Sun, 2025).

The origins of total factor productivity (TFP) can be traced back to Tinbergen. In 1957, Solow formally introduced the concept of total factor productivity. Through the “Solow residual method” (Solow, 1957), technological progress was systematically incorporated into economic growth models.

Solow defined TFP as the portion of output growth attributable to unobservable, exogenous, Hicks-neutral technological progress. Because TFP is widely used to assess the quality of economic development, it has attracted substantial academic attention since its introduction.

At the firm level, TFP reflects the overall efficiency with which different input factors are transformed into final output. Existing studies typically examine TFP determinants from two perspectives. One perspective focuses on endogenous firm characteristics. The other perspective emphasizes exogenous external environments.

On one hand, from an internal perspective, both R&D intensity and intelligent development significantly boost enterprise TFP, but only moderate R&D intensity produces the strongest impact (Mao et al., 2013). From an external environment perspective, the extensive use of internet technology can facilitate knowledge sharing through spillover effects, thereby enhancing enterprise TFP (Ahangama et al., 2025). At the same time, the digital economy drives innovation that promotes the sustainable growth of enterprises’ TFP (Sturgeon, 2021). Research has expanded from macro to micro levels—early scholars mainly studied at national, regional, and industrial scales, while later studies increasingly focused on micro-level entities like enterprises, leading to innovations in measurement methods.

Manufacturing remains the cornerstone of a nation’s economy and its overall strength. As a manufacturing powerhouse, China has developed a comprehensive industrial system. However, the sector still grapples with being “large but not strong,” remaining largely in the third tier of the global manufacturing hierarchy. In this macroeconomic context, promoting high-quality growth by enhancing Total Factor Productivity (TFP) is critical (Xie et al., 2024). To drive this transition from factor-based growth to innovation-driven growth, the emergence of New Quality Productivity (NQP) is essential (Zhou and Fang, 2026).

To establish a rigorous foundation, this study delineates the conceptual and operational boundaries between NQP and TFP across four distinct dimensions:

Theoretical Origins and Paradigms: NQP is deeply rooted in contemporary political economy, mapping how systemic scientific revolutions structurally alter the classic combination of labor, instruments of labor, and objects of labor. Conversely, TFP originates as a derivative accounting residue within neoclassical growth frameworks, treating technological progress primarily as an exogenous, Hicks-neutral efficiency factor (Liu, 2025; Guo and Fan, 2022).

Conceptual Attributes: NQP represents a “qualitative state transition” of the economic structure, targeting adjustments toward future-oriented, strategic emerging industries. TFP, in contrast, functions as an “efficiency metric” evaluating how intensely and seamlessly existing resources are utilized (Gong and Dai, 2024).

Driving Operational Logic: NQP focuses on the endogenous generation and innovative synthesis of production factors—such as big data and high-tier human capital—pushing out the production-possibility frontier (Zhou and Fang, 2026). TFP acts as a tracking tool for resource allocation efficiency given a fixed configuration of production boundaries (Gong and Dai, 2024).

Macro-Strategic Application: NQP serves as a grand economic development paradigm designed to build new structural production relations and guide national industrial policies. TFP primarily functions as an international quantitative standard utilized to measure performance and test the quality of growth (Solow, 1957; Zhou and Fang, 2026).

In summary, NQP and TFP are profoundly interconnected yet separate: NQP describes the future-oriented qualitative transformation of productive forces, providing the directional vector, whereas TFP serves as the core quantitative yardstick to test whether this structural transformation achieves actual development efficiency.

Furthermore, the structural upgrading of the manufacturing sector cannot be detached from the global imperative of the circular economy (CE). Traditional manufacturing models operate on linear paths (‘take-make-dispose’), which are highly vulnerable to ecological and resource constraints. The transition to a circular model—which prioritizes resource reduction, reuse, and closed-loop recycling—requires systemic efficiency shifts at the firm level. However, a critical research gap remains in the existing literature: while the theoretical necessity of CE is widely acknowledged, few empirical studies demonstrate how emerging macroeconomic paradigms like NQP structurally translate into micro-level efficiency gains (TFP) that make circular loops economically viable for enterprises. By investigating this connection, this study bridges the gap between green technological transformation and circular resource optimization.

This study is positioned at the micro-level of enterprises, aiming to investigate the total factor productivity of manufacturing firms to establish a stronger foundation for advancing high-quality development in the manufacturing sector. While existing literature has produced valuable insights, this study aims to contribute in the following ways: ① Exploring the mechanisms by which new quality productivity influences the total factor productivity of manufacturing; ② Empirically testing the impact of new quality productivity on manufacturing TFP, including robustness checks and heterogeneity analyses; and ③ Theoretically clarifying the causal pathways through which new quality productivity affects manufacturing TFP, utilizing mediation models to investigate transmission mechanisms involving technological innovation and digital transformation empirically.

2 Theoretical analysis and research hypotheses

2.1 Theoretical analysis of how new quality productivity influences enterprise TFP

The development of NQP leads to significant transformations in the workforce and production methods. A skilled workforce familiar with advanced technologies helps manufacturing firms improve efficiency and solve longstanding structural constraints (Zheng et al., 2025). This shift to intelligent resource utilization, driven by big data and AI, facilitates improved decision-making and optimization (Masulis et al., 2025; Yang et al., 2026). These changes collectively support the high-quality development of manufacturing enterprises, leading to our first hypothesis (Feng et al., 2025):

H1: New quality productivity significantly enhances the TFP of manufacturing enterprises.

2.2 The mediating role of technological innovation

New technologies and innovative business models represent disruptive forces that shift the production frontier (Tang and Chen, 2026). By integrating advanced tools like AI and the Internet of Things (IoT) into current production processes, firms can transition toward high-value, customized output (Acemoglu and Restrepo, 2016). This strategic integration prevents resource misallocation, reduces operational idle times, and eliminates inefficient manufacturing stages (Huang, 2019). Additionally, the development of NQP demands high-tier human capital, prompting continuous skill renewal among employees (Yu et al., 2025). Therefore, NQP acts as a catalyst for systemic technological innovation, which in turn drives overall firm efficiency:

H2: New quality productivity promotes TFP growth through the mediating effect of technological innovation.

2.3 The mediating role of digital transformation (RBV perspective)

Drawing on the Resource-Based View (RBV) and Dynamic Capabilities Theory, digital transformation serves as a critical mechanism through which firms configure valuable, rare, and inimitable digital assets to generate sustainable competitive advantages (Zhu et al., 2023). NQP inherently fosters an automated, flexible, and data-driven manufacturing environment (Wang et al., 2024).

First, digitalization mitigates information asymmetry and transaction costs. Internally, real-time data flows enable managers to make informed decisions, reducing departmental barriers and resource misallocation (Guo et al., 2025). Externally, firms can detect shifts in customer demand and supply chain dynamics at lower costs, facilitating agile operational adjustments. Second, digital integration accelerates research and development (R&D) cycles through digital simulation and AI-driven modeling, enhancing overall innovation performance (Cong and Jiayang, 2024). Consequently, digital transformation optimizes the deployment of existing assets and unlocks higher productivity levels:

H3: New quality productivity enhances the TFP of manufacturing enterprises by accelerating the depth of digital transformation.

3 Research design and model specification

To analyze the effect of new quality productivity on corporate TFP, (see Figure 1) we specify the following two-way fixed-effects model:

TFP it = cNQP it + β n Control it + δ i + γ t + ε it

Where the dependent variable TF P it represents the TFP of the firm i in year t; the core explanatory variable NQ P it denotes the level of new quality productivity; Contro l it represents the vector of control variables; δ i and γ t denote firm and industry/year fixed effects, respectively, and ε it is the error term.

3.1 Instrumental variables and identification strategy

To address potential endogeneity arising from omitted variables or reverse causality, this study employs a multi-instrumental 2SLS estimation strategy. We introduce three instrumental variables (IV 1,IV 2, IV 3):

• 1) Historical Telecommunications Endowment (IV1): Calculated as the city-level volume of postal and telecommunications services in 1984, interacted with a time trend. This historically preset milestone satisfies the “exogeneity condition” as it predates the sample period (2015–2024), while its progressive technological diffusion directly relates to modern digital infrastructure.
• 2) Confucian Cultural Heritage (IV2): Measured as the interaction term between the number of historical Confucian academies in the prefecture-level city and a time trend. Confucian culture fosters long-term educational accumulation and human capital cultivation—key components of NQP—while exerting no direct impact on modern firm-level accounting TFP.
• 3) Natural Geographic River Density (IV3): Calculated as the interaction of city river density with a time trend. Geographic features historically shaped trade routes and talent agglomeration, correlating with regional innovation capacity, while satisfying the exclusion restriction for individual firm-level productivity (Lu and Lian, 2012; Liu, 2025). The final measurement indicators for the NQP level of enterprises are shown in Table 1.

Table 1. Evaluation indicator system for new quality productivity in manufacturing enterprises.

Data source: CSMAR database.

3.1.1 Instrumental variables

The instrumental variables used in this study are: (1) Technological Innovation (Innovation). Based on the work of relevant scholars (Li et al., 2024), this study measures techno-logical innovation levels using the natural logarithm of the total number of patent applications filed by enterprises for invention patents, utility model patents, and design patents, plus one. (2) Digital Transformation. Drawing on the research of Li and Tian (Li and Tian, 2023), text analysis is used to identify keywords related to digital transformation within listed companies’ annual reports and to count their frequencies. This results in a “textual intensity” for corporate digital transformation. The logarithm of the frequency of digital keywords in the annual report plus one is employed as a proxy variable to create a corporate digital transformation index, which serves as the mediating variable in the mechanism analysis.

3.1.2 Control variables

Based on theoretical analysis, the framework constructed here, and common practices in significant existing literature (Guo and Zhou, 2025), the following control variables were selected: fixed asset ratio (FAR), growth capability (GROW), operating cash flow ratio (OCFR), Tobin’s Q (TBQ), capital intensity (FATA), ownership structure (SOE), dual-role management (DUAL), firm age (AGE), proportion of independent directors (PID), and ownership share (OC10). These control variables are widely recognized as key factors influencing a firm’s TFP. Including them in the model helps accurately identify and interpret the impact of new quality productivity.

3.1.3 Moderating variable

Drawing on Wu et al. (2020), the debt-to-equity ratio is used as the moderating variable. This metric indicates the level of debt financing and financial risk in enterprises, serving as a key channel for analyzing how new quality productivity affects total factor productivity through the debt-to-equity ratio.

3.2 Model specification

• (1) Basic Model Specification. To analyze the effect of new quality productivity on enterprises’ total factor productivity, a fixed-effects model was used for estimation. Additionally, this study adjusts its standard deviation to account for potential estimation bias caused by heteroscedasticity. The specific formula is as follows:

TF P it = cNQP F it + β n Contro l it + δ i + γ t + ε it (1)

Where the dependent variable TF P it represents the TFP of firm i in year t; the core explanatory variable NQP F it denotes the level of new quality productivity for the firm i in year t; Controlit constitutes the control variables for the firm i in year t, δ i, γ t, and ε it respectively represent the time fixed effect, industry fixed effect, and error term.

In Equation (1), δ_i denotes firm-level individual fixed effects that control for time-invariant unobserved heterogeneity across firms, while γt denotes the industry fixed effects (or the relevant grouped fixed effects) included in our two-way fixed-effects estimation. This specification ensures that the estimated impact of NQP on TFP is identified net of persistent firm characteristics and industry structural heterogeneity.

• (2) Mechanism testing analysis. To investigate the mediating effects of technological innovation and digital transformation, this study applies the stepwise regression procedure proposed by Wen et al. to construct the empirical models for mediation analysis (Wen and Ye, 2014).

M it = aNQP F it + β n Contro l it + δ i + γ t + ε it (2)

TF P it = c ′ NQP F it + b M it + β n Contro l it + δ i + γ t + ε it (3)

In this model, TF P it denotes total factor productivity, NQP F it represents new quality productive forces, and M it stands for the mediating variable (technological innovation or digital transformation). Contro l it is a vector of control variables, and ε it is the random disturbance term. Within this framework, the coefficient in Equation (1) measures the total effect of NQPF on TFP; the coefficient in Equation (2) estimates the effect of NQPF on the mediating variable (technological innovation or digital transformation); and in Equation (3), the coefficient b quantifies the effect of the mediator on TFP after controlling for NQP, while c ′captures the direct effect of NQP after accounting for the mediator. Accordingly, the mediating effect is calculated as a × b, and the total effect decomposes as c = c ′ + a × b. With the other variables remaining the same as in Equation (1).

To further verify the robustness of the results, a bias-corrected non-parametric percentile bootstrap method (with 1,000 resamples) was employed to directly estimate the mediating effect (a × b) and compute its 95% confidence interval. If the confidence interval does not contain zero, the mediating effect is deemed statistically significant. Details on variable selection are provided in Table 2, while descriptive statistics are shown in Table 3.

Table 2. Variable selection table.

Table 3. Descriptive statistics.

All baseline and robustness regressions are estimated with fixed effects to control for unobserved heterogeneity, including firm fixed effects.

4 Empirical analyses

This section presents empirical evaluations that test the study’s core hypotheses. We begin with baseline fixed-effects regressions that examine the direct impact of new quality productivity on firm-level total factor productivity. To ensure the reliability of our findings, we conduct a series of endogeneity treatments using historical and geographical instrumental variables, alongside rigorous robustness checks. Furthermore, we explore the contextual boundaries through moderation and heterogeneity analyses across geographical regions and manufacturing subsectors, followed by an empirical examination of the underlying mechanism channels.

4.1 Benchmark regression analyses

Based on the results of the Hausman test, the fixed effects model was chosen. Table 4 presents the estimation results for the basic two-way fixed effects model. We estimated a two-way fixed-effects model that included firm-level fixed effects and industry fixed effects, which jointly control for time-invariant firm heterogeneity and industry-specific unobserved shocks. The impact of new quality productivity on the TFP of manufacturing enterprises is significantly positive at the 1% level. These findings strongly support Hypothesis H1, which states that new quality productivity has a significant positive effect on TFP.

Table 4. Benchmark regression results.

*, , * denote significance at the 1, 5, and 10% levels, respectively; figures in parentheses are t-statistics based on cluster-robust standard errors. All regressions include firm fixed effects; the table additionally reports year and industry fixed effects as indicated.

Regarding the economic significance of the coefficient, we clarify that since the dependent variable TFP is specified in natural logarithmic form (lnTFP) while NQP is in level form, the model is a log-level model. Therefore, the coefficient of 0.032 represents a semi-elasticity, implying that a one-unit increase in NQP leads to an approximate 3.25% increase in firm-level TFP. Given the sample distribution, moving a firm from the 25th to the 75th percentile in NQP results in a substantial 31.3% increase in TFP, which is economically meaningful compared to the low baseline growth rate of manufacturing TFP.

4.2 Endogeneity and robustness tests

4.2.1 Endogeneity analysis

Although other variables influencing manufacturing enterprises’ total factor productivity are controlled, endogeneity bias likely persists in assessing the impact of new quality productivity on TFP due to data limitations and unobservable factors. For example, fintech (Zeng and Wang, 2023), digital transformation, high-quality labor agglomeration, and financial agglomeration (Gu et al., 2023) all significantly enhance TFP. Furthermore, potential measurement errors in composite indicators of new quality productivity, especially those related to unobservable factors affecting corporate TFP, may create endogeneity bias in the model. Additionally, higher regional levels of new quality productivity often correlate with stronger green technological innovation capabilities and resource efficiency, which directly increases local manufacturing enterprises’ TFP. Therefore, applying instrumental variable methods is further justified to address potential endogeneity issues.

Firstly, drawing on existing research (Cheng et al., 2023), we select the total volume of postal and telecommunications services in cities during 1984 as the instrumental variable. Since China’s urban economic system reform began in 1984, it makes sense to use that year as the baseline for historical cross-sectional data. Because it is a historically preset milestone, it meets the requirements of “historical exogeneity” and effectively allays worries about reverse causality because it occurs enough before our sample period (2015–2024). Two factors further support this decision: first, the 1984 postal and telecommunications volume reflects the region’s early information infrastructure level, which established a strong foundation for contemporary digital technologies and innovation activities, improving the quality and efficiency of new productivity development; second, because historical data primarily relates to communications infrastructure, it has no direct connection to current manufacturing TFP, satisfying the exogeneity condition. Given its time-invariant nature, we multiply it by the year to create a time trend term, which serves as the final instrumental variable, denoted as IV1. Drawing on the empirical paradigm established by (Nunn and Qian, 2011; Autor and Dorn, 2013), we justify our instrument construction as follows: the interaction of historical cross-sectional data with a time trend constructs a time-varying panel instrument. The underlying theoretical logic is that the impact of historical telecommunications endowment on contemporary new-quality productivity is not static; rather, it is progressively released and cumulatively amplified alongside technological progress and economic development. Consequently, the interaction term with the year variable is the necessary specification to capture this dynamic cumulative effect.

Secondly, the value orientations promoted by Confucian culture—such as the tradition of valuing education and accumulating human capital—can influence the development of new quality productivity through channels that affect innovation awareness and knowledge accumulation, thereby satisfying the instrument variable’s correlation requirement. At the same time, as a non-formal institution formed over historical time, Confucian culture primarily transmits its value concepts through implicit means such as cultural ethics and social norms. It has no direct correlation with current TFP levels, thus meeting the exogeneity condition. Therefore, based on existing research (Zhao et al., 2025), this study uses the interaction term between the logarithm of the number of Confucian academies in the prefecture-level city where manufacturing listed companies are registered and the year as an instrumental variable, denoted as IV2. Data on Confucian academies are obtained from the CNRDS Confucian Culture Database.

Thirdly, based on the research of Li et al. (2025), the interaction term between a prefecture-level city’s river density and the time trend is used as an instrumental variable, called IV3. As a natural geographic feature, river density can, on the one hand, reduce transportation costs, increase appeal to talent, and attract human capital. However, regional natural geographic factors limit the number of urban rivers. It satisfies the requirements for an instrumental variable since it has no direct impact on firm total factor productivity.

To ensure the statistical validity of our identification strategy, all three instrumental variables (1984 postal and telecommunications volume, number of Confucian academies, and river density) were introduced simultaneously into the first-stage regression. The joint inclusion of multiple instruments enables an overidentification test (Hansen J test, with a p-value > 0.10), confirming that the instruments were jointly exogenous and satisfied the exclusion restriction. Furthermore, the weak instrument test revealed a Kleibergen-Paap Wald F-statistic well above the critical value of 10, confirming instrument relevance.

The results of the 2SLS regression using instrumental variables are shown in Table 5. The null hypothesis of weak instrument identification is rejected at the 1% level by the K-P rk LM statistics for each of the three sets of instrumental variables. The lack of weak instruments is indicated by the C-D Wald F-statistic exceeding the crucial value for Stock–Yogo at the 10% level. This demonstrates that the selected instrumental variables are reliable and valid. The calculated coefficients for new quality productivity are considerably positive at the 1, 1, and 10% levels, respectively, after fully addressing any endogeneity issues from reverse causality, as seen in columns (2), (4), and (6). In particular, river density and the volume of postal and telecommunications services have a beneficial impact on new quality productivity, whereas Confucian academies have a negative effect. This reflects the different mechanisms through which various historical factors affect innovation development. This aligns with the benchmark regression results, further confirming that new quality productivity helps improve TFP levels.

Table 5. Robustness analysis – endogeneity test.

Data within {} denotes p-values, while data within [] indicates critical values for the Stock–Yogo weak identification test at the 10% significance level.

4.2.2 Robustness test

In economics, productivity and efficiency are often used interchangeably. Common methods for measuring TFP include the OLS, OP, LP, and GMM methods. The fixed effects method has limited ability to correct endogeneity problems inherent in least squares estimation, whereas the OP method effectively addresses issues arising from theoretical measurement bias and sample selection bias (Lu and Lian, 2012). Therefore, this study employs the OP method to recalculate manufacturing total factor productivity as the primary dependent variable, followed by regression estimation using Equation (1). As reported in Column (1) of Table 6, the estimated coefficient for new quality productivity remains significantly positive at the 1% level, confirming that corporate new quality productivity continues to foster firm TFP even under alternative variable measurements. To mitigate the potential bias introduced by extreme values in the financial and operational distributions, we perform a bilateral trimming procedure by removing the top and bottom 1% tails of all continuous variables. As displayed in Column (2) of Table 6, the estimated coefficient for new quality productivity on firm TFP remains statistically significant at the 1% level, indicating that the observed positive relationship is not driven by extreme outliers and reaffirming the robustness of the benchmark results. During the sample observation period, the exogenous shock of the COVID-19 pandemic between 2020 and 2022 generated structural fluctuations in business operations. To prevent these abnormal macro-level conditions from confounding our estimates, we exclude the observations spanning 2020 through 2022 and re-estimate the baseline model. Column (3) of Table 6 shows that the coefficient for enterprise new quality productivity remains positive and significant at the 1% level, demonstrating that the structural productivity-enhancing effect persists independently of extraordinary external disruptions. Given the specialized administrative hierarchy, resource concentration, and policy advantages of municipalities directly governed by the central government, firms located in these areas operate under distinct environmental dynamics compared to those in typical prefecture-level cities. To ensure our findings are generally applicable across broader regional frameworks, we rerun the regression after excluding firm observations from Beijing, Shanghai, Tianjin, and Chongqing. The empirical results presented in Column (4) of Table 6 confirm that the coefficient of new quality productivity remains significantly positive at the 1% level, proving that the productivity gains induced by NQP represent a pervasive baseline phenomenon across China’s regional economic landscape. Addressing potential reverse causality where higher firm TFP might simultaneously facilitate the adoption of new quality productive forces, we re-estimate the model by using the one-period lagged term of new quality productivity (L.NQP) as an explanatory variable. As shown in Column (5) of Table 6, the coefficient for the lagged NQP variable remains positive and statistically significant at the 1% level, supporting the temporal precedence of NQP in driving TFP growth and alleviating reverse causality concerns.

Table 6. Robust test results.

NQP, new quality productivity; *, , * denote significance at the 1, 5, and 10% levels, respectively; figures in parentheses are t-statistics based on cluster-robust standard errors.

To address potential functional specification bias and ensure transparency regarding variable transformations, we re-estimate our baseline model by performing a natural logarithmic transformation on our core explanatory variable, thereby altering the specification from a log-level model to a log–log elasticity model. The specification is modeled as:

ln TF P it = α ln NQP F it + β n Contro l it + δ i + γ t + ε it (4)

The regression results are integrated into Column (6) of Table 6. The estimated coefficient for ln NQPF is 0.086, which is statistically significant at the 1% confidence level (t-statistic = 11.42).

When comparing economic significance across the two functional forms, the specifications yield highly consistent interpretations. In our benchmark log-level setup, the semi-elasticity coefficient of 0.032 indicates that moving an enterprise from the 25th percentile to the 75th percentile of the NQP score results in a cumulative TFP improvement of roughly 31.3%. Under the log–log configuration, executing an identical cross-percentile translation projects an estimated TFP improvement of 29.8%. The narrow variance between these two evaluations firmly rejects any functional form anomalies, proving that the positive economic impact of new quality productive forces on firm efficiency is highly robust and structurally invariant to logarithmic adjustments.

4.3 Moderating effect analysis

Following Wu et al. (2020), we use the debt-to-asset ratio as the moderating variable. Table 7, Columns (1)–(3), shows regression results for the moderating effect of the debt-to-asset ratio. Column (2) shows a significantly positive coefficient for the debt-to-asset ratio, indicating that a moderate debt level provides firms with essential financial support for resource allocation transformation, thus directly improving TFP. Column (3) reveals that the interaction term coefficient between new quality productivity and the debt-to-asset ratio is significant at the 10% level. This suggests that the debt-to-asset ratio “amplifies” the positive effect of new quality productivity on TFP. The higher the debt-to-asset ratio, the stronger the positive effect of new quality productivity on TFP.

Table 7. Moderation effect test.

NQP, new quality productivity; *, , * denote significance at the 1, 5, and 10% levels, respectively; figures in parentheses are t-statistics based on cluster-robust standard errors.

4.4 Heterogeneity analysis

4.4.1 Regional heterogeneity analysis

Significant regional disparities in China’s business environment and institutional frameworks greatly influence corporate decision-making (Chen, 2025). To ensure a certain degree of comparability across groups, following the approach of Huang Hongbin et al., this study employs the Chow test to examine intergroup coefficient differences (Huang et al., 2016). Therefore, this study divides the sample into four groups—Eastern, Central, Western, and Eastern region—based on the regional classification from the National Bureau of Statistics. The regression results are shown in Table 8. Findings indicate that the positive impact of new quality productivity on firms’ TFP varies notably across regions. First, NQP exerts a significantly positive effect on TFP in the eastern, central, and western regions, whereas the coefficient for the northeastern region is negative and statistically insignificant. This pattern is broadly consistent with China’s regional development landscape. The eastern region, benefiting from advanced digital infrastructure, vibrant innovation ecosystems, and abundant human capital, is better positioned to translate NQP into efficiency gains. The central and western regions, supported by industrial relocation and targeted policy incentives, are gradually unlocking the growth potential of NQP. In contrast, the northeastern region, constrained by industrial rigidity, institutional frictions, and talent outflows, has yet to fully realize the enabling effect of NQP (Xiaodong et al., 2022). Second, although the coefficient for the eastern region (0.035) appears numerically higher than the coefficients for the central (0.030) and western (0.032) regions, Chow test cross-model coefficient comparisons reveal that these differences are not statistically significant. This indicates a pervasive and balanced driving effect of NQP across these three primary macro-regions. In contrast, the coefficient for the Northeast region is statistically insignificant, which reflects its ongoing structural constraints but must be interpreted cautiously due to a smaller sample size (n = 857).

Table 8. Regional heterogeneity analysis.

NQP, new quality productivity; *, , * denote significance at the 1, 5, and 10% levels, respectively; figures in parentheses are t-statistics based on cluster-robust standard errors.

These findings align with the “gradient advancement” characteristic of China’s regional development.

4.4.2 Analysis of industry heterogeneity

If we examine this issue solely at the macro level of the manufacturing sector, we observe that new quality productivity contributes to enhancing the overall TFP efficiency of China’s manufacturing enterprises. However, without in-depth comparative analysis across subsectors, there is a risk of underestimating the heterogeneous nature of this impact. This not only hinders a precise understanding of the intrinsic mechanisms linking new quality productivity to TFP but also complicates the formulation of differentiated development policies tailored to the distinct characteristics of various manufacturing industries. Therefore, this study adopts the methodology proposed by Xiaodong et al. (2022). The entire manufacturing sector within the National Economic Industry Classification is grouped into 14 industries: Industry 1 covers agricultural and sideline product processing, food manufacturing, and the production of alcoholic beverages, beverages, and refined tea; Industry 2 includes textiles, textile and apparel, and leather, fur, feather, and related products, along with footwear manufacturing; Industry 3 in-volves wood processing, as well as the manufacturing of wood, bamboo, rattan, palm, and grass products; and Industry 4 pertains to study and study products, printing and reproduction of recorded media, and educational materials. Artistic, sporting, and entertainment goods manufacturing; Industry 5: petroleum, coal, and other fuel processing; Industry 6: chemical raw materials, chemical products, pharmaceutical manufacturing, chemical fiber manufacturing, rubber, and plastic products manufacturing; Industry 7: non-metallic mineral products manufacturing; Industry 8: ferrous metal smelting and rolling, non-ferrous metal smelting and rolling, metal products manufacturing; Industry 9: general machinery and specialized equipment manufacturing; Industry 10: motor vehicle manufacturing, railway, ship-building, aerospace, and other transport equipment manufacturing; Industry 11: electrical machinery and apparatus manufacturing; Industry 12: computer, communication, and other electronic equipment manufacturing; Industry 13: instrument and control apparatus manufacturing; Industry 14: other manufacturing, resource recovery from waste, and repair of metal products, machinery, and equipment.

Table 9 presents the heterogeneous effects of new quality productivity (NQP) on the TFP of listed manufacturing firms across different industries. First, NQP exerts a significantly positive effect on most manufacturing subsectors. The coefficients are particularly prominent in Industry 3 (wood processing and furniture manufacturing), Industry 6 (chemical raw materials, chemical products, and pharmaceutical manufacturing), and Industry 8 (ferrous and non-ferrous metal smelting and rolling), all of which are significant at the 1% level. This suggests that NQP plays a critical role in breaking the “low-end lock-in” of traditional factor-dependent industries through technological upgrading and equipment modernization. Second, notable heterogeneity exists across sectors. The coefficient for Industry 5 (petroleum, coal, and other fuel processing) is significantly negative, which may be attributed to the stringent environmental regulations and carbon reduction pressures faced by energy-intensive industries, where the short-term transition costs of NQP outweigh efficiency gains. In contrast, Industry 2 (textile and apparel)​ and Industry 14 (other manufacturing and repair) show insignificant coefficients, reflecting the limited sensitivity of labor-intensive industries to productivity shocks due to weaker technological absorptive capacity. Third, regarding statistical significance, the Chow test confirms that the coefficient differences for most industries are significant at the 1% level. For Industry 3, the test yields a p-value of 0.054, indicating marginal significance at the 10% level. This is largely driven by the relatively small sample size (Observations = 339) and high within-industry heterogeneity, which increases estimation variance. Nevertheless, the standardized coefficient (0.078) for Industry 3 is among the largest across all sectors, indicating that the economic magnitude of NQP in this industry remains substantial. Fourth, concerning sample size, industries such as Industry 5 (n = 133) and Industry 14 (n = 165) have limited observations, which may affect estimation precision. However, the main conclusions remain robust after adjusting for cluster-robust standard errors and conducting Chow tests.

Table 9. Heterogeneous impact of new quality productivity on China’s listed manufacturing industries.

NQP, new quality productivity; *, , * denote significance at the 1, 5, and 10% levels, respectively; figures in parentheses are t-statistics based on cluster-robust standard errors.

4.5 Further analysis

The theoretical analysis presented earlier suggests that new quality productivity impacts TFP through technological innovation and digital transformation within enterprises. Therefore, this study investigates the presence of mediating effects to validate the operational mechanism by which new quality productivity influences enterprise total factor productivity. (1) Mediating Variable: Technological Innovation. Existing literature often uses R&D expenditure or patent counts to measure technological innovation capacity. Drawing on scholars (Li et al., 2024), this study assesses corporate technological innovation levels using the natural logarithm of the total patent applications (including invention patents, utility model patents, and design patents) plus one. The results are shown in Table 10. Column (1) indicates that new quality productivity significantly enhances the TFP levels of manufacturing enterprises. In Model (2), the coefficient for new quality productivity is significantly positive, suggesting that new quality productivity considerably promotes corporate technological innovation levels. In Column (3), after controlling for new quality productivity, the coefficient for technological innovation on corporate-rate-rate TFP is 0.0611 and significant at the 1% level, demonstrating that corporate technological innovation drives improvements in TFP. The mediation effect of techno-logical innovation is 0.00785, accounting for 24.5% of the total effect. This confirms that technological innovation plays an important mediating role in the process of new quality productivity affecting TFP, thus validating Hypothesis H2. Mediating variables of digital transformation. Existing literature mainly uses word frequency methods to measure it, counting keywords related to digitalization in corporate annual reports and aggregating them to assess the extent of digital transformation. Drawing on the research of Li and Tian (Li and Tian, 2023), text analysis is used to identify and count keywords related to digital transformation within listed companies’ annual reports. This yields a “textual intensity” measure for corporate digital transformation. The logarithm of the annual report’s digital keyword frequency plus one serves as a proxy variable to construct a corporate digital transformation index, acting as the mediating variable in the mechanism analysis. The specific test results are shown in columns (4) and (5) of Table 10. Column (4) indicates that the coefficient for new quality productivity on digital transformation is significantly positive at the 1% level, suggesting that new quality productivity can significantly drive the digital transformation of manufacturing enterprises. Column (5) reveals that after including the digital transformation indicator, the coefficient for new quality productivity on total factor productivity is 0.0286, significantly positive at the 1% level, with a Bootstrap 95% confidence interval of [0.0048–0.0053]. This confirms the significant mediating effect of digital transformation, showing that new quality productivity increases TFP levels by promoting digital transformation within companies. Hypothesis H3 is therefore supported.

Table 10. Test results for the mediating effect mechanism.

*, , * denote significance at the 1, 5, and 10% levels, respectively; figures in parentheses are t-statistics based on cluster-robust standard errors.

The mediation effect of digital transformation stands at 0.00339 (10.6% of the total effect). While numerically modest compared to technological innovation (24.5%), it underscores that digital infrastructure upgrades represent an incremental structural shift rather than an exhaustive driver of firm TFP. This highlights that manufacturing firms are in a critical transition state of capturing digital dividends, where structural adjustments yield narrower immediate gains than pure technology shocks.

5 Discussion

5.1 Alignment with and extension of circular economy and innovation literature

Our empirical findings provide critical theoretical and practical insights that bridge the gap between emerging macroeconomic paradigms and the circular economy (CE) literature (Ghisellini et al., 2016; Kirchherr et al., 2017). While foundational literature conceptualizes the circular economy primarily through resource closed-loops and waste minimization frameworks, recent scholarship stresses that transitioning from linear ‘take-make-dispose’ models toward systemic circularity fundamentally hinges on micro-level efficiency breakthroughs and technological modernization (Rejeb et al., 2026; Chauhan et al., 2022). Our baseline result—confirming that New Quality Productivity (NQP) significantly enhances corporate TFP—validates that NQP provides the vital operational efficiency needed to make circular industrial practices economically feasible rather than cost-prohibitive (D'Amato and Korhonen, 2021).

Furthermore, the validation of our transmission mechanisms—specifically technological innovation (H2) and digital transformation (H3)—directly enriches the understanding of the enablers of circular manufacturing systems: (1) Digital Transformation as a Circular Enabler (Twin Transition): Recent circular economy research highlights the “twin transition” (digital and green), where digital architectures serve as the backbone for circular supply chains (Kristoffersen et al., 2020; Otasowie et al., 2026). Our finding that digital transformation mediates the NQP–TFP nexus reinforces evidence that smart data systems, Industrial IoT, and real-time tracking reduce informational asymmetries across product lifecycles, optimize asset utilization, and streamline closed-loop material flows (Kristoffersen et al., 2020; Schöggl et al., 2024). (2) Technological Innovation and Eco-Efficiency: In line with contemporary eco-innovation literature, our empirical confirmation of the technological innovation channel demonstrates that NQP equips manufacturing firms to shift the production-possibility frontier away from virgin-material dependency toward clean production, remanufacturing, and secondary resource integration (Chauhan et al., 2022; Mora-Contreras et al., 2025). (3) Contextualizing Sectoral Divergence: Our industry heterogeneity analysis also contributes to circular transition debates. The negative friction observed in energy-intensive industries (petroleum and coal processing) aligns with recent observations that traditional extractive sectors face substantial structural lock-in and high switching costs during early circular transformations, requiring targeted policy incentives to reconcile environmental compliance with productivity gains (Rejeb et al., 2026; Fu and Zailani, 2025).

5.2 Sectoral and regional heterogeneity: contextual rigidities

Contrary to generalized assumptions that green and digital productivity transitions generate uniform benefits across all economic sectors, our subsectoral heterogeneity findings reveal crucial friction points. The observed negative impact of NQP on TFP within energy-intensive industries (Industry 5: petroleum, coal, and fuel processing) underscores that capital-intensive sectors face high short-term structural costs. For these firms, compliance expenses, heavy fixed-asset dependencies, and technological switching costs temporarily outweigh immediate productivity gains (Xiaodong et al., 2022).

Geographically, while the positive effects of NQP on TFP are balanced across the Eastern, Central, and Western regions, the non-significant effect observed in the Northeast region highlights how localized factors—such as institutional rigidity, historical reliance on heavy industrial legacy, and talent outflows—can suppress the enabling capacity of NQP (Su and Sun, 2024; Xiaodong et al., 2022). This confirms that macro-level productivity policies must be tailored to regional absorptive capacities rather than applied as uniform mandates.

5.3 Limitations and future research directions

Despite the robust methodology and comprehensive dataset spanning 2015 to 2024, this study has several limitations that open avenues for future research. First, our sample focuses exclusively on Chinese A-share listed manufacturing companies. While listed firms represent the backbone of China’s industrial economy, future studies could extend the empirical framework to small and medium-sized enterprises (SMEs) and unlisted private firms to test broader market generalization. Second, due to data accessibility, our measure of corporate-level NQP relies on a multi-dimensional composite index constructed via accounting metrics; future work could incorporate granular microdata such as industrial patent quality scores, real-time IoT deployment rates, or carbon emissions reductions. Finally, while this study establishes technological and digital transmission channels, future research could explore additional institutional mechanisms, such as corporate governance structures, green finance accessibility, and environmental policy intensity.

6 Conclusion

This study systematically investigates the direct impact and transmission mechanisms of new quality productivity on firm-level total factor productivity across Chinese A-share listed manufacturing enterprises over the period 2015 to 2024. Utilizing a two-way fixed effects estimation strategy alongside comprehensive endogeneity and robustness checks, the empirical findings demonstrate that new quality productivity significantly elevates corporate total factor productivity. Mechanism analyses reveal that this productivity-enhancing effect operates primarily through two critical operational pathways: fostering technological innovation and accelerating digital transformation. Heterogeneity evaluations further indicate that while NQP exerts a balanced, positive influence across the Eastern, Central, and Western regions of China, its effect remains statistically insignificant in the Northeastern region due to localized industrial structural rigidities. Across manufacturing subsectors, NQP drives substantial productivity gains in high-tech, equipment manufacturing, and traditional process industries undergoing technological modernization, whereas energy-intensive subsectors experience short-term productivity frictions during their structural transitions.

These findings carry actionable implications for industrial stakeholders, enterprise executives, and policymakers aiming to foster high-quality economic development. Policymakers should design targeted, industry-specific strategies that move beyond generalized investment incentives. For high-tech and equipment manufacturing sectors, policy frameworks should prioritize building collaborative innovation ecosystems, encouraging joint research consortia between enterprises and research institutes, and facilitating the commercialization of disruptive technologies. For traditional and energy-intensive industries, government support should center on providing structured digital transformation guidance, specialized credit facilities, and fiscal incentives to alleviate short-term transition costs and lower technical barriers. Regionally, tailored interventions are required to address local disparities; central and western provinces should continue enhancing digital infrastructure to capture industrial relocation spillovers, while northeastern rust belt regions require structural institutional reforms, business environment optimization, and targeted talent attraction initiatives to unlock the transformative potential of new quality productivity.

For corporate managers, these insights highlight the strategic necessity of deeply integrating digital architectures with core production workflows. Enterprise leadership should treat digital transformation not merely as an IT upgrade but as a foundational capability that optimizes resource allocation, minimizes operational idle times, and accelerates product innovation cycles. By aligning internal capital allocation with sustainable, technology-driven practices, manufacturing firms can build long-term competitive advantages, decouple industrial growth from excessive resource consumption, and effectively transition toward high-efficiency, sustainable operations. Future research should expand upon these insights by investigating the long-term interaction between new quality productivity and green supply chain resilience across diverse international settings.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Konstantinos P. Tsagarakis, Technical University of Crete, Greece

Reviewed by: Zhaocheng Xu, Shanghai Dianji University, China

Maoyu Gong, Shandong University of Finance and Economics, China