1. Introduction

In the context of intensifying global competition and rapid technological advancements, the conventional large-scale, low-cost manufacturing model, which has been a fundamental aspect of traditional manufacturing, is facing challenges in adapting to the dynamic requirements of modern markets. The transformation of the manufacturing industry represents a central focus of current industrial development and is of utmost importance. It serves as a crucial catalyst for refining the global economic structure and represents a strategic pathway towards achieving sustainable development objectives and enhancing national competitiveness. As total factor productivity (TFP) serves as a vital metric for assessing the advancement of enterprises, enhancing TFP within manufacturing enterprises, and consequently promoting high-quality growth in the manufacturing industry has emerged as a critical issue for both industry practitioners and academic researchers. The accelerated pace of technological advancement and industrial transformation has created unprecedented opportunities for innovation within the manufacturing industry (Liu et al., 2020) [1]. Among these digital technologies, industrial robots are particularly noteworthy as innovative catalysts for industrial progress. For example, Tesla employs a highly automated production line that uses advanced robotic technology for body welding and assembly. These devices are able to complete complex welding tasks in a very short time. This not only improves production efficiency, but also significantly reduces labor costs and the incidence of human error, making the production capacity of its electric vehicles significantly increased. Therefore, conducting a comprehensive analysis of the impact of industrial robot application on the TFP of manufacturing enterprises, and the underlying mechanisms, holds considerable practical significance for boosting productivity, accelerating industrial transformation, and promoting high-quality development within the manufacturing industry.

A substantial body of research has been conducted to examine the economic impact of industrial robots. The principal insights derived from this research can be summarized as follows: Firstly, industrial robots, representing technological progress, have the capacity to displace low-skilled labor more significantly, generating a “substitution effect” (DeCanio, 2016; Acemoglu and Restrepo, 2020) [2,3]. This substitution effect will lead to a decrease in labor demand, resulting in widespread unemployment in the market (Autor et al., 2003; Frey and Osborne, 2017; Agrawal et al., 2019) [4–6]. Secondly, the application of industrial robots can lead to the creation of new employment opportunities, thereby stimulating enterprises to increase their demand for highly skilled personnel. This process optimizes the workforce structure and enhances enterprises’ innovation capacity, a phenomenon known as the “creation effect” (Graetz and Michaels, 2018; Balsmeier and Woerter, 2019; Koch et al., 2021) [7–9]. Lastly, the application of industrial robots exerts a diverse influence on the scale and composition of an enterprise’s workforce, demonstrating distinct effects across diverse types of enterprises. While the existing literature primarily concentrates on the macro-level labor market implications of industrial robots, there is insufficient examination of their impact on the total factor productivity of micro-enterprises. Duan et al. (2023) [10] examines how industrial robots affect enterprises’ TFP through these two aforementioned effects. However, further investigation is needed to determine whether additional mechanisms contribute to the consequences of applying industrial robots in micro-enterprises. This study aims to investigate the economic implications of industrial robots at the micro-enterprise level and provide compelling evidence for evaluating their function.

The contemporary theory of economic growth identifies the primary drivers for sustained economic growth as encompassing the level of technological progress, accumulation of human capital, and material production factors (Romer, 1990) [11]. High-quality development represents an economic model that efficiently provides society with superior products and services, characterizing it as a form of “intensive growth” achieved by enhancing the quality and efficiency of factor utilization (Brandt et al., 2012) [12]. The concept of TFP at the enterprise level measures the maximized output generated from various factor inputs such as capital, labor, technology, management, organizational structure, and other inputs. It serves as an indicator of efficiency in converting these factor inputs into final outputs (Hsieh and Klenow, 2009) [13].

Total factor productivity (TFP), which reflects the comprehensive productivity of all factors within an enterprise, is widely acknowledged by scholars as a pivotal indicator of high-quality growth. Prior research has provided a robust theoretical foundation and practical pathways for enhancing enterprises’ TFP. The existing literature has comprehensively analyzed the pathways to enhance TFP, with a primary focus on financial development (Sturm and Williams, 2004) [14], technological innovation (Baumann and Kritikos, 2016; Anzoategui et al., 2019) [15,16], and the labor market (Bennedsen et al., 2019) [17]. However, there is a significant deficiency in the existing literature regarding thorough analysis of the influence mechanisms exerted by artificial intelligence and automation technologies, particularly industrial robots, on enterprises’ TFP. As a paradigmatic manifestation of digital technological progress, industrial robots have the capacity to assist enterprises in reducing costs and increasing efficiency. Therefore, it is paramount to investigate the impact of industrial robot application on enterprises’ TFP. The objective of this study is to enhance our understanding of digital technology on enterprises’ TFP. Furthermore, it offers valuable insights for facilitating industrial transformation and fostering high-quality progression.

The Fourth Industrial Revolution has established digital technology as a crucial instrument for countries across the globe to drive the transformation and modernization of their manufacturing industry. China is currently at a critical juncture of economic transformation and upgrading. This presents both challenges and opportunities in terms of industrial structure adjustment and innovation-driven development. Currently, China is confronted with the challenge of a gradually diminishing demographic dividend, characterized by a decline in employment opportunities for low-skilled and repetitive labor. The resulting labor shortage has led to a notable increase in the cost of human capital within the manufacturing industry. Consequently, promoting intelligent and digital technologies has emerged as a novel trend in the context of manufacturing transformation. As the foremost manufacturing country globally, China boasts vast manufacturing industries that provide rich data resources for the application of industrial robots. In this context, examining the role of technological progress in catalyzing economic transformation and upgrading becomes significantly important.

This study utilizes panel data derived from Chinese A-share listed manufacturing enterprises, covering the period from 2007 to 2022, with the aim of achieving a comprehensive understanding of the implications of industrial robot application on enterprises’ TFP. The study examines both the relationship between industrial robot application and manufacturing enterprises’ TFP and the underlying mechanisms. The findings indicate that industrial robot application has a positive impact on enterprises’ TFP. This conclusion remains consistent even after conducting a series of robustness tests. Furthermore, an in-depth examination of the mechanisms suggests that industrial robot application can reshape enterprises’ production and operational structures, thereby attracting highly skilled personnel who are proficient in navigating this new technology. Consequently, there is an increase in human capital levels and its TFP, providing empirical support for the concept of a “talent aggregation effect”. Additionally, industrial robot application has been shown to enhance TFP by facilitating innovation capabilities, thus providing empirical support for the “innovation effect”. Moreover, heterogeneity analysis reveals considerable variation in the extent to which industrial robots enhance TFP across different enterprises. Notably, the impact of industrial robot application is more pronounced in high-labor productivity enterprises as well as state-owned and high-tech enterprises.

The findings of this study contribute to the existing body of research in the following ways:

Firstly, this research presents a comprehensive analysis of the impact of industrial robots on micro-enterprises, with a particular focus on total factor productivity. This endeavor builds upon and extends the existing academic literature on the economic implications of industrial robot application. Previous studies have predominantly focused on examining the “substitution effect” and “creation effect” of industrial robots on the macroeconomic labor market and employment rates. However, little attention has been paid to the impact of these robots on the decision-making processes of micro-enterprises, particularly in relation to their total factor productivity. The assessment of total factor productivity, defined as the efficiency with which an enterprise transforms inputs into outputs, is paramount for optimizing resource allocation within firms and fostering sustainable high-quality economic growth. Based on rigorous investigation, this study reveals that industrial robot application at the micro-enterprise level yields beneficial outcomes. This finding not only enhances research on the economic consequences of industrial robots but also provides compelling micro-level evidence for the effectiveness of digital technology in boosting enterprise productivity and fostering high-quality development. Furthermore, it offers invaluable insights for government policymakers in formulating strategies to accelerate enterprises’ transformation and modernization.

Secondly, the present study examines the impact of industrial robots on total factor productivity in enterprises within the context of digital technology. This contributes to and develops the existing analysis of factors influencing enterprise TFP. Prior studies have predominantly focused on identifying determinants of enterprises’ TFP, including financial growth (Sturm and Williams, 2004) [14], technological innovation (Baumann and Kritikos, 2016; Anzoategui et al. 2019) [15,16], and labor market (Bennedsen et al. 2019) [17]. However, the complex mechanisms through which industrial robots affect enterprises’ TFP have been largely overlooked in existing literature. The findings suggest that industrial robots have the potential to enhance the structure of enterprises’ human capital, thereby creating a positive aggregation effect regarding talent. Moreover, as a significant catalyst for technological advancement, industrial robots serve as a crucial driver of enterprises’ innovation capabilities. This research contributes to understanding enterprises’ TFP from a digital technology perspective and provides actionable insights for manufacturing industry to develop strategies for substituting human labor with machine-based processes and enhancing their competitive advantage in the market.

Thirdly, this study conducts a comprehensive examination of the mechanisms through which industrial robots affect TFP. The findings of our research indicate that the application of industrial robots has a positive impact on enterprises’ TFP primarily by promoting talent aggregation and enhancing innovation capabilities. Although some researchers have examined the influence of industrial robot application on TFP (Duan et al., 2023; Xu et al., 2024; Qin et al., 2025) [10,18,19], there is a notable absence of studies investigating these impacts from the perspective of technical and labor production factors. The objective of this research is to address this knowledge gap by investigating the fundamental mechanisms through which industrial robot application affects TFP, thereby making a significant contribution to the existing academic discourse. The aim is to clarify the complex relationship between industrial robots and TFP, thus supporting both the “talent aggregation effect” and “innovation effect” associated with industrial robots.

2. Hypothesis development

The improvement of TFP is crucial for enterprises seeking sustainable development and market competitiveness, as it drives economic growth and facilitates sustained expansion (Solow, 1956) [20]. Enhancing TFP is not only essential for microeconomic entities to refine resource allocation and improve operational efficiency but also a catalyst for achieving high-quality growth and transformative upgrading at the macroeconomic level. TFP represents a comprehensive metric that evaluates an enterprise’s maximum output with a given set of inputs, leading to improvements in both production efficiency and economic returns. Increasing the total factor productivity of enterprises effectively promotes industrial structure optimization, economic transformation, and upgrading. In the context of rapid globalization and the growing role of information technology, enhancing the total factor productivity of enterprises, as fundamental units within national economies, holds great significance in improving national competitiveness and achieving sustainable economic development.

The relentless advancement of big data, artificial intelligence, the Internet of Things, and associated technologies has resulted in a profound integration between digital technology and the real economy. Consequently, industrial robots are increasingly replacing repetitive traditional manual work and assuming more responsibilities, thereby marking a transformation that is set to enhance enterprises’ TFP. Firstly, the application of industrial robots has prompted a transformation in the workforce structure, shifting from an operational and skill-centric model to a knowledge-centric one (Du and Lin, 2022) [21]. This transition inevitably leads to an increase in the skill requirements for the workforce, resulting not only in elevated demand for highly skilled personnel but also in a greater proportion of such skilled workers within enterprises. Subsequently, this optimizes the composition of the workforce and enhances the quality of human capital. Secondly, the integration of cutting-edge technologies into industrial robotics stimulates innovation in technology research and development, production processes, and product-related services. This integration results in more efficient allocation of resources and enhanced production efficiency (Makridakis, 2017; Graetz and Michaels, 2018) [22,23]. Moreover, their intelligent functionalities facilitate the modernization of enterprises’ organizational frameworks and management methodologies, thereby creating an environment conducive to sustained innovation. The aggregation of talent associated with industrial robots increases the density of knowledge networks and facilitates knowledge transfer, fostering the growth of intra-group innovation networks. Ultimately, this drives continuous innovation and transformation leading to a sustained enhancement of total factor productivity (Che and Zhang, 2018) [24]. Based on the analysis above, this paper formulates the following research hypothesis:

Hypothesis 1: The industrial robot application can enhance the total factor productivity of enterprises.

A theoretical examination has been conducted to determine the direct implications of industrial robot application on enterprises’ total factor productivity. The following analysis will examine in greater detail the mechanisms underlying this influence, with a particular focus on two key areas: talent aggregation and innovation capacity.

The implementation of industrial robots has had a significant impact on the production and operational structure of enterprises, as evidenced by substantial restructuring. This transformation has facilitated the recruitment of highly skilled employees who are aligned with the changes brought about by industrial robot application. Consequently, enterprises have been able to enhance their human capital, resulting in an increase in their TFP.

Firstly, the automation and intelligence inherent in industrial robots have facilitated the refinement of internal talent structures within enterprises, enabling the convergence of highly skilled individuals in pivotal roles and creating a talent aggregation effect. The application of industrial robots has accelerated technological advancements, optimized resource allocation, and alleviated the burden on employees engaged in arduous repetitive tasks (Gaggl and Wright, 2017; Graetz and Micheals, 2018) [23,25]. Consequently, these employees can now devote more time to tasks that require advanced cognitive and creative abilities (Goos et al., 2009; Michaels et al., 2014) [26,27]. Exceptionally skilled individuals are crucial for innovation as they possess exceptional professional skills, robust innovative capabilities, and adept knowledge application abilities. Exceptionally skilled individuals are capable of quickly comprehending and effectively utilizing advanced technologies, thereby amplifying the spillover effects of technology within enterprises (Jung and Lim, 2020) [28]. The rise of such individuals reinforces the labor force structure, elevates human capital levels, and injects new vitality into innovation and development. As a result, this leads to increased innovation and promotes technological progress (Aghion et al., 2015; Wang et al., 2020) [29,30]. Furthermore, the application of industrial robots can facilitate the creation of employment opportunities with distinctive advantages, thereby generating new forms of attractive employment and work modalities for highly educated professionals. For example, the use of industrial robots may lead to an increased demand for innovative roles such as robot engineers. This prompts enterprises to enhance their recruitment and training activities in relevant domains, thereby improving the caliber and competitiveness of their human resources. The appealing remuneration packages and career progression opportunities brought by industrial robots can attract and retain a larger number of exceptional talents, thus establishing a virtuous cycle. Based on the above analysis, this paper formulates the following research hypothesis:

Hypothesis 2: The industrial robot application enhances total factor productivity by facilitating the aggregation of talent.

Technological advancement is the key to enhancing the total factor productivity of enterprises, as it helps refine the distribution of production elements and upgrade the technological proficiency of enterprises, thereby having a favorable impact on TFP enhancement. Among the key factors influencing the growth of innovation capacity, industrial robot application plays a significant role and has a considerable impact on the enhancement of TFP.

Firstly, the application of industrial robots serves to stimulate innovation within the sphere of production technologies. Enterprises persistently focus on researching, developing, and deploying industrial robots in order to explore new technologies, processes, and methodologies. This not only enhances the accuracy and efficiency of product manufacturing but also fosters the infiltration and integration of new technologies across various stages including product design, production workflows, and quality assurance. As a result, technological innovation becomes more dynamic. The synergistic application of cutting-edge manufacturing technologies with industrial robots has been shown to enhance production precision, reduce energy consumption, and optimize production workflows thereby propelling continual advancements in production technologies (Loebbecke and Picot, 2015; Acemoglu and Restrepo, 2020) [3,31]. This technological innovation serves to enhance enterprises’ production efficiency and afford them heightened competitive advantages in the marketplace thereby contributing to the enhancement of TFP (Balsmeier and Woerter, 2019) [8]. Secondly, the application of industrial robots serves to stimulate innovation in enterprises’ management practices. In order to optimize the efficacy of industrial robots, it is essential for enterprises to continuously refine their production management processes, increase the level of digitization, and strengthen interdepartmental collaboration. By enhancing knowledge accumulation and integrating both internal and external knowledge resources, enterprises ultimately achieve the goals of optimizing production workflows, improving product quality, and increasing innovation capabilities (Huang et al., 2017) [32]. Concurrently, the application of industrial robots facilitates a reduction in resource and energy consumption as well as production costs (Impullitti et al., 2022) [33], an increase in output per unit of input, and consequently enhances TFP. Based on the above analysis, this paper formulates the following research hypothesis:

Hypothesis 3: The industrial robot application enhances total factor productivity by improving their innovative capabilities.

3. Methodology and data

3.2. Variable definition

3.2.1. Dependent variable.

The dependent variable under investigation is the Total Factor Productivity (TFP) of enterprises. This metric not only embodies the efficiency of their input-output operations but also serves as an indication of their innovation-oriented capabilities and growth potential (Brandt et al., 2012) [12]. Following the methodologies of Levinsohn and Petrin (2003) [34], this paper employs two methods to measure the TFP of enterprises: TFP calculated using the Levinsohn-Petrin method (TFP_LP) and TFP calculated using the fixed effects method (TFP_FE). The TFP_LP approach controls for potential endogeneity between input choices and unobserved productivity by using intermediate input demand (e.g., material costs) as a proxy. And, the TFP_FE approach estimates productivity as the residual from a Cobb-Douglas production function with firm-level fixed effects, which absorbs time-invariant unobservable heterogeneity across firms.

3.2.2. Independent variable.

The independent variable is the application of industrial robots (ROBOT). Following the research of Acemoglu and Restrepo (2020) [3], the level of industrial robot application is measured by the penetration rate of industrial robots at the enterprise level. The specific calculation process is as follows:

First, the industry-level industrial robot penetration was calculated using formula (1):

(1)

Where signifies the installation of industrial robots in Chinese industrial sector during year . represents the number of workers in Chinese industry during the base period of 2007. The year 2008 was the base year t. denotes the penetration of industrial robots within the Chinese industrial sector in year .

Second, firm-level industrial robotics penetration is calculated using Equation (2):

(2)

Where denotes the industrial robot penetration of firm in industrial sector j within the manufacturing industry in year . denotes asset expenditures, such as fixed and intangible assets, of firm in industry sector within the manufacturing industry in year . denotes the median asset expenditures of all firms in the manufacturing industry in year .

3.2.3. Control variables.

Drawing on previous research (Acemoglu and Restrepo, 2020; Liang et al., 2023) [3,35], this paper has selected a set of variables that reflect corporate financial and governance conditions as control variables. These include company size (SIZE), return on assets (ROA), Tobin’s Q (TOBINQ), firm age (AGE), cash flow ratio (CASH), gross profit margin (GPM), management expense ratio (MFEE), tangible asset ratio (TANG), as well as industry and year variables. The definitions of each variable are presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Variable definitions.

https://doi.org/10.1371/journal.pone.0331378.t001

3.3. Model specification

3.3.1. Baseline model.

In order to verify the effect of industrial robots on the total factor productivity of enterprises, refers to the research of Liang et al.(2023) [35], we construct the following baseline regression model (3):

(3)

where is the dependent variable, denoting total factor productivity of enterprises. is the independent variable, denoting industrial robot application. represents all control variables and industry-year fixed effects. represents the random disturbance term. All regression analyses conducted in this paper are clustered at the firm level to mitigate potential issues of heteroskedasticity and autocorrelation. The primary focus is on examining the regression coefficient associated with the independent variable, industrial robot application. If this coefficient is found to be significantly greater than 0, it suggests that the impact of industrial robot application on enterprises’ total factor productivity is significantly positive. This implies that an increase in the extent of industrial robot application enhances the total factor productivity of enterprises, thereby confirming Hypothesis 1.

4. Empirical analysis and results

4.1. Descriptive statistics and correlation analysis

The descriptive statistics for the primary variables under consideration are presented in Table 2. It is evident from the table that there is a significant level of industrial robot penetration within enterprises, with an average (median) prevalence of 1.8491 (1.5393). Therefore, it is important to examine the impact of industrial robot penetration on total factor productivity (TFP) in enterprises. The mean (median) TFP, calculated using the LP method, is 8.9277 (8.8361), while the corresponding values obtained using the fixed-effects method are 11.3640 (11.2375). Additionally, both methods indicate variability in productivity across different enterprises with regards to TFP – 0.9969 for LP and 1.2298 for fixed-effects method.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Descriptive statistics.

https://doi.org/10.1371/journal.pone.0331378.t002

Table 3 presents the correlation coefficients between variables. Spearman correlation coefficients are located above the principal diagonal in the correlation matrix, while Pearson correlation coefficients are situated below it. According to the Pearson correlation coefficients, there is a significant positive correlation between industrial robot penetration (ROBOT) and total factor productivity (TFP). This finding suggests that applying industrial robots enhances enterprises’ total factor productivity to some extent, providing initial support for Hypothesis H1 proposed in this research. The interpretation of Spearman’s correlation coefficients aligns with the above and will not be reiterated here.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Correlation coefficient.

https://doi.org/10.1371/journal.pone.0331378.t003

4.2. Baseline regressions

Table 4 presents the results of the baseline regression analysis, which investigates the correlation between industrial robot application and TFP. Specifically, columns (1) to (3) showcase the estimated TFP results obtained using the Levinsohn-Petrin (LP) method, while columns (4) to (6) display the estimated TFP outcomes derived from the fixed effects approach. In particular, columns (1) and (4) present regression results without any control variables and without considering fixed effects for industry and year. Columns (2) and (5) reveal regression results that exclude control variables but incorporate fixed effects for industry and year. Columns (3) and (6) present regression results that include control variables and account for both industry and year fixed effects. Across all columns from (1) to (6), consistently demonstrate that the regression coefficients for industrial robots are significantly positive at a 1% level, suggesting that their application notably enhances firm-level TFP thus supporting Research Hypothesis 1. Furthermore, computed VIF values for all models are below 10 indicating no issues with multicollinearity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Baseline regressions.

https://doi.org/10.1371/journal.pone.0331378.t004

4.3. Robustness test

To further enhance the robustness of the findings, this section conducts a series of robustness tests, including replacing both the independent and dependent variables, changing the regression model, conducting persistence tests, eliminating the influence of other factors, and performing IV estimation.

4.3.1. Change variables.

Firstly, we conduct a validation test by substituting the dependent variable. Specifically, we choose to use Total Factor Productivity estimated through the Ordinary Least Squares method, referred to as TFP_OLS, instead of the TFP values originally calculated using the LP and fixed-effects methods in our baseline regression model. The results of this verification exercise are presented in column (1) of Table 5, indicating a significant positive relationship between the regression coefficient and industrial robot application. Secondly, we perform a second test involving the substitution of the independent variable. In this case, we use the count of operational industrial robots (ROBOT2) as an alternative measure to replace the penetration rate of industrial robots used in the baseline analysis. The outcomes of this test are shown in columns (2) and (3) of Table 5. Examination of these columns reveals that both estimated coefficients associated with industrial robot application are significantly positive.The above empirical findings confirm that using substitute variables yields results that align substantially with those obtained from the baseline regression, thereby reinforcing the robustness of conclusions presented in this study. Furthermore, we replace the original variables with the number of changes in the robots (ROBOT3) and the research results still support our conclusion in columns (4) and (5) of Table 5.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Change variables.

https://doi.org/10.1371/journal.pone.0331378.t005

4.3.2. Change models.

To ensure the reliability of our research findings, we conducted additional analyses using various specifications of the fixed-effects model. The results of these analyses are presented in columns (1) to (4) of Table 6. Specifically, columns (1) and (2) display the results that incorporate fixed effects for both companies and years, while columns (3) and (4) show the outcomes with fixed effects for industries, years, and provinces. The findings reveal that the coefficient estimates associated with industrial robot application exhibit significant positivity, consistent with the baseline regression results. This consistency underscores the robustness of our research conclusions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Change models.

https://doi.org/10.1371/journal.pone.0331378.t006

4.3.3. Cluster at the city level.

The standard errors are currently clustered at the firm level. We also try clustering at the city level to account for autocorrelation in error terms among firms in the same city. The results of these analyses are presented in columns (1) to (6) of Table 7. The findings reveal that the coefficient estimates associated with industrial robot application exhibit significant positivity, consistent with the baseline regression results.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. Cluster at the city level.

https://doi.org/10.1371/journal.pone.0331378.t007

4.3.4. Persistent test.

We investigate the long-term influence of industrial robots on total factor productivity by lagging the dependent variable by one and two time periods. The resulting estimation results are displayed in Table 8. Specifically, columns (1) and (2) present the estimates for the dependent variable one period ahead, while columns (3) and (4) illustrate the findings two periods ahead. The outcomes reveal that all coefficient estimates associated with industrial robot application are significantly positive. This provides additional evidence that the beneficial impact of industrial robot application on enterprises’ TFP persists over time, supporting the conclusion derived from the baseline regression analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. Persistent test.

https://doi.org/10.1371/journal.pone.0331378.t008

4.3.5. Eliminate the influence of other factors.

To ensure the rigor and scientific validity of our regression analysis, this study employs two distinct methods to mitigate the confounding effects of extraneous factors. The detailed results are outlined in Table 9. Firstly, provinces are considered pivotal conduits for macroeconomic policy enforcement, resource distribution, and localized cultural norms; their inherent disparities can exert both direct and indirect influences on the decision-making processes of microeconomic units. To counteract any potential disruption from regional factors on the interplay between core economic variables, we incorporate three provincial control variables into our analysis: the marketization index (FINDEVE), per capita fixed asset investment (INVP), and supplier concentration (CTION). The resulting regression outputs are presented in columns (1) and (2). Importantly, even after considering these control variables, the regression results align with our primary findings, indicating that regional factors do not significantly alter the conclusions of this investigation. Secondly, we exclude data from the financial crisis period to eliminate its potential interference with the regression outcomes. Specifically, we omit samples from 2007 and 2008, and provide detailed regression results subsequent to this exclusion in columns (3) and (4). Our research conclusions remain consistent with earlier findings, confirming the robustness of our core regression conclusions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 9. Eliminate the influence of other factors.

https://doi.org/10.1371/journal.pone.0331378.t009

4.3.6. IV estimation.

Industrial robot application may be affected by omitted variables or reverse causality, which raises concerns about endogeneity. For instance, enterprise with higher total factor productivity (TFP) often incorporate industrial robots more extensively into their production and operational procedures, potentially leading to a reverse causality issue. To address this concern, we adopt the methodology used by Acemoglu and Restrepo (2020) [3], utilizing the penetration rate of industrial robots in the United States as an instrumental variable and employing a two-stage least squares (2SLS) estimation approach. The rationale for selecting this variable is twofold: Firstly, during the study period, both the United States and China demonstrated comparable levels of progress and application in industrial robots while following similar developmental paths. Secondly, as a global leader in robotics embodying industry-wide technological trends, the United States’ influence on enterprises’ TFP is expected to reflect solely exogenous technological progress and spillover effects. Consequently, the impact of the U.S. industrial robot penetration rate on Chinese enterprises’ TFP primarily mirrors analogous industries’ technological characteristics, meeting the relevance requirement for an instrumental variable. Moreover, it remains largely independent of other domestic factors that influence industrial robot application in China while fulfilling the exogeneity requirement and thus mitigating endogeneity concerns. The empirical results are summarized in Table 10. Specifically, column (1) presents the first-stage estimation results revealing a notably positive regression coefficient between the instrumental variable (USROBOT_IV) and the independent variable (ROBOT), which aligns with the relevance condition. Columns (2) and (3), respectively utilizing Levinsohn-Petrin(LP) method and fixed-effects method, present second-stage TFP estimation results. Both sets of findings indicate significantly positive regression coefficients for industrial robot application consistent with baseline regression results, confirming both robustness of our conclusions as well as alleviation of endogeneity concerns. Furthermore, we continued to use the European robots as the instrumental variable, and the results in Table 11 are consistent with the previous conclusion, confirming the existing conclusion.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 10. The result of using American robots as IV.

https://doi.org/10.1371/journal.pone.0331378.t010

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 11. The result of using European robots as IV.

https://doi.org/10.1371/journal.pone.0331378.t011

5. Mechanism tests

To achieve a deeper understanding of the mechanism underlying the impact of industrial robots on enterprises’ TFP, this section employs a mediation effect model. This model is used to explore the various channels through which industrial robots affect TFP, with a specific focus on dimensions such as talent aggregation and innovation capacity. The aim is to clarify the inherent connections within this relationship and provide empirical support for promoting enterprises’ transformation and achieving high-quality development.

5.1. Talent aggregation

The primary manifestation of corporate human capital lies in employees who possess high levels of educational achievement, technical expertise, and skill proficiency. For clarity, highly educated employees (HR1) are defined as the ratio of those holding bachelor’s degrees or higher within the overall workforce; highly technical employees (HR2) represent the proportion of technical personnel in the total workforce; and highly skilled employees (HR3) signify the fraction of research and development (R&D) staff within the entire workforce. Table 12 outlines the results of the mechanism test conducted to assess talent aggregation. Specifically, columns (1) to (3) examine the influence of industrial robot application on corporate human capital. The findings reveal that the coefficients estimating the effect of applying industrial robots on human capital show significant positivity, indicating that applying industrial robots significantly enhances the quality of corporate talent. These findings suggest that applying industrial robots has the potential to elevate the level of corporate human capital and subsequently boost an enterprise’s total factor productivity. This evidence supports the “talent aggregation effect” associated with industrial robots and confirms Hypothesis 2 proposed in this research.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 12. Mechanism test results.

https://doi.org/10.1371/journal.pone.0331378.t012

6. Heterogeneity analysis

To achieve a comprehensive understanding of how the application of industrial robots affects total factor productivity, this section undertakes an exhaustive examination of the various impacts exerted by such application on TFP. The objective is to provide stronger empirical evidence supporting the overall enhancement of TFP.

6.1. The heterogeneity of production efficiency

The interplay between production efficiency and the application of industrial robots in relation to TFP is the focus of this section. In scenarios where enterprises already exhibit high levels of production efficiency, the application of industrial robots emerges as a pivotal catalyst for further augmenting their TFP. Firstly, considering technological compatibility and synergistic advantages, enterprises characterized by high production efficiency typically possess well-established production technologies and streamlined process systems (Firth et al., 2015) [36]. By incorporating industrial robots, these enterprises can seamlessly blend and refine their existing production workflows, demonstrating heightened technological assimilation and innovation prowess. Leveraging the precision and efficiency of industrial robots, these enterprises can drastically reduce errors and waste during production, thereby optimizing resource allocation and enhancing TFP. Secondly, enterprises with superior production efficiency often excel in the efficient allocation of labor and capital. The application of industrial robots facilitates the substitution of low-skilled labor, prompting enterprises to expedite the upgrading of their human capital and foster an effective integration of highly skilled labor with advanced capital (Goos et al., 2014) [37]. This integration not only strengthens the enterprise’s innovative capacity but also sustains continuous enhancement in TFP. Consequently, this paper predicts that within high-efficiency enterprises, the application of industrial robots will have a more pronounced effect on TFP.

The methodology for calculating labor productivity (LP) involves subtracting non-operating income from operating income and then dividing the result by the total workforce, with data categorized based on the median. Regression results related to production efficiency are presented in Table 13, where columns (1) and (2) display estimated outcomes of TFP derived using the LP approach and calculated through the fixed-effects method, respectively. The research findings reveal that coefficients of the interaction term between industrial robot application and enterprises’ production efficiency (ROBOT×LP) are significantly positive at a level of 1%, indicating that within enterprises with high production efficiency, the impact of industrial robot application on stimulating TFP is more substantial, aligning with hypotheses posited in this study.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 13. Heterogeneity analysis results.

https://doi.org/10.1371/journal.pone.0331378.t013

7. Conclusion and policy implications

Supporting information