The Impact of Digital Talent Inflow on the Co-Agglomeration of the Digital Economy Industry and Manufacturing

Abstract

The co-agglomeration of the digital economy industry and manufacturing is significant for addressing issues such as being “large but not strong” and “comprehensive but not refined” in China’s manufacturing sector. This study uses 269 cities in China from 2006 to 2022 as the research sample, innovatively employing data from digital economy enterprises and manufacturing enterprises to measure industrial co-agglomeration, and comprehensively analyzes the mechanism of how the inflow of digital talents influences the co-agglomeration of the digital economy industry and manufacturing. The findings are as follows: (1) From 2006 to 2022, the inflow of digital talents and the level of co-agglomeration between the digital economy industry and manufacturing in Chinese cities have consistently risen, generally moving towards higher inflow and higher levels of co-agglomeration. However, the inflow of digital talent in the central and western regions is relatively low, with most cities still facing difficulties due to inadequate policy support and resource investment. Industrial co-agglomeration exhibits characteristics of “core–periphery”, “multi-core agglomeration”, and “gradient diffusion” coexisting. (2) The flow of digital talents can significantly promote the co-agglomeration of the digital economy industry and manufacturing, and this conclusion remains valid after robustness testing. The flow of digital talents drives the co-agglomeration of the digital economy industry and manufacturing by enhancing the level of digital technology innovation, promoting the spillover and flow of digital knowledge, increasing the entrepreneurial activity of urban digital economy enterprises, and upgrading industrial structures. Furthermore, digital economy policies play a regulatory role in this process. (3) The promotion effect of digital talent inflow is more pronounced in low- and mid-end manufacturing, high-grade cities, well-developed digital infrastructure, and non-resource-based cities. By contrast, this effect is relatively weaker in high-end manufacturing and low-grade cities. In cities with weak digital infrastructure and resource-based cities, this effect is not significant. (4) The inflow of digital talents and the co-agglomeration of digital economy industry and manufacturing have a significant promotion effect on cities with similar economic development levels and adjacent geographical locations, demonstrating a positive diffusion effect.

1. Introduction

In recent years, China’s manufacturing has emerged as a significant component of global manufacturing, demonstrating a development pattern that is the largest in scale, the most comprehensive in categories, and the most intact in systems worldwide, with robust international competitiveness. However, China’s manufacturing sector still faces numerous challenges, such as being “large but not strong”, “comprehensive but not refined”, having “excessive low-end supply” and “insufficient high-end supply”, a “weak industrial foundation”, and a “lack of international voice and influence”. To address these issues, the report of the 20th National Congress of the Communist Party of China explicitly proposes implementing the industrial foundation reconstruction project and the major technical equipment research project [1], supporting the development of specialized and innovative enterprises, and promoting the manufacturing to move towards high-end and intelligent directions. The rapid development of the digital economy industry has injected new impetus and vitality into the manufacturing, accelerating the deep integration between the digital economy and manufacturing.

It is noteworthy that the deep integration between the digital economy industry and manufacturing is not merely a simple application of digital technology but requires a high level of integration and optimal allocation of resources, technology, and markets across various industries, which is precisely the core content of industrial co-agglomeration [2]. Through industrial co-agglomeration, the digital economy industry and manufacturing can share the outcomes of digital technology innovation, thereby driving the transformation of China’s manufacturing towards a knowledge-intensive direction. In this transformation process, high-skilled labor, especially digital talent, plays a crucial role. Existing research has shown that high-skilled labor is a key factor in promoting industrial co-agglomeration, and its inflow can impact the scale of regional industrial co-agglomeration [2]. However, current academic research on the co-agglomeration of the digital economy industry and manufacturing is still insufficient, lacking deep and systematic exploration. What impact does the inflow of digital talent have on the co-agglomeration of the digital economy industry and manufacturing? How does the underlying mechanism of this impact operate? Is this impact constrained by other factors? These questions all require further investigation.

This study will delve into the three core issues mentioned above, aiming to uncover the impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing, as well as the underlying mechanisms behind it. Additionally, it will examine other factors that may either constrain or facilitate this impact. This study not only helps reveal how the digital economy industry and manufacturing can achieve co-agglomeration but also provides an in-depth analysis of the specific role of digital talent inflow in this process. It offers policymakers theoretical grounds for formulating industrial policies, optimizing talent cultivation and recruitment strategies, and promoting regional industrial development, thereby carrying significant theoretical and practical importance.

2. Literature Review

Currently, there are three main categories in the literature most relevant to this paper. The first is research on digital talent. This research primarily employs qualitative methods, often using literature reviews, case studies, and other approaches to explore the concept and role of digital talent [3,4,5] and to analyze the current status and existing issues of digital talent development in China [3]. Quantitative research has mainly investigated the role of digital talent, with studies indicating that digital talent can influence green technology innovation [6], individual innovation behavior [7], optimization of the manufacturing structure [8], and open innovation [9].

The second category is research on the digital economy industry and manufacturing agglomeration. There is a considerable amount of literature in this area, providing valuable insights for this study. Many scholars measure the scale of the digital economy industry in a region by calculating the industrial value-added, main business income [10,11], and number of employees in the digital economy and manufacturing [12,13,14,15,16,17]. They also use the location entropy index to assess the degree of agglomeration of the digital economy industry and manufacturing [10,11,12,13,14,15,16,17,18,19]. The location entropy index is one of the commonly used indicators for measuring manufacturing agglomeration and is widely applied, with fewer scholars adopting other measurement methods [20,21]. Research on influencing factors suggests that factors such as the digital economy [12], data element marketization [19], transportation infrastructure [20], government policies [22], and Marshall externalities [23] can impact the agglomeration of the digital economy industry and manufacturing. Research on effect mechanisms indicates that the agglomeration of the digital economy industry and manufacturing can influence regional innovation capacity [10], green development efficiency [14], high-quality economic development [16], manufacturing technology innovation [18], environmental pollution [13,15,21], carbon intensity [24], green total factor productivity [25], and enterprise digital transformation [26].

The third category is research on the measurement of industrial co-agglomeration. Currently, the primary methods include the E-G index proposed by Ellison et al. [27], the D-O index proposed by Duranton and Overman [28], the γ index proposed by Chen and Chen [29], and the Θ index proposed by Chen et al. [30]. The E-G index is based on industry-average employment data, but its application presupposes similar labor/capital ratios across target industries, a condition often difficult to satisfy in reality. The D-O index, on the other hand, requires precise spatial location data of enterprises and assumes continuous spatial distribution of manufacturers without regional marginal constraints. However, its practicality in China is limited, and there is currently no literature using this index to specifically measure industrial co-agglomeration in China. Against this backdrop, Chen and Chen [29], and Chen et al. [30] proposed the γ index and Θ index. The Θ index, as a further optimized version of the γ index, has received widespread attention. Existing research often employs industrial output value, main business income, and the number of employees to measure the scale of regional industries, calculates the degree of industrial agglomeration in a region through location entropy, and ultimately utilizes the Θ index to evaluate the level of co-agglomeration between two industries. The Θ index can simultaneously reflect the quality and extent of co-agglomeration and has become the most frequently used indicator by scholars both domestically and internationally.

The fourth category is research on the co-agglomeration of the digital economy industry and manufacturing. Currently, there is limited literature directly exploring the co-agglomeration of the digital economy industry and manufacturing. Yang et al. [31] investigate the impact mechanism of the co-agglomeration of the digital industry and manufacturing on enterprise green innovation. In measuring the degree of co-agglomeration of the digital industry and manufacturing, they use the E-G index method and quantify the level of industrial co-agglomeration in 27 Chinese provinces from 2003 to 2015 based on employment data from the digital industry and manufacturing. By constructing a fixed-effects model, they deeply analyze the driving effect of co-agglomeration on enterprise green innovation. The research findings indicate that the co-agglomeration of the digital industry and manufacturing not only enhances the advanced nature of the regional labor force but also promotes the convergence and exchange of innovative talent, thereby providing a strong impetus for enterprise green innovation. Additionally, the co-agglomeration effect helps break down information barriers, accelerates the dissemination and application of green technology, and thereby enhances enterprises’ innovation efficiency and green competitiveness.

Through the induction and analysis of existing research, it can be seen that studies on digital talent, the measurement and agglomeration characteristics of the digital economy industry and manufacturing, influencing factors, and effect mechanisms are relatively mature, providing valuable references for this study. However, there are still areas with potential for expansion:

First, while the existing literature has started to focus on the impact of the co-agglomeration of the digital economy industry and manufacturing on corporate green innovation, and has revealed the social benefits it brings, relevant research remains relatively limited. Specifically, only one study has explored the effects of the co-agglomeration of the digital economy industry and manufacturing, and none of the literature has delved deeply into the specific factors influencing this co-agglomeration. Therefore, there is a clear research gap in exploring the motivations and mechanisms behind the co-agglomeration of the digital economy industry and manufacturing. Further research in this area is crucial for a more comprehensive understanding of the phenomenon of the co-agglomeration of the digital economy industry and manufacturing.

Second, existing research has touched upon the mechanism by which digital talent influences green technology innovation, individual innovative behavior, and open innovation performance, fully revealing the importance of digital talent in innovation activities. However, to effectively form a cluster of digital talent, a region must attract the inflow of such talent. Unfortunately, current research on digital talent has not fully explored the issue of digital talent mobility, especially the specific effects and mechanisms of digital talent inflow on a region. This research gap limits our comprehensive understanding of the phenomenon of digital talent clustering and hinders our ability to more effectively utilize digital talent resources to promote regional innovation and development in practice. Therefore, there is a clear gap in the existing research regarding the inflow effects and mechanisms of digital talent, and this area holds significant value for further exploration.

Third, when exploring the measurement of industrial agglomeration, most scholars tend to adopt the location entropy method, utilizing data on industrial output value, main business income, and employment to estimate the degree of industrial agglomeration. However, these data primarily originate from the China City Statistical Yearbook and the statistical yearbooks of various provinces and cities, where multiple segments of the digital economy industry are not adequately represented. This leads to a lack of detailed data for these segments. Therefore, existing research on measuring the agglomeration of the digital economy industry is often limited to rough estimations based on a small subset of industries within the sector. For instance, Xin [11] uses only the main business income of digital manufacturing and digital services as the measurement criterion, while Chen et al. [25] rely solely on the number of employees in the information and software services industry. These studies are constrained by data availability and granularity, making it difficult to comprehensively and accurately reveal the agglomeration status of the digital economy industry.

Furthermore, location entropy mainly reflects the degree of industrial specialization and neglects the actual spatial distribution of industries. In practical applications, the presence of leading enterprises can easily lead to inflated data on regional output value, main business income, and employment, subsequently resulting in exaggerated agglomeration indices. However, in reality, there may not be many enterprises present in the region, lacking a truly diversified business ecosystem, which does not align with the essential connotation of industrial agglomeration.

Fourth, when measuring the co-agglomeration of the digital economy industry and the manufacturing, a key challenge lies in the overlap between these two sectors, necessitating a clear distinction between them. However, existing data on output value, main business income, and employment have not effectively separated the digital economy industry from the manufacturing, leading to the issue of double-counting of data from both industries during the measurement process. For example, Yang et al. [31] used indicators of digital industry employment and manufacturing employment in their measurements. Since the manufacturing employment includes some digital industry employees, this has led to double-counting of employment indicators between the digital economy industry and manufacturing.

Based on this background, this research aims to delve into the impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing, and to analyze the specific mechanisms through which digital talent influences this co-agglomeration process. Furthermore, this study will also examine the factors that may constrain the exertion of such impacts, with the expectation of providing valuable insights for both theory and practice. Compared with existing research, the innovations of this study are mainly reflected in the following aspects:

First, innovation from a research perspective. This study incorporates digital talent inflow and the co-agglomeration of the digital economy industry and manufacturing into a unified analytical framework, deeply exploring the internal mechanisms through which digital talent inflow promotes the co-agglomeration of the digital economy industry and manufacturing in both theoretical and empirical dimensions. It empirically verifies the correlation between digital talent inflow and industrial co-agglomeration, providing a new perspective and evidence for research in the field of co-agglomeration of the digital economy industry and manufacturing.

Second, innovation in variable measurement. In terms of measuring the explained variable, this study selects the number of urban digital economy enterprises and manufacturing enterprises from 2006 to 2022 as indicators to measure industrial agglomeration. The advantage of this method is that urban enterprise data can be precisely retrieved and classified by industry, thus ensuring that various segments of the digital economy industry are covered. This effectively compensates for the lack of detailed industry data in China’s Urban Statistical Yearbook, making the measurement of the digital economy industry more comprehensive and accurate. At the same time, relying on enterprise data from segmented industries, the digital economy industry can be effectively separated from the manufacturing, avoiding the problem of double-counting data from the two industries in previous studies. In addition, starting from the essence of industrial agglomeration, this study directly uses the number of enterprises to reflect the agglomeration capacity of enterprises in a region, rather than estimating the level of industrial agglomeration in the region through indirect indicators. This approach provides a new data source for the calculation of the Θ index, making it no longer limited to traditional data such as output value, main business income, and employment. It further enriches the measurement dimensions of industrial agglomeration in the calculation of the Θ index and provides a new perspective and method for the study of industrial agglomeration.

In terms of explanatory variable measurement, this study focuses on the emerging and crucial field of digital talent. We select employees in the information transmission, software, and information technology services sector as the core indicator of digital talent. By employing the gravity model and identifying five key factors—housing prices, employment environment, ecological environment, salary levels, and education levels—as significant influences on the inflow of digital talent, this study conducts a quantitative measurement of digital talent inflow. This measurement not only closely aligns with the current macro trend of digitization but also fills a gap in the research field of digital talent inflow measurement, providing valuable reference and insights for subsequent academic research and practical applications.

Third, innovation in research mechanisms. This study delves into the mechanisms of digital technology innovation, digital knowledge spillover, digital knowledge inflow, entrepreneurial activity in the digital economy industry, and urban industrial structure upgrading in the impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing. At the same time, it also analyzes the moderating role of digital policies, revealing the reasons for the co-agglomeration of the digital economy industry and manufacturing from a deeper level.

Fourth, this study comprehensively considers the heterogeneous characteristics of the urban manufacturing hierarchy, digital infrastructure, urban hierarchy, and economic structure, thoroughly examining the differences in the effects of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing. Additionally, this study further explores the spatial spillover effects of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing, comprehensively examining the direct and indirect effects of digital talent inflow.

3. Theoretical Mechanism

New economic geography divides the internal driving forces of economic spatial agglomeration into two types: the first is the connection derived from traditional economic activities, known as “economic linkage”; the second is the connection brought by knowledge innovation, diffusion, or dissemination, known as “knowledge linkage”. Both types of linkages possess the power to generate agglomeration and constitute the main components of economic agglomeration forces in the real world. Berliant and Fujita [32], as well as Fujita [33], established a dynamic model of knowledge innovation and diffusion to explain how the inflow of intellectuals can promote the agglomeration of innovation sectors and manufacturing sectors in the same region, thereby forming a co-agglomeration effect between the innovation industry and the manufacturing. Within this theoretical framework, this study argues that the inflow of digital talent can facilitate the agglomeration of the digital economy industry and manufacturing in the same region, thereby fostering a co-agglomeration effect between the two industries.

On the one hand, as a crucial production factor in digital economic activities, digital talent, both in quantity and quality, significantly impacts the transformation and upgrading of the regional digital economy industry and manufacturing. According to the labor market search and matching theory, an efficient labor market matching mechanism can enhance resource allocation efficiency and subsequently promote economic growth [34]. When selecting a location, enterprises, in order to reduce the cost of factor selection, tend to prefer cities with an abundant labor force and a highly matched supply–demand structure. The inflow of digital talent not only strengthens the labor force in these cities but also optimizes the urban supply–demand structure, bringing a significant talent agglomeration effect and effectively filling the gap in digital talent in cities. Meanwhile, the inflow of digital talent also significantly improves the efficiency of urban labor allocation, enabling the digital economy industry and manufacturing to more quickly match digitally skilled talents that meet their needs, thereby reducing recruitment costs and talent search costs for digital economy enterprises and manufacturing enterprises. The advantage of cost reduction further attracts more digital economy enterprises and manufacturing enterprises to agglomerate there, thereby forming a co-agglomeration effect between the digital economy industry and manufacturing.

On the other hand, the inflow of digital talent signifies the agglomeration of a large number of talents who possess advanced digital technology and information technology in the same city, providing a shared digital talent pool for digital economy enterprises and manufacturing enterprises. Digital talents usually possess interdisciplinary knowledge backgrounds and can serve as a bridge between digital economy enterprises and manufacturing enterprises. They can introduce advanced technological concepts from the digital economy industry into manufacturing and feedback on the actual demands of manufacturing to the digital economy industry, thereby driving collaborative innovation between the two industries in terms of products, services, and technology. In this process, communication and cooperation between digital economy enterprises and manufacturing enterprises can proceed smoothly, and collaboration costs arising from information asymmetry are effectively reduced.

Furthermore, in the process of collaborative cooperation between the digital economy and manufacturing, by sharing digital talent resources, manufacturing enterprises can avoid independently cultivating or extensively recruiting digital talents, achieve optimal allocation and efficient utilization of talent resources, and subsequently reduce costs related to talent recruitment and training, thereby alleviating the operational burden of manufacturing enterprises. The reduction in collaboration costs and operational costs further enhances the synergy between digital economy enterprises and manufacturing enterprises, promoting their agglomerated development in the same city.

In addition, this study further proposes that the inflow of digital talents exerts multiple pathways in promoting the co-agglomeration of the digital economy industry and manufacturing. Specifically, these pathways include fostering digital technology innovation, accelerating knowledge spillovers and flows, enhancing the entrepreneurial vitality of urban digital economy enterprises, and facilitating the optimization and upgrading of urban industrial structures. Meanwhile, the implementation of digital policies can provide robust support for the inflow of digital talent, further reinforcing this positive effect. The elaboration is as follows:

The inflow of digital talent can promote digital technology innovation. According to the resource-based theory, the competitive advantage of a firm is derived from its unique, valuable, and difficult-to-imitate and -substitute resources [35,36]. In the digital economy sector, digital talents themselves constitute a core resource. When a large number of digital talents flow into the digital economy industry of a city, the innovative elements such as digital knowledge, skills, and technologies they bring along also converge [37], providing a more favorable external environment for technological innovation in digital economy enterprises [38,39]. This prompts digital economy enterprises to integrate and optimize innovation resources more effectively, thereby achieving breakthroughs in technological innovation. Meanwhile, the inflow of digital talent intensifies competition in the talent market. The survival-of-the-fittest mechanism in the talent market drives digital talents to continuously enhance their professional skills and innovation capabilities to gain a competitive edge. Furthermore, the existence of collaborative mechanisms enables digital talents with common interests to establish close cooperative relationships [40]. Digital talents with diverse knowledge backgrounds jointly address the challenges of digital technology innovation and promote technological innovation and development in the digital economy sector by sharing resources, exchanging ideas, and other means [41].

The inflow of digital talent can promote the spillover and flow of digital knowledge. According to the theory of knowledge spillover, the agglomeration of labor factors can promote the effect of knowledge spillover, driving down the cost of economic development and enhancing market efficiency [42]. Through academic exchanges, technical seminars, project collaborations, and other means, digital talents can swiftly share and disseminate new digital knowledge and technologies, thereby promoting the spillover and circulation of digital knowledge [43]. The spillover effect of digital knowledge encompasses two dimensions: Firstly, the spillover of explicit knowledge, such as digital technology patents, academic papers, or publications, which can be communicated through formal documents and reports [44], enabling long-distance dissemination and rapid learning and application by other digital talents in new cities. This enhances the technical proficiency and innovation capability of the entire digital talent pool in the inflow city.

Secondly, the spillover of tacit knowledge, including innovative thinking and problem-solving methods related to digital technology. Although such knowledge is difficult to codify and disseminate, it can effectively circulate among digital talents through informal means such as interpersonal interactions and work-related communications, thereby promoting the spillover and circulation of digital knowledge [36,45]. According to the “learning by doing” theory, individuals often learn more effectively through practical experience than through theoretical study alone [46]. Digital talents can gradually master and apply this tacit knowledge during their interactions, further enhancing their digital skills and innovation capabilities.

The inflow of digital talents can enhance the entrepreneurial activity of digital economy enterprises in cities. The “Action Plan for Accelerating the Cultivation of Digital Talents to Support the Development of the Digital Economy (2024–2026)” emphasizes the need to increase investment in entrepreneurial training for digital talents to promote their innovation and entrepreneurship activities in the digital economy sector [47]. This policy highlights the dual role of digital talents: they are both active participants in the construction of the regional digital economy industry and entrepreneurs who inject new vitality into this industry. On the one hand, based on the rational economic agent hypothesis, entrepreneurs will engage in rational analysis during the process of entrepreneurship and business operation, weighing costs and benefits to make optimal decisions [48]. In regions with a significant inflow of digital talent, the abundance of digital human capital can reduce the costs and time investment in talent recruitment and training for digital economy enterprises [49], thereby attracting more entrepreneurs to choose to establish digital economy enterprises locally.

On the other hand, with their professional expertise and comprehensive capabilities, digital talents can transform digital technology achievements into commercial projects and establish and operate digital economy enterprises. This not only expands the number of entrepreneurial groups but also enriches the local entrepreneurial ecosystem, further enhancing the entrepreneurial activity of digital economy enterprises. In addition, to attract more digital talents to choose to start businesses in the city, the government will also introduce a series of policy measures, such as providing government subsidies and tax incentives [36], to reduce entrepreneurial risks and costs, thereby attracting more entrepreneurs to select this city for their entrepreneurial activities.

The inflow of digital talent can promote the upgrade of urban industrial structures. On the one hand, according to the theory of innovation and industrial upgrading, innovation is the core driving force for the upgrading of industrial structure [50]. Through their digital technology innovation effect, digital talents not only promote the adoption of advanced digital technologies by traditional industries in the region but also accelerate the digitization and intelligent transformation of these industries, thereby significantly enhancing production efficiency. This transformation promotes the shift from low-value-added industries to high-value-added and high-tech industries, further realizing the rationalization and upgrading of the industrial structure.

On the other hand, according to the theory of the diffusion effect of leading industries, leading industrial sectors drive the overall economic growth and the upgrading of industrial structure through their rapid growth and diffusion effects [51]. The inflow of digital talent provides a solid foundation for cities that have chosen the digital economy as their leading industry. Digital talents can exert a creative effect on emerging job positions, promoting the upgrading of the region’s industrial structure. By establishing digital economy enterprises, digital talents can generate a large number of digital job positions, such as those in artificial intelligence, the Internet of Things, big data, cloud computing, blockchain, and smart manufacturing engineering and technology, driving the region’s industries towards service-oriented and knowledge-intensive development, and achieving a more diversified and high-end industrial structure.

Digital economy policies can support the inflow of digital talent. On the one hand, regions with robust digital economic policies often experience vigorous development of the digital economy and continuous expansion of market demand. According to signaling theory [52], such policies clearly convey the message of talent demand to the outside world, enabling digital talents to keenly perceive the broad development opportunities and career prospects offered by the region. Meanwhile, policy support accelerates the maturation of related industry chains and the establishment of industrial ecosystems, providing digital talents with more diverse career options and growth paths. Furthermore, regions with strong digital economic policies indicate more investment opportunities, abundant innovation resources, and continuous policy support in the future, which are highly attractive to digital talents pursuing career growth and innovation opportunities. On the other hand, the improvement of regional digital economic policies prompts governments to increase investment in digital infrastructure, thereby optimizing the digital economic ecosystem. Top-tier digital infrastructure creates an efficient working environment for digital talents, effectively attracting their inflow [53], which not only enriches the local talent pool for the digital economy but also provides strong technical support for the digital transformation of traditional manufacturing, promoting deep integration and mutual development between the digital economy and manufacturing.

Based on the above analysis, this study proposes the following research hypotheses:

H1. 

The inflow of digital talent can promote the co-agglomeration of the digital economy industry and manufacturing.

H2. 

The inflow of digital talent can promote digital technology innovation.

H3. 

The inflow of digital talent can promote the spillover and flow of digital knowledge.

H4. 

The inflow of digital talents can enhance the entrepreneurial activity of digital economy enterprises in cities.

H5. 

The inflow of digital talent can promote the upgrade of urban industrial structure.

H6. 

Digital economy policies can support the inflow of digital talent.

The impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing is shown in Figure 1:

4. Indicator Selection and Model Setup

4.1. Explanatory Variable

Digital talents refer to a group of professionals in the information transmission, software, and information technology services sector who possess skills in information and communication technology, data analysis capabilities, and innovative thinking, enabling them to drive innovation in business models and processes. Considering that this category of professionals plays a crucial role in digital tasks such as information technology applications, software development, and information transmission, and possesses a high level of digital knowledge and skills, this study references the research of Huang [6] and selects employees in the information transmission, software, and information technology services sector as an indicator for measuring digital talents. The indicators affecting talent inflow are selected based on the research of Shi et al. [37] and Hu et al. [54], including five factors: housing price level, employment environment, ecological environment, salary level, and education level of the city. The calculation method for digital talent inflow is based on the research of Sun et al. [55] and Li and Chan [56], and the gravity model is constructed as follows:

$${\mathrm{DT}}_{\mathrm{ijt}}={\left({\displaystyle \frac{{\mathrm{hpr}}_{\mathrm{jt}}}{{\mathrm{hpr}}_{\mathrm{it}}}}\right)}^{{\mathsf{\beta }}_1}\times {\left({\displaystyle \frac{{\mathrm{job}}_{\mathrm{jt}}}{{\mathrm{job}}_{\mathrm{it}}}}\right)}^{{\mathsf{\beta }}_2}\times {\left({\displaystyle \frac{{\mathrm{env}}_{\mathrm{it}}}{{\mathrm{env}}_{\mathrm{jt}}}}\right)}^{{\mathsf{\beta }}_3}\times {\left({\displaystyle \frac{{\mathrm{wag}}_{\mathrm{it}}}{{\mathrm{wag}}_{\mathrm{jt}}}}\right)}^{{\mathsf{\beta }}_4}\times {\left({\displaystyle \frac{{\mathrm{edu}}_{\mathrm{it}}}{{\mathrm{edu}}_{\mathrm{jt}}}}\right)}^{{\mathsf{\beta }}_5}\times {\displaystyle \frac{{\mathrm{P}}_{\mathrm{it}}^{{\mathsf{\alpha }}_1}\times {\mathrm{P}}_{\mathrm{it}}^{{\mathsf{\alpha }}_2}}{{\mathrm{d}}_{\mathrm{ij}}^{\mathrm{k}}}}$$

(1)

$${\mathrm{DT}}_{\mathrm{it}}={\displaystyle {\sum }_{\mathrm{j}}^{268}{\mathrm{DT}}_{\mathrm{ijt}}}$$

(2)

In Equations (1) and (2), DT_ijt represents the total number of digital talents flowing from city j to city i in year t. DT_it represents the total number of digital talents flowing from the other 268 cities to the city i in year t. hpr_it, job_it, env_it, wag_it, and edu_it represent the factors influencing the inflow of digital talents, which respectively correspond to the city’s housing price level, employment environment, ecological environment, salary level, and education level. These factors are measured by the average sales price of residential properties, unemployment rate, green coverage rate of built-up areas, average wage of employees in the city, and the proportion of education expenditure in fiscal expenditure. Among them, hpr_it and job_it are negative indicators, while the others are positive indicators; P_it and P_jt, respectively, represent the number of digital talents in the cities i and j in year t. d_ij represents the geographical distance between cities i and j. α is the gravitation parameter for digital talents in the region. Referring to existing research [56], let α₁ = α₂ = 0.5; k is the influence parameter of the geographical distance between cities. Considering that the development of modern transportation has weakened the hindrance of distance to travel, referring to existing research [56], k is set to 1. β is the index weight for attracting digital talents, and the entropy weight method is used to calculate the weights of the five indicators.

4.2. Explained Variable

The measurement of the co-agglomeration between the digital economy industry and manufacturing is primarily divided into two steps: Firstly, the agglomeration levels of the digital economy industry and manufacturing are measured separately. According to the definitions provided in this paper, the agglomeration of the digital economy industry and manufacturing refers to the phenomenon of a high concentration of digital enterprises and manufacturing enterprises within the same region. A greater number of agglomerated enterprises in a region indicates a higher level of industrial agglomeration in that area. Therefore, referencing the research conducted by Yang et al. [24] and Zhang et al. [57], this paper employs the number of digital economy enterprises and manufacturing enterprises to gauge the agglomeration of the digital economy industry and manufacturing in cities. The data on digital economy enterprises and manufacturing enterprises are sourced from the CRM Enterprise Directory website (URL: https://www.curtao.com (accessed on 15 July 2024)), which collects enterprise lists across all industries and regions in China, offering multi-dimensional search functions by enterprise industry, type, status, etc., enabling users to accurately filter and locate the required enterprise information by time, industry, and region.

During the process of enterprise data retrieval, it is necessary to set the industry, enterprise type, and enterprise status. Regarding the industry settings for digital economy enterprises, according to the definition provided in the “Statistical Classification of Digital Economy and Its Core Industries (2021)” issued by the National Bureau of Statistics [58], the digital economy industry encompasses five major categories: digital product manufacturing, digital product services, digital technology applications, digital factor-driven industries, and digital efficiency enhancement industries. Among these, the first four categories are defined by the National Bureau of Statistics as industries that can provide digital technology, products, infrastructure, and solutions for the development of other industries, serving as crucial sectors that enable other industries to achieve digital transformation. The digital efficiency enhancement industry refers to industries that have significantly improved efficiency through the application of digital technology, industries deeply empowered by digital technology.

Given the extensive range of industries involved in digital efficiency enhancement, which covers nearly all sectors of the national economic industry classification, and the difficulty in obtaining relevant data, some scholars only conduct quantitative analysis on the first four categories when studying the digital economy industry [59]. Furthermore, the discussion in this paper focuses on the empowering effect of the digital economy industry on manufacturing, how the digital economy industry provides necessary digital technology, products, infrastructure, and solutions for manufacturing during the process of co-agglomeration, thereby effectively promoting the digital and intelligent transformation of the manufacturing. This focus of discussion is highly aligned with the definition of core digital economy industries provided by the National Bureau of Statistics. Therefore, based on the definition of the National Bureau of Statistics and in conjunction with academic practices in the quantitative research of the digital economy industry, this paper defines the digital economy industry as encompassing four major categories: digital product manufacturing, digital product services, digital technology applications, and digital factor-driven industries. The specific industry settings for retrieval are shown in

Table 1. Industry settings for digital economy industry retrieval.

Core Classification	Industry Segmentation of the National Economy	Core Classification	Industry Segmentation of the National Economy
Digital product manufacturing	Manufacturing of computer	Digital technology applications
	Manufacturing of communication equipment		Software development
	Manufacturing of broadcasting and television equipment		Integrated circuit design
	Manufacturing of radar and supporting equipment		Information system integration services
	Manufacturing of nonprofessional audiovisual equipment		IoT technology services
	Manufacturing of smart consumer devices		Operation and maintenance services
	Manufacturing of electronic devices		Information processing and storage support services
	Manufacturing of electronic components and specialized electronic materials		Information technology consulting services
	Manufacturing of other electronic devices		Digital content services
Digital product services	Retail of computers, software, and auxiliary equipment		Other information technology service industries
	Retail of communication equipment		Telecommunication
Digital factor-driven industries	Internet platform		Broadcasting and television transmission services
	Broadcast		Satellite transmission services
	Television		Internet access and related services
	Film and television program production		Internet search services
	Integrated broadcasting and television control		Internet security services
	Distribution of movies and radio and television programs		Internet data service
	Movie screening		Other Internet-related services
	Sound recording

.

Regarding the industry setting for manufacturing enterprises, there is a consensus in academia on the definition of manufacturing, primarily referencing the “Classification of National Economic Industries (GB/T4754-2017)”, which encompasses a total of 31 manufacturing segments [60]. In exploring the impact of the digital economy industry on the co-agglomeration of manufacturing, this study particularly focuses on the “Statistical Classification of Digital Economy and Its Core Industries (2021)” released by the National Bureau of Statistics [58], which explicitly categorizes computer, communication, and other electronic equipment manufacturing as part of the digital economy industry. This indicates that this classification provides digital technology, products, infrastructure, and solutions for the development of other industries. It also implies that, in discussing the impact of the digital economy industry on the co-agglomeration of manufacturing, manufacturing, as the empowered sector, should exclude computer, communication, and other electronic equipment manufacturing that supplies digital technology, products, infrastructure, and solutions. Therefore, the manufacturing sector referred to in this paper specifically includes the remaining 30 manufacturing segments that have not yet fully completed the digital empowerment transformation, excluding computer, communication, and other electronic equipment manufacturing.

Based on the above industry settings, digital economy enterprises and manufacturing enterprises were searched with the following conditions set: (1) The time range was set from 1 January 2006, to 31 December 2022. (2) The enterprise types were set as limited liability companies, joint-stock companies, state-owned enterprises, foreign-invested enterprises, personal exclusive investment enterprises, collectively-owned enterprises, limited partnerships, and general partnerships. (3) The enterprise status was set as surviving and operational. Finally, Python software was used to crawl the retrieved enterprise data, obtaining a total of 4,078,596 digital economy industry enterprises and 7,108,011 manufacturing enterprise data. Based on the establishment years and addresses of these enterprises, they were matched to their respective years and cities to calculate the agglomeration levels of the digital economy industry and manufacturing in various cities across different years.

Furthermore, the degree of co-agglomeration between the digital economy industry and manufacturing in cities was measured. Currently, the measurement of co-agglomeration indices mainly adopts the co-agglomeration index method and the Θ index method. The specific calculation formulas are as follows:

$${\mathrm{Cor}}_{\mathrm{it}}=\left(1-{\displaystyle \frac{\left|{\mathrm{Deagg}}_{\mathrm{it}}-{\mathrm{Managg}}_{\mathrm{it}}\right|}{{\mathrm{Deagg}}_{\mathrm{it}}+{\mathrm{Managg}}_{\mathrm{it}}}}\right)$$

(3)

$${\mathrm{Cor}}_{\mathrm{it}}=\left(1-{\displaystyle \frac{\left|{\mathrm{Deagg}}_{\mathrm{it}}-{\mathrm{Managg}}_{\mathrm{it}}\right|}{{\mathrm{Deagg}}_{\mathrm{it}}+{\mathrm{Managg}}_{\mathrm{it}}}}\right)+\left|{\mathrm{Deagg}}_{\mathrm{it}}+{\mathrm{Managg}}_{\mathrm{it}}\right|$$

(4)

Equation (3) represents the co-agglomeration index method, while Equation (4) represents the Θ index method. Cor_it indicates the degree of co-agglomeration between the digital economy industry and manufacturing. A higher Cor_it value suggests a higher level of co-agglomeration between the digital economy industry and manufacturing. Deagg_it represents the agglomeration of the digital economy industry in a city, while Managg_it represents the agglomeration of manufacturing in a city. Equation (3) indicates the comparison of agglomeration differences between the digital economy industry and manufacturing, which can reflect the level of co-agglomeration between the digital economy industry and manufacturing in a city. However, it may lead to the same level of industrial co-agglomeration for both low-level and high-level industrial agglomeration. Therefore, most scholars adopt Equation (4) for revision. This study uses the revised relative co-agglomeration index method to calculate the co-agglomeration between the digital economy industry and manufacturing.

4.3. Control Variables

By referencing the literature on the agglomeration of digital economy industries and manufacturing, this study controls for variables that may affect the co-agglomeration of digital economy industries and manufacturing. Specifically, the following ten control variables are included: (1) economic development level, measured by the natural logarithm of urban gross production; (2) level of openness, measured by the natural logarithm of per capita actual foreign capital utilization; (3) degree of government intervention, measured by the natural logarithm of general fiscal expenditure of local governments; (4) level of financial development, measured by the natural logarithm of the balance of deposits and loans of financial institutions at the prefecture level; (5) level of informatization, represented by the proportion of total postal and telecommunications business volume to GDP at the prefecture level; (6) level of urbanization, measured by the proportion of urban population to total population at the prefecture level; (7) industrial structure, represented by the ratio of the sum of secondary and tertiary industries to GDP; (8) investment in innovative personnel, measured by the natural logarithm of personnel in scientific research and technical services; (9) total market size, measured by the natural logarithm of the total retail sales of consumer goods; (10) transportation infrastructure, measured by the natural logarithm of road mileage at the prefecture level. The descriptive statistics of each variable are shown in

Table 2. Descriptive statistics of main variables.

Variable	Unit	N	Mean	SD	Min	Max
The co-agglomeration of the digital economy industry and manufacturing	/	4573	0.600	0.403	0.061	6.936
Digital talent inflow	/	4573	1.200	1.519	0.100	22.278
Level of financial development	Ten thousand yuan	4573	17.234	1.219	14.240	21.848
Level of informatization	%	4573	2.618	1.969	0.232	29.434
Economic development level	Billion	4573	7.259	0.994	4.251	10.707
Level of openness	USD/10,000 people	4573	13.135	1.686	5.492	16.877
Level of urbanization	%	4573	53.967	16.206	3.191	100.000
Degree of government intervention	Ten thousand yuan	4573	14.642	0.946	11.721	18.358
Industrial structure	%	4573	87.372	8.193	50.110	99.970
Investment in innovative personnel	People	4573	8.478	1.170	3.689	13.483
Total market size	Ten thousand yuan	4573	15.444	1.090	12.119	19.013
Transportation infrastructure	Kilometer	4573	6.427	0.973	1.099	9.614

:

4.4. Data Description

The data on digital economy enterprises and manufacturing enterprises used in this study are sourced from the publicly available information on the CRM enterprise directory website (URL: https://www.curtao.com (accessed on 15 July 2024)). Systematic crawling was conducted using Python software. The housing price data used to measure the inflow of digital talents are primarily collected based on the official statistical yearbooks of various provinces and prefecture-level cities. For cities where housing price data are not covered, the historical data provided by the Anjuke website are used for supplementation and improvement. The remaining data required for measuring digital talents and the relevant data for control variables are derived from the statistical yearbooks of various provinces and prefecture-level cities, the China City Statistical Yearbook, and the China Urban and Rural Construction Database in the EPS database, ensuring the comprehensiveness and accuracy of the data. For a small quantity of missing data, this study adopts the interpolation method to fill in the gaps, ensuring the completeness and continuity of the data series.

4.5. Benchmark Regression Model Specification

When setting the benchmark model, the variables are tested first. The specific steps are as follows: Firstly, the Hausman test is conducted. The results show that the p-value of the Hausman test is 0.000, indicating that it is more appropriate to use the fixed-effects model for estimation. Since each city has some characteristics that do not change over time, and economic variables may also exhibit time-varying characteristics, which may affect the explained variable. Therefore, this study selects a two-way fixed-effects model for both city entities and years to reduce potential biases. Secondly, the White test is conducted. The results show that the p-value of the White test is 0.000, indicating the presence of heteroscedasticity. Therefore, this study uses robust standard errors for estimation to eliminate the impact of heteroscedasticity. Based on this, the following two-way fixed-effects model is constructed to examine the impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing:

$${\mathrm{Cor}}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{DT}}_{\mathrm{it}}+{\mathsf{\alpha }}_2{\mathrm{Control}}_{\mathrm{it}}+{\mathsf{\alpha }}_3{\mathrm{Year}}_{\mathrm{t}}+{\mathsf{\alpha }}_4{\mathrm{City}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(5)

In Equation (5), Control_it represents a series of control variables; Year_t denotes the year fixed effect; City_i indicates the city-specific fixed effect; and ε_it is the random error term.

5. Empirical Results and Analysis

5.1. Analysis of the Spatiotemporal Evolution Characteristics of Digital Talent Inflow and the Co-Agglomeration of the Digital Economy Industry and Manufacturing

5.1.1. Analysis of the Spatiotemporal Evolution Characteristics of Digital Talent Inflow

Based on the calculation results, this study employs the natural breaks classification method in Python software to categorize the level of digital talent inflow in various cities from 2006 to 2022 into five grades. Specifically, these grades include high inflow level (12.709 ≤ DT < 22.277), relatively high inflow level (7.003 ≤ DT < 12.709), medium inflow level (3.206 ≤ DT < 7.003), relatively low inflow level (1.223 ≤ DT < 3.206), and low inflow level (0.100 ≤ DT < 1.223). Furthermore, Origin software is utilized to create sunburst charts illustrating the digital talent inflow in 269 cities for the years 2006, 2014, and 2022, as shown in Figure 2, Figure 3 and Figure 4.

From a temporal dimension analysis, between 2006 and 2022, there was a continuous upward trend in the inflow level of digital talent across various cities, indicating the vigorous development of China’s digital economy and the surge in demand for digital talent. The number of cities with high and relatively high inflow levels increased significantly, from zero in both categories in 2006 to three and four cities, respectively, in 2022, forming multiple clusters of cities with high inflow levels. This reflects the high attention and active investment given by the national and local governments to the development of the digital economy, which has prompted continuous improvement and optimization in multiple dimensions such as digital infrastructure construction and urban quality of life, thereby successfully attracting more digital talent.

Meanwhile, the number of cities with medium, relatively low, and low inflow levels also underwent dynamic adjustments, evolving from 2, 16, and 251 cities in 2006 to 16, 80, and 166 cities in 2022, respectively. The overall trend is towards higher inflow levels, demonstrating the vitality and dynamic balance of digital talent flow across the country. However, cities with relatively low and low inflow levels still dominate, indicating that most cities still face challenges in policy support and resource allocation for promoting digital economic development and attracting digital talent.

From a spatial dimension analysis, between 2006 and 2022, the inflow of digital talent in 269 cities in China exhibited distinct agglomeration effects and regional disparities. Cities with high and relatively high inflow levels are concentrated in economically developed urban agglomerations along the eastern coast. Among them, the Beijing–Tianjin–Hebei region, the Yangtze River Delta, and the Pearl River Delta have become preferred destinations for digital talent due to their advantageous geographical locations, well-developed infrastructure, abundant innovation resources, and open market environment. In spatial distribution, the “core–periphery“ structural characteristic is particularly prominent, with high-inflow cities serving as core nodes effectively driving the development of the digital economy in surrounding cities through mechanisms such as knowledge spillovers and industrial linkages, creating significant radiation effects.

In comparison, while the inflow level of digital talent in central and western regions is relatively lower, from the perspective of spatial evolution, with the increasing support from the state for the development of these regions and their continuous investment in infrastructure construction and industrial transformation and upgrading, their digital economies have also shown vigorous development trends. In particular, central cities such as Chengdu, Wuhan, and Chongqing have gradually become new hotspots for the inflow of digital workers due to their unique locational advantages and favorable development environments, making the spatial distribution more balanced. However, the issue of uneven regional development remains prominent. Cities with high and relatively high inflow levels are still concentrated in developed regions along the eastern coast, while central and western regions as well as some economically underdeveloped cities still face significant challenges in attracting digital workers.

5.1.2. Analysis of the Spatiotemporal Evolution Characteristics of the Co-Agglomeration of the Digital Economy Industry and Manufacturing

Based on the calculation results, this study utilizes the natural breaks classification method in Python software to precisely categorize the level of co-agglomeration between the digital economy industry and manufacturing across various cities from 2006 to 2022 into five grades. Specifically, these grades include a high level of co-agglomeration (3.109 ≤ DT < 6.935), relatively high level of co-agglomeration (1.473 ≤ DT < 3.109), moderate level of co-agglomeration (0.819 ≤ DT < 1.473), relatively low level of co-agglomeration (0.486 ≤ DT < 0.819), and low level of co-agglomeration (0.061 ≤ DT < 0.486). Furthermore, Origin software was used to create a sunburst chart illustrating the levels of co-agglomeration between the digital economy industry and manufacturing in 269 cities for the years 2006, 2014, and 2022. The results are shown in Figure 5, Figure 6 and Figure 7.

From a temporal dimension analysis, between 2006 and 2022, the level of industrial co-agglomeration in 269 cities in China achieved significant growth. In 2006, there were no cities with high or relatively high levels of co-agglomeration; only a few cities reached a moderate level, while the majority were at lower or low levels. However, with the vigorous development of the digital economy and the active promotion of national policies, more cities have gradually achieved effective collaboration between the digital economy and manufacturing through optimizing industrial structures, enhancing technological innovation, adjusting development strategies, and increasing resource investment. By 2022, the number of cities with high and relatively high levels of co-agglomeration had increased significantly, reaching 6 and 24 cities, respectively. This change indicates that the deep integration of the digital economy and manufacturing has become a prevalent trend across the country, and it is expected that more cities will achieve high levels of co-agglomeration in the future.

Meanwhile, although the number of cities with moderate, lower, and low levels of co-agglomeration fluctuated, they remained stable overall, transitioning from 29, 83, and 157 in 2006 to 86, 141, and 12 in 2022. This further indicates that the widespread penetration and deep development of the digital economy have prompted more cities to prioritize the collaborative development of the digital economy and manufacturing, actively enhancing their co-agglomeration levels through various measures. However, due to differences in resource endowments, economic foundations, and policy support, these cities still face significant challenges in the collaborative development process of the digital economy and manufacturing.

From a spatial dimension analysis, the eastern coastal regions, as the most economically developed areas in China, have become hubs of co-agglomeration between the digital economy and manufacturing. Cities with high levels of co-agglomeration are concentrated in the eastern coastal regions, reflecting the relative lag in digital economy development in the central and western regions. In terms of agglomeration characteristics, China’s co-agglomeration between the digital economy and manufacturing exhibits a pattern of “core–periphery”, “multi-core agglomeration”, and “gradient diffusion”. Cities with high co-agglomeration are concentrated in economic centers, particularly in the Yangtze River Delta, Pearl River Delta, and Beijing–Tianjin–Hebei urban agglomerations, forming the core of high co-agglomeration. In contrast, cities with lower co-agglomeration are dispersed in peripheral regions.

At the same time, cities such as Chengdu, Chongqing, Xi’an, and Changsha in the central and western regions, leveraging their own resources and location advantages, actively undertake industrial transfers from the east and gradually become new engines for the collaborative development of the digital economy and manufacturing, forming a complementary and “multi-core agglomeration” development pattern with the eastern regions. This spatial distribution pattern not only helps alleviate regional development imbalances but also promotes deep integration and collaborative development of the digital economy and manufacturing across the country. Within urban agglomerations, co-agglomeration also exhibits differentiated characteristics, with core cities such as Shanghai, Guangzhou, and Shenzhen having significantly higher levels of co-agglomeration than surrounding cities, forming a gradient distribution that promotes knowledge spillovers, technology diffusion, and industrial collaboration within urban agglomerations. Overall, the spatial distribution of China’s co-agglomeration level in the digital economy industry is still uneven, but the level of industrial co-agglomeration in the central and western regions is gradually improving, gradually forming a new pattern of multi-regional collaborative development, which alleviates the issue of regional imbalances.

5.2. Benchmark Regression Results

In the benchmark regression analysis, this study employed a two-way fixed effects model to explore the impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing and conducted regression analysis by gradually introducing control variables.

Table 3. Benchmark regression results.

Variable	Cor
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
DT	0.172 ***	0.161 ***	0.208 ***	0.125 **	0.128 **	0.199 ***
	(0.041)	(0.040)	(0.045)	(0.050)	(0.050)	(0.045)
Constant	0.394 ***	0.407 ***	0.350 ***	−0.639	−0.123	−0.175
	(0.042)	(0.042)	(0.054)	(0.472)	(0.509)	(1.249)
N	4573	4573	4573	4573	4573	4573
Adj R2	0.420	0.524	0.822	0.509	0.563	0.828
Control variables	NO	NO	NO	YES	YES	YES
Urban fixed effects	NO	YES	YES	NO	YES	YES
Fixed year effects	NO	NO	YES	NO	NO	YES

Note: ** and *** respectively indicate that the estimated coefficients are significant at the 5% and 1% levels. The values in parentheses are robust standard errors.

presents the benchmark regression results. Specifically, Model 1 serves as the benchmark model, incorporating only the two core variables of digital talent inflow and the co-agglomeration of the digital economy industry and manufacturing. Models 2 to 6 gradually control for year-fixed effects, city-specific fixed effects, and control variables. The empirical results indicate that even after strictly controlling for the two-way fixed effects of cities and years and gradually introducing various control variables, the estimated coefficient for digital talent inflow remains positive at the 1% significance level, suggesting that the inflow of digital talent significantly promotes the synergistic effect between the digital economy industry and manufacturing, injecting new vitality into the formation and development of industrial co-agglomeration. Therefore, this study preliminarily validates research hypothesis 1.

5.3. Endogeneity Test

Considering that the enhancement of the co-agglomeration of the digital economy industry and manufacturing in cities may attract more digital talent inflow, which could lead to a bidirectional causality issue, this study further employs Granger causality tests and instrumental variable methods to explore this issue.

5.3.1. Granger Causality Tests

For the Granger causality tests, this study adopts a panel VAR model and determines the optimal lag order of variables based on the Bayesian information criterion (BIC). The results in

Table 4. Lag order test and Granger causality test.

Explanatory Variable	Explained Variable	Lag Order Test		Granger Causality Test
		Order	BIC	p-Value	Conclusion
The inflow of digital talent	The co-agglomeration of the digital economy industry and manufacturing	Lag 1	−25,035.839 *	0.000	The inflow of digital talent is a Granger cause of the co-agglomeration of the digital economy industry and manufacturing.
		Lag 2	−22,261.273
		Lag 3	−19,969.573
		Lag 4	−17,206
		Lag5	−14,822.049
The co-agglomeration of the digital economy industry and manufacturing	The inflow of digital talent	Lag 1	−9562.043 *	0.108	The co-agglomeration of the digital economy industry and manufacturing is not a Granger cause of the inflow of digital talent.
		Lag 2	−8261.085
		Lag 3	−6824.807
		Lag 4	−5567.507
		Lag 5	5307.609

Note: * indicates that the estimated coefficient is significant at the 10% level.

indicate that the optimal lag order is 1 for both the impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing, and the reverse impact of such the co-agglomeration of the digital economy industry and manufacturing on digital talent inflow. Based on this key parameter, Granger causality tests are conducted. The results in Table 4 show that the p-value for the Granger causality test of digital talent inflow causing the co-agglomeration of the digital economy industry and manufacturing is 0.000, indicating that digital talent inflow is a cause of such co-agglomeration. The p-value for the Granger causality test of the co-agglomeration of the digital economy industry and manufacturing causing digital talent inflow is 0.108, suggesting that such co-agglomeration is not a cause of digital talent inflow. Therefore, there is a one-way causal relationship between digital talent inflow and the co-agglomeration of the digital economy industry and manufacturing.

5.3.2. Instrumental Variable Method

In terms of the instrumental variable method, this study selects the number of national-level intangible cultural heritage projects possessed by a city as the instrumental variable from the perspective of urban cultural heritage. This variable satisfies the two major conditions for being an instrumental variable: Firstly, national-level intangible cultural heritage, as a significant marker of the historical and cultural achievements of the Chinese nation, carries rich historical memories and cultural information, reflecting the vitality and creativity of the Chinese nation. If a city possesses a greater number of national-level intangible cultural heritage items, it indicates that the city has a profound accumulation in cultural heritage, capable of effectively protecting and inheriting these precious cultural treasures. While focusing on professional development, digital talent also values personal life experiences and enriching their spiritual world. The unique cultural charm and historical heritage of a city can provide digital talent with richer and more diverse life experiences and spiritual sustenance, thereby enhancing the city’s attractiveness. This aligns with the correlation requirement of the instrumental variable.

Secondly, the co-agglomeration of digital economy industries and manufacturing, as a complex economic phenomenon, primarily relies on multiple factors such as market demand, technological innovation, policy support, and close collaboration within the upstream and downstream of the industrial chain. These factors have no direct connection with the number of representative national-level intangible cultural heritage items possessed by a city, satisfying the “exclusion restriction” of the instrumental variable, also known as the exogeneity requirement.

The data on national-level intangible cultural heritage items in this study are sourced from the China Intangible Cultural Heritage website (URL: https://www.ihchina.cn (accessed on 15 July 2024)). This website systematically compiles the five batches of national-level intangible cultural heritage lists announced by the State Council in 2006, 2008, 2011, 2014, and 2021, covering a total of 10 categories, 1557 national-level intangible cultural heritage representative items, and 3610 sub-items. To measure the cultural heritage (IV) of each city, this study matches the application regions of the aforementioned items with specific cities and evaluates the instrumental variable based on the actual number of national-level intangible cultural heritage items possessed by each city each year.

Based on this, this study employs the two-stage least squares method to estimate the selected instrumental variable.

Table 5. Instrumental variable method.

Variable	DT	Cor
	Model 1	Model 2
DT		0.194 ***
		(0.008)
IV	0.100 ***
	(0.003)
Kleiberen–Paap rk LM statistic	22.942 ***
Kleiberen–Paap rk Wald F statistic	113.281
	{16.380}
Constant	−0.794	−2.324 ***
	(1.713)	(0.487)
N	4573	4573
Adj R2	0.861	0.839
Control variables	YES	YES
Urban fixed effects	YES	YES
Fixed year effects	YES	YES

Note: *** indicates that the estimated coefficient is significant at the 1% level. The numerical values in { } are the critical values at the 10% level of the Stock–Yogo weak identification test.

reports the regression results of the instrumental variable method, where Model 1 presents the regression analysis results from the first stage, and Model 2 presents those from the second stage. The empirical research results indicate that the regression coefficient of the instrumental variable IV is significantly positive at the 1% level, suggesting a strong positive correlation between the instrumental variable and digital talent inflow, meeting the correlation requirement of the instrumental variable. Meanwhile, the regression results of the digital talent inflow indicator are also significantly positive at the 1% level, which fully demonstrates that even after considering endogeneity issues, the positive impact of digital talent inflow on the co-agglomeration of digital economy industries and manufacturing remains robust, further validating the robustness of the core conclusions of this paper. Additionally, the Kleibergen–Paap rk LM statistic and the Kleibergen–Paap rk Wald F statistic of the instrumental variable successfully pass the under-identification test and the weak instrumental variable test, fully indicating that the selected instrumental variable meets the criteria.

5.4. Robustness Test

5.4.1. Alternative Variable Measurement Methods

Given that the measurement methods of variables can significantly impact empirical results, this paper employs alternative variable measurement methods for robustness checks. Regarding the re-measurement of the explained variable: (1) The unadjusted co-agglomeration index method is used to re-measure the co-agglomeration index of the digital economy industry and manufacturing, and the regression analysis is conducted again based on the new measurement results. Model 1 in

Table 6. Alternative variable measurement methods.

Variable	Cor1	Cor2	Cor3	Cor
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
DT	0.026 ***	0.257 ***	0.215 ***
	(0.007)	(0.045)	(0.050)
DT1				2.767 ***
				(0.595)
DT2					0.220 ***
					(0.044)
DT3						28.299 ***
						(6.053)
Constant	2.486 ***	1.546	−0.464	0.254	−0.192	0.302
	(0.540)	(1.856)	(1.423)	(1.301)	(1.230)	(1.310)
N	4573	4573	4573	4573	4573	4573
Adj R2	0.834	0.746	0.800	0.814	0.833	0.812
Control variables	YES	YES	YES	YES	YES	YES
City fixed effects	YES	YES	YES	YES	YES	YES
Fixed year effects	YES	YES	YES	YES	YES	YES

Note: *** indicates that the estimated coefficient is significant at the 1% level.

presents the regression results after re-measurement. (2) Referring to the research by Yang et al. [31], this study uses the number of employees in the digital industry and manufacturing to measure the agglomeration indicators of the digital economy industry and manufacturing, and constructs a new co-agglomeration index of the digital economy industry and manufacturing by recalculating the location entropy index and industrial co-agglomeration index. Model 2 in Table 6 presents the regression results after re-measurement. (3) In defining manufacturing for this study, it excludes computer, communication, and other electronic equipment manufacturing, encompassing the remaining 30 manufacturing segments that have not fully completed the digital transformation. To examine whether this approach significantly impacts the empirical results, this section adjusts the scope of the digital economy industry accordingly, excluding computer, communication, and other electronic equipment manufacturing while adding this industry to the manufacturing classification, resulting in a total of 31 manufacturing segments. Model 3 in Table 6 presents the regression results after re-measurement.

Regarding the re-measurement of explanatory variables: (1) When using the Lawry gravity model to measure the talent inflow indicator, scholars typically set the influence parameter k for intercity geographical distance to 1 or 2. To avoid the impact of such setting differences on the measurement results, this study refers to the research by Sun et al. [55] and chooses to set the k value to 2. Model 4 in Table 6 presents the regression results after re-measurement. (2) Considering that different choices for indicators of attracting digital talent inflow may also lead to variations in results, this paper selects the number of unemployed people in urban areas, sewage treatment rate, and the number of full-time teachers in regular institutions of higher learning to re-evaluate the urban employment environment, ecological environment, and educational environment. Meanwhile, the influence parameter k for geographical distance is set to both 1 and 2. Models 5 and 6 in Table 6 present the regression results after re-measurement.

The empirical results in Table 6 indicate that even after changing the measurement methods for the digital talent inflow indicator and the co-agglomeration index of the digital economy industry and manufacturing, the digital talent inflow indicator remains positive at a 1% significance level. This suggests that the research conclusions of this paper are not affected by changes in the measurement methods of the digital talent inflow and co-agglomeration indicators of the digital economy industry and manufacturing, thereby verifying the robustness of the core conclusions of this paper.

5.4.2. Changing the Statistical Model

Considering that the choice of statistical models may lead to variations in empirical results, this paper further employs the GLS model to re-estimate the empirical results. The GLS model is effective in addressing issues of heteroscedasticity and autocorrelation in the data, enhancing the efficiency and accuracy of estimation by adjusting the variance in the error term.

Table 7. Test results of robustness.

Variable	Cor
	Model 1	Model 2	Model 3	Model 4	Model 5
DT	0.199 ***	0.202 ***	0.253 ***	0.129 ***	0.147 ***
	(0.004)	(0.044)	(0.050)	(0.034)	(0.030)
Constant	−2.402 ***	−0.222	0.316	0.509	2.699 ***
	(0.477)	(1.484)	(0.940)	(0.763)	(0.898)
N	4573	4573	4505	3672	3766
Adj R2		0.829	0.841	0.798	0.852
Control variables	YES	YES	YES	YES	YES
City fixed effects	YES	YES	YES	YES	YES
Fixed year effects	YES	YES	YES	YES	YES

Note: *** indicates that the estimated coefficient is significant at the 1% level.

reports the regression results after changing the statistical model, with Model 1 presenting the estimation results of the GLS model. The empirical results indicate that even after switching to the GLS model, the indicator of digital talent inflow remains significantly positive at the 1% level. This robustness result suggests that the research conclusions of this paper are not affected by the change in statistical models, thereby verifying the robustness of the core conclusions of this paper.

5.4.3. Winsorization

Considering that extreme values in the sample may have a significant impact on empirical results, potentially leading to biased model estimates, this study chose to perform 1% winsorization on the data to conduct robustness checks. Model 2 in Table 7 reports the regression results after 1% winsorization. The robustness results indicate that even after considering the removal of extreme outliers, the inflow of digital talent indicator remains significantly positive, suggesting that the core findings of this study are not affected by extreme values, further verifying the robustness of the core conclusions of this paper.

5.4.4. Adjusting the Sample Period

Taking into account the uniqueness of municipalities directly under the central government and coastal cities in terms of economic development level and policy support, these factors may have a significant impact on empirical results. To ensure the robustness and universality of the research findings, this paper further adopts the method of excluding samples from municipalities directly under the central government and coastal cities for robustness testing. Furthermore, given the significant impact of traffic restrictions and lockdown measures implemented during the COVID-19 pandemic on population inflow, this paper also excludes samples from the COVID-19 period and uses data from 2006 to 2019 for re-regression. Models 3, 4, and 5 in Table 7 present the regression results after excluding samples from municipalities directly under the central government, coastal cities, and COVID-19 years, respectively. The robust results indicate that even after excluding samples from municipalities directly under the central government, coastal cities, and the COVID-19 period, the inflow of digital talent indicators remains significantly positive, indicating that the inflow of digital talent indicators has broad applicability and persistence in the sample of this study, and is not limited to samples from municipalities directly under the central government, coastal cities, or specific time intervals to maintain its significance, further verifying the robustness of the core conclusions of this paper.

5.4.5. Multi-Dimensional Fixed Effects

Although this paper has incorporated city and year fixed effects into the benchmark regression model, there may still be other unobserved factors affecting the research results. Therefore, this study further introduces provincial fixed effects and the interaction fixed effects between province and year to conduct more rigorous robustness tests.

Table 8. Test results of robustness.

Variable	Cor
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
DT	0.199 ***	0.243 ***				0.279 ***
	(0.045)	(0.053)				(0.054)
LDT1			0.208 ***
			(0.046)
LDT2				0.218 ***
				(0.046)
LDT3					0.226 ***
					(0.045)
DT2						−0.005
						(0.003)
Constant	−0.175	−0.733	−0.623	−1.087	−1.690	0.088
	(1.253)	(1.628)	(1.252)	(1.237)	(1.272)	(1.183)
N	4573	4437	4304	4035	3766	4573
Adj R2	0.827	0.871	0.840	0.850	0.862	0.834
Control variables	YES	YES	YES	YES	YES	YES
City fixed effects	YES	YES	YES	YES	YES	YES
Fixed year effects	YES	YES	YES	YES	YES	YES
Fixed effects of provinces	YES	NO	NO	NO	NO	NO
Interaction terms between provinces and cities	NO	YES	NO	NO	NO	NO

Note: *** indicates that the estimated coefficient is significant at the 1% level.

reports the regression results after incorporating multi-dimensional fixed effects, where Models 1 and 2 present the estimation results after adding provincial fixed effects and the interaction fixed effects between province and year, respectively. The empirical results indicate that even after incorporating provincial fixed effects and the interaction fixed effects between province and year, the inflow of digital talent indicator remains significantly positive, suggesting that the core conclusions of this study are not affected by other unobserved factors, further verifying the robustness of the core conclusions of this paper.

5.4.6. Inclusion of Lagged Terms

Considering that the inflow of digital talent may require an adaptation period before exerting a significant impact on the co-agglomeration of the digital economy industry and manufacturing, there may be a time lag effect. To comprehensively explore this temporal dynamic effect, this study chooses to incorporate one-period, two-period, and three-period lagged terms of the digital talent inflow indicator into the model for robustness testing. Models 3, 4, and 5 in Table 8 report the estimation results after including the one-period, two-period, and three-period lagged terms of the digital talent inflow indicator. The robustness results indicate that even after considering the time lag effect of the digital talent inflow indicator, it remains significantly positive, suggesting that the core findings of this study are not affected by the time lag effect of the digital talent inflow indicator, further verifying the robustness of the core conclusions of this paper.

5.4.7. Nonlinearity Test

Considering that the impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing may not be strictly linear, this study further conducts a nonlinearity test by adding the squared term of the digital talent inflow indicator to the benchmark model. Model 6 in Table 8 reports the test results for the nonlinear relationship between digital talent inflow and the co-agglomeration of the digital economy industry and manufacturing. The robustness results indicate that the squared term of the digital talent inflow indicator is not significant, suggesting that there is a linear relationship rather than a nonlinear relationship between digital talent inflow and the co-agglomeration of the digital economy industry and manufacturing, and the benchmark regression model of this paper is reliable.

5.4.8. Omitted Variable Test

Although this study has included ten key control variables in the benchmark regression model, there may still be other omitted variables that have not been considered. To further explore the impact of these potential omitted variables on the conclusions of this study, this paper adopts the method proposed by Oster [61] to test for omitted variables. Under the specific condition of setting α₁ = 0, the calculated δ value is 1.908, which is greater than 1. The δ value is used to quantify the relative importance of omitted variables compared to observed control variables. This robustness test result indicates that if there are indeed omitted variables, their importance would exceed all currently observed control variables by 1.908 times. Therefore, the core conclusions of this study are unlikely to be overturned due to omitted variable issues, further verifying the robustness of the core conclusions of this paper.

5.5. Analysis of the Mechanism of Action

5.5.1. Model Specification

The empirical results presented earlier demonstrate that the inflow of digital talent plays a significant role in promoting the co-agglomeration of the digital economy industry and manufacturing. However, the underlying mechanisms of action, particularly the specific roles of mediating and moderating mechanisms, have not been thoroughly explored. Based on the theoretical analysis conducted previously, this paper constructs a mediation effect model aimed at revealing the specific pathways and mechanisms through which five mediating variables—digital technology innovation, digital knowledge spillover, digital knowledge flow, digital entrepreneurial activity, and urban industrial structure upgrading—mediate the impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing. Additionally, to gain a more comprehensive understanding of the changing influence of digital talent inflow on the co-agglomeration of these industries and the underlying determinants, this study also constructs a moderation effect model to uncover the moderating role of urban digital economic policies in the impact of digital talent inflow on industrial co-agglomeration. The relevant models are as follows:

$${\mathrm{Mechan}}_{\mathrm{it}}={\mathrm{b}}_0+{\mathrm{b}}_1{\mathrm{Dwo}}_{\mathrm{it}}+{\mathrm{b}}_2{\mathrm{Control}}_{\mathrm{it}}+{\mathrm{b}}_3\mathrm{Year}+{\mathrm{b}}_4\mathrm{City}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(6)

$${\mathrm{Cor}}_{\mathrm{it}}={\mathrm{c}}_0+{\mathrm{c}}_1{\mathrm{Dwo}}_{\mathrm{it}}+{\mathrm{c}}_2{\mathrm{Mechan}}_{\mathrm{it}}+{\mathrm{c}}_3{\mathrm{Control}}_{\mathrm{it}}+{\mathrm{c}}_4\mathrm{Year}+{\mathrm{c}}_5\mathrm{City}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(7)

$${\mathrm{Cor}}_{\mathrm{it}}={\mathrm{d}}_0+{\mathrm{d}}_1{\mathrm{DT}}_{\mathrm{it}}+{\mathrm{d}}_2{\mathrm{Policy}}_{\mathrm{it}}+{\mathrm{d}}_3{\mathrm{Policy}}_{\mathrm{it}}\times {\mathrm{DT}}_{\mathrm{it}}+{\mathrm{d}}_4{\mathrm{Control}}_{\mathrm{it}}+{\mathrm{d}}_5\mathrm{Year}+{\mathrm{d}}_6\mathrm{City}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(8)

In the formula, Mecham_it represents the mediating variable, which indicates the city’s digital technology innovation, inflow of digital knowledge, transfer of digital technology, entrepreneurial activity of digital enterprises, and industrial structure upgrading. Policy_it represents the moderating variable, which indicates the city’s digital economic policies.

5.5.2. Mechanism of Digital Technology Innovation

Based on digital technology patent data, this study evaluates digital technology innovation from both quantitative and qualitative dimensions. The digital technology patent data used is sourced from the incoPat database (URL: https://www.incopat.com (accessed on 15 July 2024)). The process of data acquisition and processing mainly involved the following key steps: Firstly, by referencing the digital technology keywords provided by Wu et al. [62], search terms for digital technology patents were constructed.

Table 9. Search terms for digital technology.

Category	Keywords
Artificial Intelligence Technology	Artificial intelligence, Business intelligence, Image understanding, Investment decision support system, Intelligent data analysis, Intelligent robots, Machine learning, Deep learning, Semantic search, Biometric technology, Face recognition, Speech recognition, Authentication, Autonomous driving, Natural language processing
Big Data Technology	Big data, Data mining, Text mining, Data visualization, Heterogeneous data, Credit investigation, Augmented reality, Mixed reality, Virtual reality
Cloud Computing Technology	Cloud computing, Stream computing, Graph computing, In-memory computing, Multi-party secure computing, Brain-like computing, Green computing, Cognitive computing, Converged architecture, Hundred-million-level concurrency, EB-level storage, Internet of Things, Cyber-physical systems
Blockchain Technology	Blockchain, digital currency, distributed computing, differential privacy technology, intelligent financial contracts
Application of Digital Technology	Mobile Internet, Industrial Internet, Mobile interconnection, Internet healthcare, E-commerce, Mobile payment, Third-party payment, NFC payment, Smart energy, B2B, B2C, C2B, C2C, O2O, Online payment network, Smart wearables, Smart agriculture, Smart transportation, Smart healthcare, Smart customer service, Smart home, Smart investment advice, Smart cultural tourism, Smart environmental protection, Smart grid, Smart marketing, Digital marketing, Unmanned retail, Internet finance, Digital finance, Fintech, Financial technology, Quantitative finance, Open banking

presents the search terms for digital technology patents used in this study:

Secondly, a search was conducted using the constructed search terms, yielding a total of 500,301 digital technology patent records. Subsequently, Python software was utilized to screen and process the raw patent data. This involved eliminating patents from foreign applicants, expired patents, design patents, and patent entries with duplicate application numbers. Ultimately, a total of 374,284 digital technology patent records were obtained, spanning from 2006 to 2022. Furthermore, based on the collected digital technology patent data, this study conducted classification and statistics by year and the city where the applicant is located, thereby obtaining the number of digital technology patents for each city each year. This indicator is used to measure the quantitative performance of each city in digital technology innovation. Following common practice, the patent data were incremented by 1 and then logarithmically transformed.

Furthermore, to delve deeper into the quality performance of digital technology innovation across various cities, this study refers to the research conducted by Li et al. [63], leveraging the patent technology advancement scores calculated by the incoPat patent database to conduct a quantitative evaluation of the quality of digital technology innovation. The specific calculation method is as follows:

$${\mathrm{Quality}}_{\mathrm{it}}=\ln \left({\displaystyle {\sum }_{\mathrm{r}=1}^{\mathrm{v}}{\mathrm{Advance}}_{\mathrm{r}}}+1\right)$$

(9)

In Equation (9), Quality_it represents the quality of digital technology patents in a city, v represents the total number of digital technology patents owned by the city, and Advance_r represents the technological advancement score of the patents. The technical advancement score is a comprehensive rating assigned by the incoPat platform to evaluate the technical advancement of a patent based on factors such as its global citation count, the number of IPC subgroups involved, the scale of R&D personnel investment, and whether licensing or transfer has occurred. The score ranges from 1 to 10, with a higher score indicating a higher technical level of the patent. Given that patent data typically exhibit a right-skewed distribution, we followed common practice by applying a logarithmic transformation after adding 1 to the patent data. If readers are interested in this indicator, they can gain a deeper understanding of this score through the information provided in Appendix A.

Based on this, an analysis of the mechanism of digital technology innovation is conducted, and

Table 10. Test results of the mechanism.

Variable	Quantity	Cor	Quality	Cor	Knowledge	Cor
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
DT	0.385 ***	0.173 ***	0.212 ***	0.195 ***	0.348 ***	0.188 ***
	(0.072)	(0.048)	(0.066)	(0.046)	(0.079)	(0.046)
Quantity		0.069 ***
		(0.014)
Quality				0.019 ***
				(0.006)
Knowledge						0.032 ***
						(0.008)
Constant	−1.004	−0.106	−2.631	−0.125	−4.393	−0.035
	(3.469)	(1.257)	(3.910)	(1.259)	(4.122)	(1.264)
Sobel test	0.027 ***		0.004 ***		0.011 ***
N	4573	4573	4573	4573	4573	4573
Adj R2	0.890	0.840	0.866	0.830	0.833	0.834
Control variables	YES	YES	YES	YES	YES	YES
City fixed effects	YES	YES	YES	YES	YES	YES
Fixed year effects	YES	YES	YES	YES	YES	YES

Note: *** indicates that the estimated coefficient is significant at the 1% level.

reports the test results of the digital technology innovation mechanism. The results of Model 1 and Model 3 indicate that the regression coefficient for the inflow of digital talent is significantly positive, suggesting that the inflow of digital talent not only significantly enhances the efficiency of digital technology innovation activities but also fundamentally optimizes the quality of digital technology innovation, thereby promoting a comprehensive improvement in the level of urban digital technology innovation.

The results of Model 2 and Model 4 indicate that the regression coefficients for both the quantity and quality of digital technology innovation are significantly positive at the 1% level, indicating that the widespread application of digital technology provides a wealth of digital tools and solutions for the manufacturing, prompting the boundaries between the digital economy industry and the manufacturing to become blurred, achieving a deeper integration between the two, and further accelerating the process of co-agglomeration between the digital economy industry and the manufacturing. Therefore, research hypothesis 2 is verified. In addition, to ensure the robustness of the results, this paper further conducts a Sobel test on this mediating effect, and the results indicate that the mediating effect is robust.

5.5.3. Mechanism of Digital Knowledge Spillover

Knowledge spillover refers to the process by which recipients acquire knowledge from providers and integrate it with their own knowledge to create new knowledge. Patents, as an important carrier of knowledge, exhibit knowledge spillover through their citation process. When a technological patent is frequently cited by subsequent patents, it indicates that the technological achievements represented by the patent have had a significant impact on subsequent technological innovations, facilitating the dissemination and diffusion of knowledge. Therefore, based on the digital technology patent data obtained previously and referencing the research by Chen and Liu [64], this study measures digital knowledge spillover using the citation count of digital technology patents. The specific measurement method is as follows:

$${\mathrm{Knowledge}}_{\mathrm{it}}=\ln \left({\displaystyle {\sum }_{\mathrm{r}=1}^{\mathrm{k}}{\mathrm{Reference}}_{\mathrm{r}}+1}\right)$$

(10)

In Equation (10), Knowledge_it represents the level of digital knowledge spillover in a city, and Reference_r denotes the number of citations for a patent. Due to the typical right-skewed nature of patent data, following common practice, the patent data are log-transformed after adding 1. Based on this, the mechanism of digital knowledge spillover is analyzed, and Table 10 reports the test results of digital knowledge spillover. The results of Model 5 indicate that the regression coefficient for the inflow of digital talent is significantly positive, suggesting that the inflow of digital talent not only increases the talent density in the inflow city but also facilitates the rapid dissemination and sharing of digital knowledge and technology among digital talents through enhanced communication and interaction.

The results of Model 6 show that the regression coefficient for digital knowledge spillover is positive at the 1% significance level, indicating that digital knowledge spillover not only promotes the effective integration of advanced digital knowledge into the manufacturing sector by the digital economy industry but also greatly stimulates collaboration between the two in technology research and development and product innovation, achieving optimal allocation and integration of advantageous resources between the digital economy industry and manufacturing. This efficient deep integration and collaboration further promote the spatial agglomeration of the digital economy industry and manufacturing, forming a mutually reinforcing development pattern. Therefore, research hypothesis 3 is verified. In addition, to ensure the robustness of the results, this paper further conducts a Sobel test on this mediating effect, and the results indicate that the mediating effect is robust.

5.5.4. Mechanism of Digital Knowledge Flow

As a direct product of technological innovation, patents contain rich technical knowledge and unique innovative ideas from inventors. When a patent is transferred from one party to another, it not only represents knowledge exchange and cooperation between two innovative entities but also signifies the flow and transfer of related technical knowledge. Through this transfer process, the recipient of the patent can deeply learn and master the technical knowledge contained in the patent, which may further trigger new technological innovations. This phenomenon of patent transfer is essentially an important manifestation of knowledge flow. Therefore, based on the digital technology patent data obtained previously and referencing the research by Ren et al. [44], this study measures digital knowledge flow using the number of transfers of digital technology patents.

The calculation process for digital technology patent transfers is mainly divided into the following steps: First, identify the legal status text of digital technology patents to determine whether the digital technology patent has undergone a transfer. Second, for digital technology patents that have been confirmed to have been transferred, identify the year of legal status change to clarify the specific time of the transfer activity. Third, analyze the address information of the assignee after the change in digital technology patents to determine the final destination city of the transfer and accurately count these data as the number of digital technology patents transferred into the city to comprehensively reflect the flow and agglomeration status of digital technology in the city. Similarly, due to the typical right-skewed nature of patent data, following common practice, the data are log-transformed after adding 1.

Based on this, the mechanism of digital knowledge flow is analyzed, and

Table 11. Test results of the mechanism.

Variable	Flow	Cor	Active	Cor	Upgrading	Cor	Cor
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model7
DT	0.502 ***	0.135 ***	0.009 ***	0.112 **	6.205 ***	0.197 ***	0.115 ***
	(0.050)	(0.045)	(0.001)	(0.053)	(1.814)	(0.045)	(0.037)
Flow		0.128 ***
		(0.014)
Active				9.771 ***
				(1.820)
Upgrading						0.000
						(0.000)
Policy							0.001
							(0.001)
DT × Policy							0.004 ***
							(0.001)
Constant	−3.966	0.334	−0.078 **	0.588	−361.713 **	−0.033	−0.123
	(2.452)	(1.193)	(0.037)	(1.360)	(151.022)	(1.219)	(1.249)
Sobel test	0.064 ***		0.087 ***		0.002 ***
N	4573	4573	4573	4573	4573	4573	4335
Adj R2	0.745	0.857	0.776	0.865	0.925	0.829	0.850
Control variables	YES	YES	YES	YES	YES	YES	YES
City fixed effects	YES	YES	YES	YES	YES	YES	YES
Fixed year effects	YES	YES	YES	YES	YES	YES	YES

Note: ** and *** respectively indicate that the estimated coefficients are significant at the 5% and 1% levels.

reports the test results of the digital knowledge flow mechanism. The results of Model 1 indicate that the regression coefficient for the inflow of digital talent is significantly positive, suggesting that the inflow of digital talent helps construct and expand the digital knowledge network of the inflow city. The inflow of digital talent can enhance knowledge connections among digital talents and effectively promote the transmission and flow of digital knowledge among different digital talents through deep exchanges and cooperation with other digital talents in the city, providing important knowledge support and innovation impetus for the development of the digital economy in the city.

The results of Model 2 show that the regression coefficient for digital knowledge flow is positive at the 1% significance level, indicating that the flow of digital knowledge enables the rapid dissemination of advanced digital technologies and knowledge from the digital economy industry to the manufacturing, helping manufacturing enterprises promptly apply cutting-edge technologies from the digital economy industry to enhance their production efficiency and product quality. At the same time, the demand from the manufacturing also drives continuous innovation in the digital economy industry, forming technological complementarity and integration, and further accelerating the co-agglomeration of the digital economy industry and manufacturing. Therefore, research hypothesis 3 is verified. In addition, to ensure the robustness of the results, this paper further conducts a Sobel test on this mediating effect, and the results indicate that the mediating effect is robust.

5.5.5. Mechanism of Entrepreneurial Activity in Digital Enterprises

New ventures are the direct outcome of entrepreneurial activities, and their quantity directly reflects the frequency and vigor of such activities. It also indirectly indicates the entrepreneurial environment and policy support of a city. An increase in the number of new ventures suggests that more entrepreneurs are engaging in the entrepreneurial wave, making entrepreneurial activities more vibrant. Therefore, referencing the research by Bai et al. [65], this study employs the number of newly established digital economy enterprises in a city to measure the entrepreneurial activity of digital economy enterprises in that city. Based on the previously obtained number of digital economy enterprises, this data is matched with the year and city of establishment according to the enterprises’ incorporation dates and addresses, thereby forming the indicator of entrepreneurial activity of digital economy enterprises for this research.

Based on this, an analysis of the mechanism of entrepreneurial activity in digital economy enterprises is conducted, and Table 11 reports the test results of the mechanism of entrepreneurial activity in digital economy enterprises. The results of Model 3 indicate that the regression coefficient for the inflow of digital talent is significantly positive, suggesting that digital talent not only plays a pivotal role in digital technology innovation and digital knowledge dissemination but also acts as entrepreneurs and investors, establishing high-value-added digital economy enterprises such as AI and big data in the cities they migrate to, significantly enhancing the entrepreneurial activity of the digital economy industry in these cities.

The results of Model 4 reveal that the regression coefficient for entrepreneurial activity in digital economy enterprises is positive at a 5% significance level, indicating that more entrepreneurial activities in digital economy enterprises inject continuous vitality into the entire digital economy industry, promoting its overall technological innovation capability and market competitiveness. This enables the digital economy industry to have a stronger capability to empower manufacturing enterprises, assisting them in achieving digital and intelligent transformation, thereby deepening the integration between the digital economy industry and manufacturing, and jointly driving high-quality economic development. Therefore, research hypothesis 4 is verified. Additionally, to ensure the robustness of the results, this paper further conducts a Sobel test on this mediating effect, and the results indicate that the mediating effect is robust.

5.5.6. Mechanism of Urban Industrial Structure Upgrading

Industrial structure upgrading refers to the process of a region’s industrial structure developing towards a higher level and becoming more rational. Referencing the research by Wu and Chen [66], this study adopts the following measurement method to calculate urban industrial structure upgrading:

$${\mathrm{Upgrading}}_{\mathrm{it}}={\displaystyle {\sum }_{\mathrm{n}}^3\left[-100\times \frac{{\mathrm{Value}}_{\mathrm{itn}}}{{\mathrm{Value}}_{\mathrm{it}}}\times \ln \left(\frac{{\mathrm{Value}}_{\mathrm{itn}}}{{\mathrm{Labor}}_{\mathrm{itn}}}/\frac{{\mathrm{Value}}_{\mathrm{it}}}{{\mathrm{Labor}}_{\mathrm{it}}}\right)\right]}$$

(11)

In Equation (11), Upgrading_it represents the indicator of urban industrial structure upgrading. Industry is classified according to the three-sector divisions. Value_itn denotes the value added of industry n in the city. Value_it represents the total output value of all industries in the city. Labor_itn indicates the number of employees in industry n in the city. and Labor_it represents the total number of employees in all industries in the city. The data used to calculate these variables are all derived from the China City Statistical Yearbook.

Based on this, the mechanism of urban industrial structure upgrading is analyzed, and Table 11 reports the test results of the urban industrial structure upgrading mechanism. The results of Model 5 indicate that the regression coefficient for the inflow of digital talent is significantly positive, suggesting that the inflow of digital talent significantly optimizes the urban industrial structure, not only promoting the vigorous development of high-tech industries such as software development, data services, and internet services, but also strongly driving the transformation and upgrading of the urban industrial structure towards high-tech content and high-value-added directions, achieving advanced and rationalized industrial structure.

The results of Model 6 show that the regression coefficient for industrial structure upgrading is not significant; therefore, this study further conducts a Sobel test. The Sobel test results indicate that industrial structure upgrading can effectively promote the co-agglomeration of the digital economy industry and manufacturing, indicating that during the process of industrial structure upgrading, it will drive traditional manufacturing towards high-value-added directions. This process stimulates a stronger demand for digital technology, thereby further expanding the demand for digital technology in the digital economy industry market. This growth in technology demand, in turn, attracts more digital economy enterprises to agglomerate in the city, promotes the formation of a close cooperative relationship between the digital economy industry and manufacturing, and enhances the synergistic effect between the digital economy industry and manufacturing. Therefore, research hypothesis 5 is verified.

5.5.7. Mechanism of Digital Economy Policy

The government work report is an important document that summarizes the government’s work over the past year and deploys its work plan for the coming year, possessing high authority. The keywords mentioned in the report often reflect the focus and orientation of regional policies, which are significant for understanding and predicting the development trends of regional industries. Therefore, referring to the research by Jin et al. [67], this study adopts text analysis to analyze regional government work reports, thereby constructing an indicator for urban digital economy policy. The specific calculation steps are as follows: First, we constructed a digital economy policy keyword library, with keywords shown in

Table 12. Keywords of digital economy policy.

Indicator

Keywords of Digital Economy Policy

Digital Economy Policy

Digital economy, Intelligent economy, Information economy, Knowledge economy, Smart Economy, Digitalized information, Modern information network, ICT, Communication infrastructure, internet, Cloud computing, Blockchain, IoT, Digitization, Digital village, Digital industry, E-commerce, 5G, Digital infrastructure, Artificial intelligence, Electronic commerce, Big data, Datafication, Industrial digitization, Digital industrialization, Data capitalization, Smart city, Cloud service, Cloud technology, Cloud platform, E-government, Mobile payment, Online, Information industry, Software, Information Infrastructure, Information technology, Digital life.

:

Secondly, this study collected 4335 prefecture-level government work report documents from 2006 to 2022 by accessing the official websites of various prefecture-level cities. Thirdly, this study utilized the Jieba library in Python software for word segmentation. Subsequently, based on the pre-established digital economy policy vocabulary in Table 9, we conducted searches, matching, and statistical analysis of the frequency of characteristic words. Finally, this study aggregated the word frequencies of 39 keywords related to digital economy policies to construct an index that quantitatively reflects the intensity of digital economy policies in various prefecture-level cities.

Based on this, we analyzed the mechanism of digital economy policies. Model 7 in Table 11 reports the test results of the moderating mechanism of digital economy policies. The results of the moderating mechanism indicate that the interaction term between digital economy policies and the inflow of digital talent is positive at a significance level of 1%, suggesting that in regions with a superior digital economy policy environment, local governments can formulate and implement attractive talent introduction policies. These policies not only effectively lower the barriers and costs for the inflow of digital talent but also significantly enhance the willingness and motivation of digital talent to choose to move to these cities, achieving optimal allocation of talent resources and laying a solid foundation of talent for the coordinated development of the digital economy and manufacturing. Digital economy policies guide digital talent to incline towards the development of the digital economy industry and manufacturing sector, which not only promotes precise matching between digital talent and the demands of the digital economy industry and manufacturing but also enables digital talent to play a greater role in promoting industrial co-agglomeration, becoming a significant driving force for industrial upgrading and innovative development. Therefore, research hypothesis 6 is verified.

5.6. Heterogeneity Test

5.6.1. Heterogeneity in Manufacturing Levels

Based on the “Classification of High-tech Industries (Manufacturing) (2017)” document published by the National Bureau of Statistics [68], this study defines pharmaceutical manufacturing, medical equipment and instrument manufacturing, aerospace manufacturing, instrument, and meter manufacturing, as well as chemical raw material and chemical product manufacturing as high-end manufacturing, while classifying the remaining manufacturing as low-to-mid-end manufacturing. This classification is used to explore the heterogeneous impacts of digital talent inflow on the co-agglomeration between the digital economy industry and high-end manufacturing, as well as between the digital economy industry and low-to-mid-end manufacturing. Model 1 in

Table 13. Test results of heterogeneity.

Variable	Cor
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
DT	0.089 ***	0.185 ***	0.247 ***	0.014	0.216 ***	0.035 *	0.047	0.240 ***
	(0.034)	(0.044)	(0.065)	(0.017)	(0.057)	(0.019)	(0.036)	(0.058)
p-value of inter group coefficient	Not involving grouping		0.067		0.003		0.005
Constant	−0.647	0.456	−3.542	0.436	−3.375	2.071 ***	1.094	−1.426
	(1.285)	(1.242)	(3.573)	(0.631)	(4.210)	(0.680)	(1.021)	(2.808)
N	4573	4573	2099	2224	1751	2652	1751	2652
Adj R2	0.704	0.821	0.878	0.904	0.864	0.871	0.864	0.871
Control variables	YES	YES	YES	YES	YES	YES	YES	YES
City fixed effects	YES	YES	YES	YES	YES	YES	YES	YES
Fixed year effects	YES	YES	YES	YES	YES	YES	YES	YES

Note: * and *** respectively indicate that the estimated coefficients are significant at the 10% and 1% levels.

estimates the impact of digital talent inflow on the co-agglomeration between the digital economy industry and high-end manufacturing, while Model 2 estimates its impact on the co-agglomeration between the digital economy industry and low-to-mid-end manufacturing.

The results of the heterogeneity test indicate that the inflow of digital talent not only promotes the co-agglomeration between the digital economy industry and high-end manufacturing but also enhances the co-agglomeration effect between the digital economy industry and low-to-mid-end manufacturing, with a more significant synergistic effect observed in the low-to-mid-end manufacturing sector. This may be attributed to the fact that low-to-mid-end manufacturing, compared to high-end manufacturing, generally has lower technological content and a lower threshold for digital transformation. The basic digital transformation measures brought by the inflow of digital talent, such as the introduction of automated equipment and the application of basic data analysis tools, can quickly demonstrate results in low-to-mid-end manufacturing, significantly improving production efficiency and reducing costs without requiring excessive technological thresholds, thereby enhancing market competitiveness and achieving effective empowerment of the digital economy industry over low-to-mid-end manufacturing. Therefore, low-to-mid-end manufacturing often benefits more rapidly from the application of digital technologies, making the synergistic effect of digital talent inflow on the digital economy industry and low-to-mid-end manufacturing more pronounced.

Compared to low-to-mid-end manufacturing, high-end manufacturing has already achieved a certain level of digital transformation due to its strong independent innovation capability. The inflow of digital talent can undoubtedly bring digital technology empowerment to high-end manufacturing and promote its digital transformation and upgrading, but this process faces greater challenges and higher cost investments. Although it has a significant positive promoting effect, the speed of its effectiveness is relatively slower. Furthermore, high-end manufacturing not only relies on the empowerment of digital talent but also requires higher-level technology and innovative talent support from other industries, and not all digital talent can meet its specific needs. At the same time, high-end manufacturing often involves more research and development, innovation, and complex industrial chains, requiring deeper technological cooperation with other industries. Therefore, when making location decisions, high-end manufacturing needs to consider more the co-agglomeration with closely related industries rather than just the digital economy industry. Consequently, compared to low-to-mid-end manufacturing, the synergistic effect of digital talent inflow on the digital economy industry and high-end manufacturing is slightly weaker.

5.6.2. Heterogeneity of Digital Infrastructure

This study refers to the research conducted by Yin et al. [69] and comprehensively considers three indicators: per capita internet broadband subscription users, per capita mobile phone subscriptions, and per capita total telecommunications business volume, to measure the level of digital infrastructure in cities. The entropy method is used to calculate the level of digital infrastructure in cities. All data used to calculate the level of digital infrastructure are derived from China City Statistical Yearbook. On this basis, this study further divides cities into two groups based on the median level of digital infrastructure: one group with relatively well-developed digital infrastructure and the other with relatively weak digital infrastructure. Model 3 in Table 13 estimates the impact of digital talent inflow on industrial co-agglomeration in cities with well-developed digital infrastructure, while Model 4 estimates the impact on cities with relatively weak digital infrastructure.

Heterogeneity test results indicate that the inflow of digital talent has a significant impact only on the co-agglomeration of the digital economy industry and manufacturing in cities with well-developed digital infrastructure, but not in cities with relatively weak digital infrastructure. At the same time, the significance test of the coefficient of variation between groups is also positive at the 10% level, indicating that there are substantial differences in the co-agglomeration effects of digital talent inflow on the digital economy industry and manufacturing across cities with different digital infrastructure levels, with a significant promoting effect only in cities with well-developed digital technology infrastructure. The reasons may be as follows: In cities with weak digital infrastructure, the lack of information technology becomes a bottleneck restricting the work efficiency and innovation ability of digital talent, making it difficult for them to fully leverage their professional advantages.

At the same time, as data are the core driving force of the digital economy, its circulation and application in these cities face more obstacles, increasing the difficulty of information exchange among digital talent and limiting their ability to acquire and utilize high-quality data resources, thereby reducing their potential to promote the collaborative development of the digital economy industry and manufacturing. In addition, weak digital infrastructure also leads to difficulties in attracting and retaining digital talent in cities. Digital talent tends to prefer cities with superior digital infrastructure and higher quality of life, further exacerbating the outflow of digital talent from these weak cities.

In contrast, cities with well-developed digital infrastructure provide a better working environment for digital talent. Strong information technology support enables digital talent to work more efficiently. At the same time, the smooth circulation and application of data between cities provide rich data resources for digital talent, enabling them to conduct deeper analysis and mine new data resources. Furthermore, cities with well-developed digital infrastructure not only have advantages in the working environment but also exhibit stronger capabilities in attracting and retaining digital talent, making the inflow of digital talent significantly promote the co-agglomeration and development of the digital economy industry and manufacturing in cities. Therefore, due to technical limitations, obstacles to data circulation, and insufficient talent attraction in cities with weak digital infrastructure, the inflow of digital talent does not have a significant impact on the co-agglomeration of their digital economy industry and manufacturing. In contrast, cities with well-developed digital infrastructure, relying on strong technical support, smooth data circulation, and strong talent attraction, enable the inflow of digital talent to significantly promote the co-agglomeration and development of their digital economy industry and manufacturing.

5.6.3. Heterogeneity of City Levels

This study refers to the “2024 City Commercial Charm Ranking“ released by the New First-Tier City Research Institute and divides the sample cities into two groups [70]: one group of high-grade cities, including first-tier, second-tier, and third-tier cities, and the other group of low-grade cities, including fourth-tier and fifth-tier cities. Model 5 in Table 13 estimates the impact of digital talent inflow on industrial co-agglomeration in high-grade cities, while Model 6 estimates the impact on low-grade cities.

Heterogeneity test results indicate that the regression coefficient for digital talent inflow in high-grade cities is positive at the 1% significance level, while the regression coefficient for digital talent inflow in low-grade cities shows a positive relationship at the 10% significance level. At the same time, the significance test of the coefficient of variation between groups is also positive at the 1% level, indicating that there are substantial differences in the co-agglomeration effects of digital talent inflow on the digital economy industry and manufacturing across cities of different grade, with a particularly significant promoting effect on high-grade cities. The reasons may be as follows: From the perspective of market demand, high-grade cities provide broad development space for the digital industry and manufacturing with their large market demand and strong consumption capacity.

In contrast, low-grade cities have limited market size, weak consumption capacity, and relatively low demand for high-end digital products, making it difficult to form a significant demand-driven effect. From the perspective of industrial policy support and resources, high-grade city governments have abundant public funds and can provide more industrial policy support and incentives. At the same time, they have relatively concentrated financial resources, providing solid financial support for the collaborative development of the digital industry and manufacturing. Low-grade cities, on the other hand, are relatively lacking in industrial policy support and financial resources, making it difficult to attract and retain digital economy and manufacturing enterprises.

From the perspective of industrial supporting measures, high-grade cities have numerous industrial parks, which effectively promote the deep integration and co-agglomeration between the digital economy industry and manufacturing. Conversely, low-grade cities have a relatively weak industrial foundation and incomplete industrial chains, making it difficult to form effective industrial synergies. From the perspective of employment opportunities and salary levels, high-grade cities provide more employment opportunities and higher salary levels for digital talent, along with well-developed urban infrastructure and superior environment, making them more attractive. Low-grade cities, on the other hand, have limited employment opportunities and relatively low salary levels, making them less attractive to high-level digital talent. Therefore, high-grade cities have significant advantages in market demand, policy support, industrial supporting measures, and city attractiveness, making digital talent more inclined to flow into high-grade cities, and digital economy and manufacturing enterprises more willing to establish factories there. Therefore, the impact of digital talent inflow on industrial co-agglomeration in high-grade cities is more significant.

5.6.4. Heterogeneity of Economic Structure

Based on the “National Sustainable Development Plan for Resource-Based Cities (2013–2020)” issued by the State Council [71], this study divides the sample cities into two groups: resource-based cities and non-resource-based cities. Resource-based cities are typically rich in mineral resources such as coal, oil, and metal minerals, with their economic development highly dependent on the exploitation, processing, and utilization of natural resources. On the other hand, non-resource-based cities do not rely on the extraction and processing of specific natural resources, and their industrial structures are relatively balanced. Model 7 in Table 13 estimates the impact of digital talent inflow on industrial co-agglomeration in resource-based cities, while Model 8 estimates the impact on non-resource-based cities.

Heterogeneity test results indicate that the regression coefficient for digital talent inflow in non-resource-based cities is positive at a 1% significance level, while the coefficient for resource-based cities is not significant. Simultaneously, the significance test for the coefficient of inter-group differences is also positive at a 1% level, indicating substantial differences in the co-agglomeration effect of digital talent inflow on the digital economy industry and manufacturing in cities with different economic structures. The inflow of digital talent only significantly promotes this effect in non-resource-based cities. The reasons may be as follows: On the one hand, the industrial structure of resource-based cities is relatively homogeneous, mainly relying on resource extraction and primary processing industries, leading to lower demand for digital products and a lack of conditions necessary to attract digital talent and establish digital economy industries. In contrast, non-resource-based cities exhibit a more diversified economic structure, with longer industrial chains and complete supporting industries, resulting in a stronger market demand for digital products, which plays a decisive role in the location choice of digital industries. At the same time, a diversified industrial structure and a complete industrial chain provide more development opportunities, attracting more digital talent.

On the other hand, due to their heavy reliance on resource extraction and primary processing industries, resource-based cities may be more inclined to invest funds in resource development and related infrastructure during their urban development, while investment in digital infrastructure is relatively inadequate. Long-term dependence on resource-based industries may also lead to a relatively weak innovation atmosphere in these cities. Conversely, non-resource-based cities vigorously promote the development of the digital economy, investing heavily in digital infrastructure to create a superior working environment for digital talent, thereby attracting more digital talent to cluster in these cities. Furthermore, non-resource-based cities possess a strong innovation atmosphere and abundant innovation resources, capable of providing a high-quality innovation and entrepreneurship environment for digital talent. Therefore, digital talent mainly has a significant impact on the co-agglomeration of the digital economy industry and manufacturing in non-resource-based cities but does not exhibit a significant impact in resource-based cities.

5.7. Spatial Spillover Effect Analysis

5.7.1. Spatial Econometric Model and Spatial Weight Matrix Settings

To examine whether the impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing exhibits a spatial spillover effect, this study establishes a spatial Durbin model (SDM) with two-way fixed effects based on Equation (12) as follows:

$$\begin{array}{c}{\mathrm{Cor}}_{\mathrm{it}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{DT}}_{\mathrm{it}}+{\mathsf{\beta }}_2{\mathrm{Control}}_{\mathrm{it}}+{\mathsf{\beta }}_3{\mathrm{W}\times \mathrm{DT}}_{\mathrm{it}}+{\mathsf{\beta }}_4{\mathrm{W}\times \mathrm{Cor}}_{\mathrm{it}}+{\mathsf{\beta }}_5{\mathrm{W}\times \mathrm{Control}}_{\mathrm{it}}\\ +{\mathsf{\beta }}_6\mathrm{Year}+{\mathsf{\beta }}_7\mathrm{City}+{\mathsf{\epsilon }}_{\mathrm{it}}\end{array}$$

(12)

In Equation (12), W represents the spatial weight matrix. In this study, three matrices, namely, economic distance, geographical adjacency, and geographical distance, are selected to examine the spatial spillover effects of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing.

5.7.2. Spatial Autocorrelation Test

Before conducting spatial econometric analysis, it is necessary to perform a spatial autocorrelation test. This paper employs the global Moran’s I index to conduct a spatial autocorrelation test on the spatial agglomeration of the co-agglomeration of the digital economy industry and manufacturing.

Table 14. Calculation results of global Moran’s I index.

Year	Global Moran’s I Index
	Digital Talent Inflow			Co-Agglomeration of the Digital Economy Industry and Manufacturing
	Economic Matrix	Adjacency Matrix	Distance Matrix	Economic Matrix	Adjacency Matrix	Distance Matrix
2006	0.158 ***	0.167 ***	−0.018 ***	0.094 ***	0.453 ***	−0.074 ***
2007	0.163 ***	0.169 ***	−0.015 ***	0.093 ***	0.432 ***	−0.079 ***
2008	0.161 ***	0.176 ***	−0.014 ***	0.089 ***	0.387 ***	−0.069 ***
2009	0.158 ***	0.158 ***	−0.014 ***	0.073 **	0.343 ***	−0.059 ***
2010	0.182 ***	0.148 ***	−0.015 ***	0.044	0.304 ***	−0.051 ***
2011	0.188 ***	0.135 ***	−0.015 ***	0.031	0.252 ***	−0.042 ***
2012	0.171 ***	0.125 ***	−0.013 ***	0.038	0.210 ***	−0.033 ***
2013	0.219 ***	0.150 ***	−0.015 ***	0.054 *	0.184 ***	−0.027 ***
2014	0.237 ***	0.164 ***	−0.016 ***	0.070 **	0.162 ***	−0.022 ***
2015	0.230 ***	0.148 ***	−0.015 ***	0.085 ***	0.151 ***	−0.018 ***
2016	0.223 ***	0.151 ***	−0.015 ***	0.097 ***	0.152 ***	−0.015 ***
2017	0.215 ***	0.181 ***	−0.014 ***	0.111 ***	0.165 ***	−0.013 ***
2018	0.224 ***	0.120 ***	−0.013 ***	0.122 ***	0.175 ***	−0.011 ***
2019	0.227 ***	0.144 ***	−0.014 ***	0.130 ***	0.185 ***	−0.010 ***
2020	0.229 ***	0.151 ***	−0.013 ***	0.132 ***	0.208 ***	−0.010 ***
2021	0.222 ***	0.166 ***	−0.014 ***	0.135 ***	0.227 ***	−0.012 ***
2022	0.222 ***	0.181 ***	−0.015 ***	0.132 ***	0.236 ***	−0.012 ***

Note: *, **, and ***, respectively, indicate that the estimated coefficients are significant at the 10%, 5%, and 1% levels.

reports the calculation results of the global Moran’s I index. Among the three spatial matrices of economic distance, geographical adjacency, and geographical distance, the global Moran’s I indices for digital talent inflow and the co-agglomeration of the digital economy industry and manufacturing are significant in most years, indicating a significant positive correlation in space. Therefore, spatial econometrics can be used to explore spatial spillover effects.

5.7.3. Results of Spatial Spillover Effect Tests

Table 15. Test results of spatial spillover effect.

Variable	Cor
	Direct Effect	Overflow Effect	Direct Effect	Overflow Effect	Direct Effect	Overflow Effect
	Model 1		Model 2		Model 3
DT	0.181 ***	0.050 ***	0.186 ***	0.082 ***	0.195 ***	−0.277 **
	(0.004)	(0.014)	(0.004)	(0.011)	(0.004)	(0.124)
W×Cor		0.212 ***		0.447 ***		−0.906 ***
		(0.028)		(0.017)		(0.290)
sigma2_e	0.024 ***		0.021 ***		0.025 ***
	(0.001)		(0.000)		(0.001)
N	4573		4573		4573
R2	0.117		0.411		0.434
Control variables	YES		YES		YES
City fixed effects	YES		YES		YES
Fixed year effects	YES		YES		YES

Note: ** and *** respectively indicate that the estimated coefficients are significant at the 5% and 1% levels.

reports the test results of the spatial spillover effects of digital talent on the co-agglomeration of the digital economy industry and manufacturing. Specifically, Model 1 to Model 3 present the results of direct spillover effects and spatial spillover effects under the SDM model in the economic distance matrix, geographical adjacency matrix, and geographical distance matrix, respectively. The results of direct spillover effects indicate that the regression coefficients of digital talent inflow in the economic distance matrix, geographical adjacency matrix, and geographical distance matrix are significantly positive, suggesting that the inflow of digital talent can significantly promote the co-agglomeration of the local digital economy industry and manufacturing, which is consistent with the previous benchmark regression results.

The results of spatial spillover effects show that the inflow of digital talent has a significant promoting effect on the inflow of digital talent in cities with similar levels of economic development and adjacent geographical locations, demonstrating a positive diffusion effect. However, for cities with greater geographical distances, a negative siphon effect is observed. Furthermore, the spatial spillover effects of the co-agglomeration of the digital economy industry and manufacturing exhibit similar results, with a significant promoting effect on cities with similar levels of economic development and adjacent geographical locations, and a negative siphon effect on cities with greater geographical distances.

The possible reason is that the inflow of digital knowledge from digital talent and its knowledge spillover effects can transcend geographical and economic boundaries to propagate. This not only directly enhances the digital technology innovation capability of the host city but also powerfully promotes industrial co-agglomeration in neighboring areas through multiple mechanisms such as knowledge spillovers and technology transfer. However, for cities that are farther away, the inflow of digital talent into other cities and the co-agglomeration of the digital economy and manufacturing in economically developed areas may actually create policy lows, attracting more digital talent and industries to concentrate in cities with higher levels of economic development. This creates a competitive relationship with distant cities, resulting in unfavorable spatial spillover effects on these cities.

6. Discussion and Practical Implications

Different from existing research, this study adopts the number of urban digital economy enterprises and manufacturing enterprises as indicators to measure industrial agglomeration. It overcomes the issues of incomplete measurement, data scarcity, and double counting between the digital economy and manufacturing in previous studies, providing a novel data source for the calculation of the Θ index. This approach transcends the limitations of traditional data on output value, main business income, and employment, offering a fresh perspective and method for studying industrial agglomeration.

Furthermore, while previous research on the co-agglomeration of the digital economy and manufacturing primarily focused on its impact on green innovation in enterprises, this study shifts the focus to the factors influencing the co-agglomeration of the digital economy and manufacturing. By incorporating the inflow of digital talent and the co-agglomeration of the digital economy and manufacturing into a unified analytical framework, this study empirically examines the relationship between the inflow of digital talent and the co-agglomeration of the digital economy and manufacturing, providing new perspectives and evidence in this field. Additionally, this study proposes multiple research hypotheses and constructs several mechanistic models to explore the pathways through which the inflow of digital talent affects the co-agglomeration of the digital economy and manufacturing. These models exhibit statistically significant results, indicating the following:

The inflow of digital talent has a significant positive impact on digital technology innovation, further validating the importance of digital talent as a source of innovation and their crucial role in promoting the co-agglomeration of the digital economy industry and manufacturing. This finding highlights the significance of digital talent in technological innovation and provides a clear guidance direction for policymakers, namely, they should focus on the cultivation and introduction of digital talent to facilitate technological innovation and industrial upgrading.

The promoting effect of digital talent inflow on digital knowledge spillover and flow is also statistically significantly supported, indicating that the flow of digital talent brings not only technology but also knowledge and experience. The spillover and flow of the knowledge and experience among industries have a significant impact on promoting the co-agglomeration of the digital economy industry and manufacturing. This finding emphasizes the importance of digital talent in knowledge dissemination and exchange, providing theoretical support for building a more open and collaborative innovation ecosystem.

The inflow of digital talents has a significant impact on enhancing the entrepreneurial activity of digital economy enterprises in cities, further demonstrating the crucial role of digital talents in driving urban economic vitality and innovation, as well as their contribution to the co-agglomeration of the digital economy industry and manufacturing. This finding encourages policymakers to optimize the entrepreneurial environment to attract and retain digital talents.

The inflow of digital talents also has a statistically significant role in promoting the upgrading of urban industrial structures, indicating that the inflow of digital talents not only propels the development of the digital economy but also facilitates the overall upgrading of urban industrial structures, enabling a closer integration between the digital economy industry and manufacturing. This finding suggests that governments can formulate and implement more targeted policies to attract and retain digital talents, thereby promoting the upgrading of urban industrial structures and economic development.

The positive moderating effect of digital economic policies on the relationship between the inflow of digital talents and the co-agglomeration of the digital economy industry and manufacturing is also significantly reflected in the model, which provides a new perspective for us to understand the important role of the policy environment in both the inflow of digital talents and the exertion of their effects. It indicates that policies play a crucial role in guiding and supporting the inflow of digital talents. This finding provides a useful reference for policymakers, suggesting that more scientific and effective policies should be formulated to promote the mobility and agglomeration of digital talents.

The statistical significance of these models not only validates the theoretical hypotheses of this study but also provides strong evidence for our deeper understanding of the impact of the inflow of digital talents on the co-agglomeration of the digital economy industry and manufacturing. Together, they constitute a comprehensive and multi-dimensional analytical framework, which helps us better grasp the essence and laws of this complex phenomenon. Meanwhile, the examination of these mechanisms can provide policymakers with a theoretical basis for formulating industrial policies, optimizing talent cultivation and introduction strategies, and ultimately promoting regional industrial development, which is of significant theoretical and practical importance.

7. Conclusions and Recommendations

7.1. Conclusions

This study innovatively incorporates the inflow of digital talent and the co-agglomeration of the digital economy industry and manufacturing into a unified analytical framework, providing a new perspective for understanding collaborative industrial development in the digital economy era. At the theoretical level, this study systematically analyzes the specific mechanisms of influence between the two. At the measurement level, this study selects sample data from 269 cities between 2006 and 2022 to calculate the inflow of digital talent and the degree of co-agglomeration between the digital economy industry and manufacturing and analyzes their dynamic changes at different stages of development and in geographical spaces. At the empirical level, this study further tests the relationship and specific mechanisms of action between the inflow of digital talent and the co-agglomeration of the digital economy industry and manufacturing. The main findings of this study are as follows:

Firstly, from the measurement results, the level of digital talent inflow in Chinese cities has continued to rise, with a significant increase in the number of cities with high and relatively high inflow levels, reflecting China’s high attention and significant investment in the development of the digital economy. However, most cities still face difficulties in policy support and resource investment. The level of co-agglomeration between the digital economy industry and manufacturing in Chinese cities has also shown a significant growth trend. The number of cities with high and relatively high levels of co-agglomeration has increased significantly. From a spatial dimension, industrial co-agglomeration exhibits characteristics of “core–periphery”, “multi-core agglomeration”, and “gradient diffusion” coexisting. Overall, the spatial distribution of China’s digital economy industry co-agglomeration level is still uneven, but the improvement in the central and western regions and the gradual transfer of industries are alleviating this issue, making the spatial distribution of industries more reasonable.

Secondly, from the empirical analysis, the inflow of digital talent has a significant positive effect on the co-agglomeration of the digital economy industry and manufacturing. This conclusion remains robust after a series of robustness tests. The analysis of intermediary mechanisms indicates that the inflow of digital talent promotes the co-agglomeration of the digital economy industry and manufacturing through multiple pathways, specifically including enhancing the level of digital technology innovation in inflow cities, promoting the spillover and flow of digital knowledge, increasing the entrepreneurial activity of urban digital economy enterprises, and facilitating the upgrading of urban industrial structures. The analysis of moderating mechanisms further suggests that the impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing is moderated by the intensity of urban digital economy policies. More favorable digital economy policies can significantly enhance this impact, thereby more fully leveraging the role of digital talent.

Thirdly, the results of heterogeneity analysis show that the promotional effect of digital talent inflow is more significant in low- and mid-end manufacturing, high-grade cities, cities with well-developed digital infrastructure, and non-resource-based cities. By contrast, this effect is relatively weaker in high-end manufacturing and low-grade cities. In cities with weak digital infrastructure and resource-based cities, the effect of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing is not significant. The analysis of spatial spillover effects indicates that the inflow of digital talent and the co-agglomeration of the digital economy industry and manufacturing have a significant promotional effect on cities with similar levels of economic development and adjacent geographical locations. This promotional effect drives industrial co-agglomeration in neighboring areas through mechanisms such as knowledge spillover and technology transfer, reflecting a positive diffusion effect. However, for cities that are geographically distant, a negative siphon effect is observed.

7.2. Recommendations

Based on the above conclusions, this study proposes the following policy recommendations:

Firstly, strengthen regional policies for attracting digital talent and promote digital skill enhancement initiatives. Given the pivotal role of digital talent in fostering the co-agglomeration of the digital economy and manufacturing, and the current challenge of low digital talent inflow in central and western regions, local governments, particularly in these areas, should introduce a series of competitive talent attraction policies, such as offering generous housing subsidies and remuneration packages, to enhance the regional appeal to digital talent. Simultaneously, actively implement the “Action Plan for Accelerating Digital Talent Development to Support the Development of the Digital Economy (2024–2026)” and advance regional digital skill enhancement efforts. Strengthen deep cooperation between vocational colleges and enterprises, jointly explore and develop digital vocational training resources, promote the innovation and optimization of university curricula, ensure a smooth alignment between talent development pathways and actual enterprise needs, and cultivate a cohort of innovative, practical, and composite digital talent with expertise in both industrial and digital technologies.

By implementing targeted training programs and promoting deep integration of industry, academia, and research, precisely match the talent needs of enterprises’ digital transformation, and supply enterprises with digital talent that meets development requirements, thereby accelerating the process of enterprises’ digital and intelligent transformation, further promoting the optimization and upgrading of regional industrial structures, and establishing a mutually beneficial and mutually reinforcing virtuous cycle between vocational colleges and enterprises. Additionally, efforts should be made to facilitate the digital skill development of existing employees within enterprises, encourage local enterprises to establish customized training programs based on demand, and cultivate composite talent. At the same time, fully leverage high-quality educational resources both domestically and internationally to launch advanced training programs for high-level digital talent, providing solid digital talent support for enterprises’ digital transformation.

Secondly, refine the digital economy policy system and strengthen the guiding role of policies. Given that digital economy policies can significantly enhance the impact of digital talent inflow on the co-agglomeration of the digital economy industry and manufacturing, the government should continuously deepen reforms of the digital economy policy system, ensuring that each policy is closely aligned with the core needs of the digital economy and is both targeted and effective. In attracting digital talent, a deep analysis of their career development and living environment needs should be conducted, and a series of highly attractive policy measures should be formulated to precisely guide the convergence of digital talent to the region, laying a solid talent foundation for the co-agglomeration of the digital economy and manufacturing.

At the same time, the government needs to increase support for digital technology innovation activities, provide strong financial support for the research and development of key digital core technologies through the establishment of special funds and the guidance of social capital investment and other diversified financing methods, and stimulate enterprises’ innovation vitality. Furthermore, the government should actively establish a platform for inter-regional exchange between digital talent and enterprises; encourage close cross-regional cooperation and exchange; promote the spillover, flow, and diffusion of digital knowledge and technology; stimulate urban innovative thinking and creative vitality; and fully leverage the core role of digital talent in driving the co-agglomeration of the digital economy and manufacturing. Finally, in the realm of innovation and entrepreneurship, the government should vigorously encourage innovation and entrepreneurship activities in the digital economy sector; establish a batch of high-quality entrepreneurial carriers and incubators; provide comprehensive support and services for startups in the digital economy; assist them in growing into specialized, sophisticated, innovative, “little giant“ enterprises in the digital economy sector; and inject robust innovative vitality and growth potential into the region’s digital economy development.

Thirdly, implement precise policies and promote the precise implementation of differentiated development strategies. Given that the co-agglomeration effect of digital talent flow on the digital economy and manufacturing exhibits different intensities under varying urban manufacturing levels, digital infrastructure, city grades, and economic structures, the government should adhere to the principle of tailoring strategies to local conditions and formulate differentiated development strategies. For high-grade cities, cities with well-established digital infrastructure, and non-resource-based cities, the potential of existing digital infrastructure should be deeply tapped and leveraged; close cooperation between universities, research institutions, and enterprises should be established; an ecosystem of deep industry–academia–research integration should be constructed; the iteration of digital technology research and development and the transformation of achievements should be accelerated; and the positive effects of the co-agglomeration of the digital economy and manufacturing should be maximized.

For low-grade cities, cities with weak digital infrastructure, and resource-based cities, simultaneous attention should be given to enhancing city grades, improving digital infrastructure, and optimizing economic structures, creating more favorable conditions for the inflow and effectiveness of digital talent. Low-grade cities should focus on optimizing the business environment, stimulating the vitality of social capital, constructing a diversified financing mechanism guided by the government, led by enterprises, involving society, and injecting robust impetus into the integration of the digital economy and manufacturing. For cities with weak digital infrastructure, fiscal investment should be resolutely increased, the layout of digital infrastructure should be accelerated, and a solid foundation should be laid for the synergistic development of the digital economy and manufacturing. Resource-based cities need to seek new transformation paths, actively attract digital economy enterprises to settle in, promote the intelligent transformation of traditional industries, cultivate new growth points in the digital economy, and realize the optimization and upgrading of economic structures.