How Agricultural Innovation Talents Influence County-Level Industrial Structure Upgrading: A Knowledge-Empowerment Perspective

Abstract

Upgrading the industrial structure is an essential step for economic growth and the transformation of old and new development drivers. Counties situated at the rural–urban interface hold a comparative advantage in industrial upgrading compared to cities, converting agricultural resource dividends into economic value. However, whether agricultural innovation talent can facilitate this process requires further investigation. Based on a sample of 1771 Chinese counties, this study employs a quasi-natural experiment using China’s “World-Class Disciplines” construction program in agriculture and establishes a difference-in-differences (DID) model to examine the impact of agricultural innovation talent on county-level industrial structure upgrading. The results show that agricultural innovation talent significantly promotes industrial upgrading, with this effect being more pronounced in counties with smaller urban–rural income gaps, greater household savings, and higher levels of industrial sophistication. Spatial spillover effects are also evident, indicating regional knowledge diffusion. Knowledge empowerment emerges as the core mechanism: agricultural innovation talent drives industrial convergence, responds to supply–demand dynamics, and integrates digital and intelligent elements through knowledge creation, dissemination, and application, thereby supporting county-level industrial upgrading. The findings highlight the necessity of establishing world-class agricultural research and talent incubation platforms, particularly emphasizing the supportive role of universities and the knowledge-driven contributions of agricultural innovation talents to county development.

1. Introduction

The upgrading of industrial structures plays a significant role in promoting national economic growth and enhancing residents’ well-being [1,2]. While prior studies have extensively discussed the implementation pathways for industrial structure upgrading [3,4,5,6], limited attention has been paid to the differential pathways between urban and county-level regions. Rural counties, lacking the advanced factor endowments of cities and facing industrialization disparities among themselves [7], cannot simply emulate urban development models when pursuing industrial upgrading. Instead, they must leverage their unique endogenous advantages to forge distinctive development paths.

Rural counties are situated in the intermediate zone between cities and rural areas, connecting to cities on the one hand and leveraging rural factor resources on the other. This positioning creates conditions for the characteristic upgrading of the industrial structure in rural counties. This paper argues that the innate advantage of rural county industries lies in their deep integration with agriculture and the vertical integration advantage of the agricultural industry chain. Rural counties can utilize the abundant rural resources such as mountains, waters, farmlands, forests, lakes, grasslands, and deserts around them, as well as layout industries such as agricultural product processing and cold chain logistics, continuously accumulating experience in agriculture-oriented industrial structure upgrading through “learning by doing.” On this basis, rural counties are expected to build regional brands around local specialty agricultural products, blurring the boundaries between agriculture, manufacturing, and services. For example, agriculture has ecological and cultural functions [8], which creates opportunities for the upgrading of the industrial structure in rural counties.

The key challenges in leveraging agricultural strengths to upgrade the industrial structure of rural counties lie in the layout, introduction, and innovative recombination of factor resources. Overcoming these challenges requires agricultural innovation talents to transform county-level industries through the application of knowledge. On the one hand, it is necessary to decode the implicit logic of agricultural strengths, including identifying “what the strengths are” and “how to utilize these strengths.” The process of uncovering these strengths is essentially one of opportunity recognition, which can be more swiftly accomplished by individuals with compound knowledge [9]. Utilizing these strengths involves innovating industrial business models, with the flow of advanced knowledge playing a crucial role in this process [10]. Agricultural innovation talents are thus the embodiment of these dual requirements. On the other hand, it is essential to address all difficulties encountered during the iteration, allocation, and recombination of factor resources. For example, the application of emerging factors such as big data and artificial intelligence in various industries has been widely documented [1,4,11,12,13]. Their primary role is reflected in digital transformation, intelligent upgrading, and network connectivity. Although digital and intelligent factors can promote stable industrial development by reducing costs, increasing efficiency, and accelerating the circulation of goods [5,12,14], the insufficient digital literacy of farmers, disconnection from the actual environment, and inadequate investment have created a barrier to the entry of these factors [15]. This barrier is essentially due to the lack of knowledge among multiple stakeholders, which leads to the failure of knowledge interaction mechanisms. Intellectual resources and effective external support are essential in driving the application of digital and intelligent factors in agricultural industries [16]. Moreover, farmers and agricultural enterprises are not adept at promoting industrial integration or aggregating and transforming market demand information related to agricultural strengths. Therefore, the knowledge-empowering role of agricultural innovation talents is crucial in this context, forming the tightly interlinked theoretical framework of “talent–knowledge–industry” in this paper.

It is widely acknowledged in the theoretical community that the improvement of human capital plays a positive role in industrial structure upgrading [9,17,18,19]. The positive impacts of knowledge spillover and industrial integration on industrial structure upgrading have also been affirmed [2,20]. The significant role of digital and intelligent transformation in driving industrial development has likewise been extensively discussed [4,6,21,22]. In addition, some studies have explored the pathways for the manifestation of agricultural strengths from perspectives such as demand matching, worker literacy, and industrial layout [8,15,16,23,24].

However, the above studies may have three limitations. First, the discussion on the talent required for industrial structure upgrading in rural counties is too broad. Injecting agricultural strengths into the industrial structure upgrading of rural counties not only requires talent to connect the interests between agricultural strengths and advanced factors but also demands sensitivity to market demand information and opportunities for industrial integration. This sensitivity originates from systematic and specialized agricultural training, which has not been sufficiently discussed in existing studies. Second, the perspectives on the accessibility and training objectives of agricultural innovation talents need to be enriched and refined. Theoretical discussions on the cultivation of innovative talents generally focus on two dimensions: education level and literacy [17,25]. However, without delving into the disciplines and their construction levels to explore the effectiveness of innovative talent cultivation, it may not be possible to derive actionable policy implications. Moreover, empirical evidence on how various educational policies respond to industrial development strategies is still lacking, and we may overestimate the functions of educational policies. Third, the mechanism of knowledge empowerment rarely appears concretely in the actual context of industrial structure upgrading in rural counties and mostly departs from agricultural strengths. While knowledge is acknowledged to promote high-quality industrial development [9,20], existing studies mostly focus on the transformation of traditional agriculture using manufacturing knowledge [16] and the accumulation of knowledge within agriculture itself [8,24,26]. A knowledge structure that leads other industries’ development with agricultural strengths as the driving force has yet to be established, and the question of who will build it is equally worth exploring.

In light of the above, the research questions of this paper are as follows: (1) Does agricultural innovation talent have an empowering effect on the upgrading of industrial structures in rural counties? Does this effect exhibit spatial spillover effects? (2) What is the intrinsic logic of knowledge empowerment in the realization process of the above-mentioned empowering effect? (3) According to the theoretical logic of economics, the consumption and investment demands of industries, as well as the difficulty of transformation, may influence the process of industrial structure upgrading. How is the functioning of agricultural innovation talent affected by the conditions for knowledge to be transformed into value and differences in industrial endowments? This paper attempts to answer the above questions and thereby fill the research gap.

China’s “World-Class Disciplines” initiative in agriculture provides an opportunity to address the aforementioned research questions. The program explicitly prioritizes the cultivation of “top-tier innovative talents” and mandates that selected disciplines must “align with national food security priorities” and “establish industry–academia–research–application integration bases”, transforming disciplinary platforms into incubators and talent magnets for county-level agricultural innovation. Rooted in disciplinary excellence, the initiative directs targeted investments in research funding, platforms (such as national key laboratories), and high-caliber faculty positions, while channeling substantial financial resources into agricultural production hubs at the county level to construct provincial–ministerial key laboratories, engineering research centers, and other innovation carriers. From every analytical perspective, China’s “World-Class Disciplines” initiative serves as an optimal observational window for examining the functions of agricultural innovation talents.

The contributions of this study are threefold. First, it establishes a theoretical analytical framework of “talent–knowledge–industry.” In the empirical part, it adopts a quasi-natural experiment, treating China’s “World-Class Agricultural Disciplines” construction project as an exogenous shock to the cultivation of agricultural innovation talent. It investigates the impact of this shock on the upgrading of industrial structures in rural counties and whether this impact exhibits spatial spillover effects. This not only helps to clarify the necessity of cultivating agricultural innovation talent through high-level discipline construction and integrating such talent into rural county industries but also provides a new perspective for the design of integrated policies on education and industrial development. Second, it emphasizes the irreplaceability and diffusibility of the endogenous agricultural advantages in rural counties. From the perspective of knowledge empowerment, it concretely examines the internal mechanism by which agricultural innovation talent, with its “smart brains,” promotes the transformation of the industrial structure in rural counties. Relying on the analytical approach of creating, disseminating, and applying knowledge, this paper covers three key links that induce the upgrading of the industrial structure in rural counties based on agricultural advantages: industrial integration, supply and demand matching, and the effective use of digital and intelligent factors. It aims to systematically investigate the dividends gained by knowledge empowerment in the process of agricultural innovation talent transforming rural county industries. Third, it constructs a “core–periphery” system to provide comprehensive support for agricultural innovation talent in upgrading the industrial structure of rural counties. Based on the positive externalities of agricultural innovation talent, this paper further analyzes the external conditions for knowledge to be transformed into value and whether differences in industrial endowments in rural counties affect the functioning of agricultural innovation talent. Thus, the implementation of complementary policies to support the “periphery” maximizes the value of agricultural innovation talent located at the “core.”

The remainder of this paper is organized as follows: Section 2 reviews the literature and background. Section 3 presents the theoretical analysis and research hypotheses. Section 4 describes the research design. Section 5 analyzes and discusses empirical results. Section 6 summarizes the research conclusions, theoretical contributions, and policy implications, and suggests areas for future research advancement.

2. Background and Literature Review

2.1. Industrial Structure Upgrading

The concept of industrial upgrading has evolved through diverse theoretical lenses. Early contributions by Clark [27] and Chenery [28] established an industrial transition paradigm, positing that economic dominance sequentially shifts from primary to secondary and ultimately tertiary sectors. While this trajectory reflects macro-level development patterns, contemporary research increasingly emphasizes inter-industrial synergy in specific contexts like county economies, framing upgrading as the transformation of industrial sectors from low-value to high-value activities [29]. Building on this insight, numerous studies measure industrial upgrading through sectoral output share relationships [30,31,32]. In the context of this paper, global value chain (GVC) theory posits that agricultural value enhancement fundamentally requires the transcending of primary production to embrace high-value domains like R&D, brand marketing, and agri-tourism integration. This evolution necessitates two structural transformations: first, agricultural industrialization drives demand for deep processing of agricultural products, propelling the secondary sector’s upgrade from low-end manufacturing to technology-intensive processing (e.g., food engineering, bio-manufacturing); second, agricultural servitization spawns tertiary sector innovations in agri-tech services, supply chain finance, and rural tourism. As these transformations must be grounded in county-level contexts, the tertiary-to-secondary sector value-added ratio effectively captures the “depth of servitization transition” in agricultural value chains. This metric thus serves as a valid indicator of a county’s capacity to leverage agricultural strengths for industrial upgrading, reflecting the dual process of sectoral value migration and systemic restructuring in regional development.

Research on the determinants of industrial structure upgrading is quite extensive. Factors such as education [33], human capital [9,17,18,19], knowledge [20], industrial integration [2], technological innovation [34,35], trade [3], institutions [36], national finance and FDI [33], industrialization and urbanization [37], land use [38], urbanization [39], factor allocation [26], technological progress [40,41], and public sector support and talent support [42,43] have all been shown to influence the upgrading of industrial structure.

2.2. Agricultural Innovation Talents

The concept of agricultural innovation talent can be deconstructed as the integration of agricultural innovation literacy and human capital. The role of human capital encompasses two aspects. On the one hand, human capital is an indispensable element of production, determining the quantity of output [34] and directly influencing production efficiency [44]. It also guides the direction of industrial development through opportunity recognition [9]. On the other hand, human capital can influence the conduct of innovative activities [45] and has a complementary relationship with emerging factors represented by digital and intelligent tools [11]. Focusing on the agricultural sector, high-quality agricultural human capital is the main force in transforming traditional agriculture [46]. In reality, there is often a lack of intermediaries to connect agricultural research results with farmers [42]. Agricultural innovation talents can act as “innovation brokers” in agriculture, promoting the efficient use of new technologies in agro-industries [47] and having a long-term preference for assisting the quality transformation of agro-industries [44]. In addition, agricultural innovation talents possess strong comprehensive management abilities in agro-industries [48] and can also deeply explore the value of agricultural multifunctionality, skillfully converting the cultural and ecological functions of agriculture into value [8]. It is evident that agricultural innovation talents cultivate sustainable momentum for the upgrading of the industrial structure in rural counties.

Agricultural innovation talents have multiple sources. They can be nurtured from higher education [18,25] or evolve from labor forces with agricultural skills [41]. However, the latter requires the accumulation of practical knowledge through “learning by doing” in agro-industries over a long period and does not consider new factors sufficiently. Therefore, the agricultural innovation talents discussed in this paper are the products of high-level discipline construction in higher education, focusing on “quality” rather than “quantity.” Their impact on the upgrading of the industrial structure in rural counties through leveraging agricultural strengths should be disruptive and holistic.

2.3. The Impact of Innovative Talents on Industrial Structure Upgrading

Innovative talents belong to the category of high-quality human capital and have two types of empowering effects on industrial structure upgrading. First are the research and application of technology, injecting momentum for the development of advanced technologies into industries. Innovative talents contribute to technological innovation [45,49,50] and, together with emerging factors represented by digital technology, help drive the quality transformation of industries [11]. In addition to being innovators of technology, innovative talents are also early adopters and followers of new technologies, promoting their widespread adoption [51]. Moreover, innovative talents accelerate the flow of knowledge [10], thereby facilitating the transition from old to new growth drivers. However, this effect is constrained by the degree of human resource misallocation; they should be leaders rather than rent-seekers [52]. The second type involves the recombination of factor resources and optimization of industrial layout in a location-specific manner, uncovering endogenous advantages for industries. Innovative talents possess compound knowledge reserves, enabling them to identify market opportunities [9] and optimize the market environment and trade system [17], all of which can provide support for the upgrading of industrial structure.

Agricultural innovation talents, in addition to the aforementioned roles, also have the characteristics of understanding and loving agriculture, and are motivated to explore solutions to complex problems in agro-industries [44]. During the process of scaling up and strengthening agro-industries, they are not only propelled by high-quality talent but also repelled by non-specialized personnel. Agricultural innovation talents must both unleash the propelling force and mitigate the noise caused by the repelling force. This noise includes the insufficient technical literacy of farmers [15], profit-seeking, and risk-avoidance preferences [24,53], which act as barriers to the structural upgrading of agro-industries. Meanwhile, the substitution of labor by capital can also, to some extent, harm the interests of farmers [26]. Agricultural innovation talents are best suited to help reduce the economic pressure they face, enhance their willingness to adopt technology, and promote a balanced distribution of benefits. This intermediary role acts as a booster for the iterative innovation of agro-industries [47]. Moreover, driving industrial development starting from agriculture also requires organizers to provide appropriate incentives and coordination for participants in the industrial and value chains [43], which is also within the scope of the functions of agricultural innovation talents.

2.4. Literature Summary

The existing literature has provided a robust foundation for this study. However, it primarily faces three limitations: First, the discussion on the talent required for industrial structure upgrading in rural counties is overly broad. Second, there is a notable deficiency in research that examines the cultivation of talent for rural county industries from the perspective of linking agricultural strengths with high-level discipline construction. Third, the concrete manifestations of knowledge empowerment are not sufficiently reflected in the existing studies on the upgrading of industrial structures in rural counties, and even fewer studies take the utilization of agricultural strengths as the logical starting point. This study aims to address these limitations and explore the experience of talent cultivation and utilization for the upgrading of industrial structures in rural counties within the framework of “talent–knowledge–industry.”

2.5. Background of China’s “World-Class Disciplines” Construction Project

China’s “World-Class Disciplines” construction project is a quintessential example of high-level discipline construction and a fertile ground for producing innovative talents. Correspondingly, the “World-Class Agricultural Disciplines” construction project is also capable of cultivating agricultural innovation talents. The rationale for this approach is twofold: On the one hand, from the perspective of policy design, the core objective of the “World-Class Disciplines” construction project is to build the core competitiveness of disciplines through the integration of resources and institutional innovation. The enhancement of innovation capability is listed as a key indicator of its evaluation system. This is clearly reflected in the “Overall Plan for the Coordinated Promotion of the Construction of World-Class Universities and Disciplines” issued by the State Council of China in 2015, as well as the “Interim Measures for the Implementation of the Coordinated Promotion of the Construction of World-Class Universities and Disciplines” formulated by the Ministry of Education, Ministry of Finance, and National Development and Reform Commission of China in 2017. On the other hand, during the construction process, the “World-Class Agricultural Disciplines” have objectively provided systematic agricultural knowledge training for agricultural innovation talents by establishing interdisciplinary research platforms, introducing cutting-edge technologies, and building collaborative mechanisms among industry, academia, and research institutions.

To date, two rounds of the “World-Class Disciplines” construction project have been implemented, with the first and second lists of universities involved in the “World-Class Disciplines” construction being announced in 2017 and 2022, respectively. In this study, the “World-Class Agricultural Disciplines” cover disciplines such as Agricultural Resources and Environment, Animal Husbandry, Veterinary Medicine, Horticulture, Crop Science, Plant Protection, Agrostology, Fisheries, Forestry, Soil and Water Conservation and Desertification Control, and Agricultural and Forestry Economics and Management.

3. Theoretical and Hypothesis Development

3.1. Direct Influence of Agricultural Talents

The core of the transition from traditional agriculture to modern agriculture is the introduction of new factors [46], and modern agriculture serves as the foundation for expanding manufacturing and service industries based on agricultural strengths. Agricultural innovation talents facilitate the introduction of these new factors. On the one hand, small-scale farmers are often at a disadvantage, and the introduction and utilization of new factors typically require the assistance of agricultural social services, in which agricultural innovation talents play a guiding role. Agricultural innovation talents not only optimize industrial layout in a location-specific manner, thereby enhancing the efficiency and fit of agricultural social services in county-level industries, but also fill the gap in the cultural and ecological functions of modern agriculture, thus promoting the extension of the agricultural value chain. In particular, since land is the object of transformation in agricultural production and is prone to being exploited at the expense of the environment for profit [54], the cultural and ecological functions of modern agriculture can be converted into value [8]. The role of agricultural innovation talents is to achieve this goal, not only reconciling the conflict between the environment and profit but also upgrading the sophistication of the industry, completing the entire process of injecting agricultural strengths into county-level manufacturing and service industries. On the other hand, the agglomeration of high-end factors relies on market entities such as family farms, cooperatives, and leading enterprises; this involves both internal and external division of labor within the industrial chain. Agricultural innovation talents, having received systematic and complex knowledge training, have in-depth thoughts on how to deepen the industrial division of labor system in counties to diffuse agricultural strengths and are quick to address practical difficulties. In Chinese counties, the subdivision of management rights offers opportunities to reshape the advantages of the division of labor. Agricultural innovation talents, as composite talents with strong opportunity recognition capabilities [9], not only allocate high-end factors to the appropriate links in the division of labor system but also ensure the organic connection between high-end factors and other factor resources, providing an inexhaustible driving force for the upgrading of the industrial structure in rural counties.

Agricultural innovation talents are irreplaceable in promoting the upgrading of industrial structures in rural counties. This is not only because human capital is an important factor of production in the production function [34] and has a complementary relationship with capital and land factors, but is also because the absence of agricultural innovation talents would lead modern agriculture to “involution,” making it difficult for county-level industries to leverage agricultural strengths for transformation and upgrading. For example, drawing from the experience of the United States, farms have two options when investing in agricultural machinery: large-scale self-investment or outsourcing machinery operation services. The latter sees high transaction costs and significant uncertainty, leading most U.S. farms to choose the former, yet without guaranteed investment efficiency. This is still based on the premise that the relationship between people and land in the United States is not strained. In contrast, in Japan, where farms operate on a small scale, the purchase of self-use machinery has plunged Japanese agriculture into a vicious cycle of inefficient machinery investment, resulting in a low-level trap in agricultural production. The apparent reasons for this phenomenon seem to be cultural and natural endowment factors, but in fact, the root cause is the lack of high-quality intermediaries skilled in integrating agricultural resources, which makes it difficult to deeply transform agricultural business models. Farmers or entrepreneurs often exhibit short-sighted behavior driven by profit motives [53,55], which is not conducive to serving the overall interests of industrial development. Agricultural innovation talents, on the other hand, are coordinators of various interests and reconstructors of the agricultural development order, capable of finding the Pareto optimal state of scaled agricultural operations. This, in turn, drives changes and upgrades in manufacturing and productive service industries, creating opportunities for county-level industries to advance based on high-quality agricultural development. Moreover, the adoption of innovative technologies requires the consideration of the life cycle of local agricultural products, as well as the socio-cultural and institutional environment. It is not advisable to make simple decisions on whether to adopt them [56]. Agricultural innovation talents are the judges of whether labor means and objects are compatible.

From the actual situation of rural counties in China, the “core–periphery” structure between counties and cities determines that the risk of labor and capital outflow from counties is relatively high, which is unfavorable for the upgrading of the industrial structure in rural counties. Agricultural innovation talents not only induce factor flow through the development of urbanization in counties and improve the efficiency of land resource utilization but also contribute to the transfer of agricultural population, scaled agricultural operation, the aggregation of human capital, and the transformation and upgrading of rural industries. In this process, issues such as the high vacancy rate of homestead land, the shortage of construction land in counties, the abandonment of farmland due to population outflow, the difficulty in utilizing the multifunctionality of agriculture in the integration of three industries, and the difficulty of labor employment can be addressed simultaneously. These issues are all constraints on counties’ ability to leverage agricultural strengths. It is evident that agricultural innovation talents have cleared numerous obstacles in the upgrading of the industrial structure in rural counties. Figure 1 presents the theoretical analytical framework of this study. In view of the aforementioned considerations, this paper posits Hypothesis 1 as the central proposition, followed by an elucidation of the transmission mechanisms derived from H1 through Hypotheses 2–4, which decompose the theoretical pathways underpinning the primary effect.

H1: 

An increase in agricultural innovation talents helps promote the upgrading of the industrial structure in rural counties.

3.2. Mechanism of Influence

The basic meaning of industrial integration is the convergence of the primary, secondary, and tertiary sectors [57]. Some scholars focus on the integration of manufacturing and services, defining it as the path through which traditional manufacturing enterprises enhance their market competitiveness by integrating product and service resources [58]. It is evident that industrial integration can be regarded as a subset of industrial quality transformation. It is not only influenced by external favorable shocks but also occurs as a result of internal structural optimization, having a positive impact on the upgrading of industrial structures [2]. From the perspective of transaction cost theory, market transactions involve costs such as information search, negotiation, and monitoring [59]. Industrial integration, as the consolidation of multiple industries, can internalize external issues and naturally reduce the transaction costs of inter-industry cooperation, while also avoiding the instability of contractual relationships. This establishes a form of market power derived from industrial integration. The greater the market power, the stronger the innovation incentives of market entities [45]; the increase in technological density will drive the upgrading of industrial structures [35]. In rural counties in China, the realization of industrial structure upgrading must address how industries can be integrated. If industrial integration is abandoned, the added value of agricultural primary processed products will remain low, and profits will still be absorbed by cities. If rural counties engage in homogeneous competition with urban manufacturing, the competitiveness of county-level industries will always be weaker than that of cities, and they may even become “urban factories.” Therefore, the upgrading of the industrial structure in rural counties must either involve the development of agricultural deep processing or the development of agro-related service industries based on the multifunctionality of agriculture. Moreover, industrial integration can propel county-level industries to climb to higher positions in the value chain. This is not only because it allows counties to extend their industrial chains and enhance their value chains, increasing the added value and market competitiveness of products, but also because the process of industrial integration involves the allocation of new and old factor resources and is a process of the mutual penetration of urban and rural factor resources. This creates opportunities for digital factors to enter county-level industries, and digital factors are also a driving force for the upgrading of industrial structures [1], which can also alleviate funding constraints in this process [22].

The supply of agricultural innovation talents is capable of bringing new knowledge of industrial integration to rural counties in a location-specific manner. On the one hand, industrial integration is an innovation in industrial business models, and the supply of agricultural innovation talents provides knowledge to enable its realization and sustainable development. From the perspective of the diffusion of innovation theory, the adoption of industrial innovation models requires innovators, early adopters, and followers to transform them into practical experience within a certain social network, promoting the generalization of new things and concepts in rural counties [51]. In rural counties, to promote industrial integration based on agricultural strengths, it is first necessary to break the fragmented situation of the integrated development of the primary, secondary, and tertiary industries, and also to create new knowledge of industrial integration to find the right fit, and agricultural innovation talents act as innovators in this process. Subsequently, it is necessary to connect urban and rural factor resources at a low cost, reconcile the interests of multiple parties, and dynamically optimize the form of industrial integration to form a virtuous cycle of knowledge about industrial integration. In addition, agricultural innovation talents create knowledge on how to rely on industrial linkages to sustainably promote industrial integration. Talents incubated by high-level disciplines have both theoretical and industrial practice knowledge, and can deconstruct upstream and downstream industrial relationships in a complex industrial integration system, forming a hierarchical map of industrial forward and backward linkages to ensure the stable operation of industrial integration. On the other hand, agricultural innovation talents are good at clarifying the implementation path of industrial integration based on the actual situation of rural counties and promoting the iteration of localized industrial integration knowledge in rural counties. For example, the prerequisite for the smooth connection of agricultural product processing and service industries with modern agriculture is high agricultural production efficiency, and it is necessary to properly adjust the proportion of agricultural machinery and effective labor participation in production [26]. Another form of industrial integration in rural counties is the integration of culture and tourism, where the challenges lie in how to mobilize consumer participation and effectively develop cultural and tourism resources [60]. The smooth passage of these links requires agricultural innovation talents to do a good job in terms of the top-level design of factor resource allocation for industrial integration in the actual context of rural counties, dynamically adjust it for a period of time, and then elevate it to the level of the necessary stage of industrial integration theory in rural counties. At the same time, the process of industries moving from a fragmented to an integrated development pattern is essentially also a process of the innovation ecosystem extending from a single discipline to multiple disciplines, with value co-creation being the main driving force for the development of the innovation ecosystem [61]. Agricultural innovation talents are pioneers in coupling cross-industry knowledge and transforming it into practical value, and are also coordinators of interests across governments, markets, and enterprises, unblocking the bottlenecks in achieving industrial integration in value co-creation. In addition, during the process of industrial integration, agricultural innovation talents can establish and participate in various forms of industrial consortia, cooperatives, and other organizational forms to closely unite the main bodies of different links, forming a risk-sharing community of interests. They can also guide fiscal incentives to pursue the weak links of industrial integration in the opposite direction, endowing policy practice with empirical knowledge. Given this, this paper proposes Hypothesis 2.

H2: 

An increase in agricultural innovation talents is conducive to creating knowledge and promoting the upgrading of the industrial structure in rural counties through industrial integration.

Responding to external demand is an essential prerequisite for ensuring the sound development of industries, and demand-induced industrial structure upgrading is in line with economic laws [62]. A reduction in the degree of demand information asymmetry implies that the scale of demand will continue to expand. On the one hand, this process enables firms to achieve economies of scale and enhance their profitability [45]. The fixed R&D costs of enterprises are spread over a larger scale of output, significantly reducing the marginal cost of innovation per unit of product [34]. On the other hand, firms’ risk tolerance for uncertainties in technological pathways will also be strengthened [63]. These dual forces lay a solid financial foundation and enhance the development resilience for their innovation-driven transformation, thereby accelerating the pace of industrial structure upgrading in rural counties. From the theory of monopolistic competition, it is known that consumption diversity can endow industries with endogenous incentives for innovation [64]. Driven by the profit-seeking goals of entrepreneurs, industrial structure upgrading is more likely to coagulate the momentum of factor agglomeration. Similarly, from the perspective of the political economy of industrial capital circulation, consumption is the purpose of production. Traditional industries can no longer stand firm in the consumption and investment markets. The pressure to recover advanced capital and the motive to extract surplus value will naturally force the upgrading of the industrial structure in rural counties. Moreover, when the scale of domestic demand grows to a certain extent, the market structure will spontaneously evolve [65], and homogeneous competition among county-level industries will intensify. Downstream differentiated competitive demands force upstream firms to develop customized products to obtain monopoly profits higher than those of standardized products [66]. Productive service industries are thus born, or alternatively, the proportion of living service industries is increased under the pressure of overcapacity, thereby shaping the momentum for the upgrading of the industrial structure in rural counties. The fact of global deindustrialization also serves as evidence for this view. From any perspective, it is beneficial to promote the upgrading of the industrial structure by unblocking the channels for responding to external demand for counties.

The key to responding to demand information lies in unblocking the supply–demand connection path, especially ensuring that demand can effectively guide the types, models, and scenarios of supply, and that quality transformation signals from the supply side are fed back to the demand side, achieving dynamic two-way interactive upgrading. However, in reality, there are bottlenecks in the ability of county-level industries to sense external demand, which are reflected in two aspects: First, county-level industries have a lag in tracking market demand. A significant manifestation of changes in residents’ consumption is the upgrading of consumption structure [67,68], which essentially reflects the hierarchical nature of human needs and their sequential progression [69,70]. Therefore, the questions of how to process external demand information into empirical knowledge and how to complete the process of transforming knowledge into specific industrial projects to create characteristic industrial brands are both challenging. Leading with the agricultural strengths of counties to create waves of expanding demand space based on new products and services is even more difficult. Second, the enthusiasm of the supply side to respond to market demand is insufficient. For counties, external demand information is an exogenous shock. Once there is a lack of empirical knowledge for the connection between old and new industries, the allocation of labor and capital will be disorderly and inefficient, leading to a decline rather than an increase in productivity levels [71]. Even if high-quality products or services are created, it is difficult to avoid adverse selection in the market. Moreover, most farmers are not optimistic about their ability to successfully run a business, nor are they confident in scaling up and strengthening their industries [24], which is not conducive to leveraging agricultural strengths for industrial structure upgrading in counties. Agricultural innovation talents boost the confidence of the supply side to respond to market demand. First, as a typical example of high-level discipline cultivation, the “World-Class Disciplines” construction project places great emphasis on unblocking the bottlenecks in industrial transformation [18]. Agricultural innovation talents cultivated by the “World-Class Agricultural Disciplines” also possess the ability to integrate and synthesize complex external agricultural demand information and promote a new round of knowledge innovation in a location-specific manner in industrial practice. This process simultaneously releases signals that incentivize high-quality talents to gather in counties, which can also be transformed into positive expectations of investors towards county-level industries. Since expectations are positively correlated with the investment enthusiasm of market entities [72], the confidence of farmers in cultivating characteristic county-level industries will also be enhanced. Second, as demand knowledge has local characteristics and difficulties in terms of mobility [73,74], demands that cannot be realized in cities, once they find a landing scene in counties, will become a powerful force driving the growth of county-level consumption. Agricultural innovation talents act as a booster for the two-way connection between urban industrial knowledge, demand gaps, and county-level factor resources. Finally, agricultural innovation talents belong to high-quality human capital, which has a positive impact on economic growth and economic structure optimization [49,50]. They optimize the economic environment for transforming demand into industry from two dimensions: residents’ purchasing power and the diversification of consumption scenarios. Given this, this paper proposes Hypothesis 3.

H3: 

An increase in agricultural innovation talents is conducive to disseminating knowledge and promoting the upgrading of the industrial structure in rural counties through responding to demand information.

The integration of digital and intelligent tools can generate momentum for the upgrading of the industrial structure in rural counties. Advances in digital technology mark the beginning of disruptive innovation in industries and can empower the upgrading of industrial structures [75], with this effect being particularly pronounced in rural areas [1]. The core value of digital and intelligent tools lies in the data element. Incorporating it into production can create greater value for enterprises by integrating it with other factor resources [11]. It is beneficial for enterprises to conduct innovative activities, reduce the degree of factor resource misallocation, and achieve cost reduction and efficiency improvement [12,14]. It also enhances the management and operational efficiency of enterprises [21], thereby clearing potential obstacles in the upgrading of industrial structures. Robots, the digital economy, digital finance, and e-commerce are all derivatives of digital technology. From the production side, the application of industrial robots is a boost to factor collaborative recombination and the upgrading of industrial structures [76]. The digital economy and digital finance have also been proven to promote the upgrading of industrial structures. Their mechanisms involve the digital economy breaking down information asymmetry, leading to enhanced liquidity and the accessibility of factor resources, while digital finance lowers the barriers to innovation and entrepreneurship [4,22]. From the consumption side, e-commerce drives the upgrading of industrial structures by expanding the scale of trade and increasing the degree of industrial agglomeration [5]. The development of digital finance and the popularization of robots respectively alleviate consumers’ financial constraints and increase residents’ labor income share [4,13], releasing positive externalities for the extension of demand space to upgrade industrial structures. In rural counties with agricultural strengths, leveraging the multifunctionality of agriculture to drive the upgrading of industrial structures is their comparative advantage over cities. Digital technology can also empower this process. On the one hand, it can upgrade the industrial structure of counties by innovating traditional industries, such as the application of remote sensing monitoring, precision operations, and the intelligent management of water, fertilizer, and pesticides in the agricultural production process. On the other hand, digital technology drives the upgrading of the industrial structure in counties by extending the industrial chain. For example, in the integration of culture and tourism, digital technology can drive the transformation and upgrading of the tourism industry through knowledge learning effects, strategic change effects, and industrial model optimization effects [6]. It is evident that the effective use of digital and intelligent tools is a boost to the upgrading of the industrial structure in rural counties.

In reality, the digital divide is prevalent [77]. For example, promoting the transformation and development of agriculture through digital technology requires joint support from the public sector and new types of agricultural talents to be realized [16]. Agricultural innovation talents can play an important role in promoting the integration of digital and intelligent tools with agricultural strengths and effectively embedding them into county-level industries. Under the analytical framework of Marx’s theory of productive forces, laborers are the most active factor in productive forces. Digital and intelligent tools belong to new types of labor means, while traditional industrial resources, digital resources, virtual space, biological genes, and micro-particles constitute the objects of labor. The role of agricultural innovation talents is to guide laborers to use new labor means to transform the objects of labor, thereby promoting the integration of digital and intelligent tools with county-level industries. This process reveals relative advantages, compatibility, and visibility, and widely promotes their use in all links of the industrial chain, similar to the role of innovators and followers in the diffusion of innovation theory after the emergence of new technologies [51]. The supply of agricultural innovation talents is an important way to promote a virtuous cycle of education, science and technology, and talent, forming a three-dimensional knowledge feedback mechanism among research institutions, technology enterprises, and local industrial practices. First, the knowledge learned is used to optimize the layout of county-level industries and refine agricultural strengths. Then, the local practice knowledge is fed back to research institutions and technology enterprises, unblocking the pathway for knowledge reflux. Repeating this process can ultimately form a Pareto optimal state for the integration of digital and intelligent tools into county-level industries and harmonize the complex production relations among the three elements of productive forces. Given this, this paper proposes Hypothesis 4.

H4: 

An increase in agricultural innovation talents is conducive to applying knowledge and promoting the upgrading of the industrial structure in rural counties through the effective utilization of digital and intelligent tools.

4. Research Design

4.1. Data Source

To empirically examine the impact of agricultural innovation talents on the upgrading of the industrial structure in rural counties and to test the research hypotheses proposed in the theoretical analysis, this paper utilizes data from 1771 rural counties across 29 provinces in China over the period of 2013–2021 as the research sample. The total sample size in the benchmark regression is 15,939. The data for the dependent variable, core independent variable, and control variables are sourced from the China Rural County Statistical Yearbook and the Ministry of Education of China. For the mechanism variables, the night-time light data of rural counties are obtained from the China “DMSP-OLS-like” night-time light remote sensing dataset. The calculation method for robot installation density follows the approach of Acemoglu and Restrepo [78], with data sourced from the International Federation of Robotics. The e-commerce enterprise data are obtained from the Ministry of Industry and Information Technology of China.

The reason for not selecting samples after 2021 is twofold. First, the representation of agricultural innovation talents is closely related to the “World-Class Disciplines” construction. Although the second round of the “World-Class Disciplines” construction project began in 2021, the specific list was not announced until 2022. Therefore, this paper considers the second round of the “World-Class Disciplines” construction project to have taken effect from 2022 onwards and treats 2021 as a continuation of the first round. This setting also aligns with the reality that policy effects often have a lag. Thus, the chosen time period can fully capture the effects of the first round of the agricultural “World-Class Disciplines” construction project and more accurately observe whether the benefits of the new agricultural labor supply exist. Second, in 2022, the National Health Commission of China declared the end of the COVID-19 pandemic. The lifting of pandemic restrictions led to a surge in commercial activity, which was a positive shock to residents’ income but increased the risk of “free-riding” on the policy effects in this study. The year 2013 is chosen as the starting point for the analysis because the National Bureau of Statistics of China changed the statistical caliber of rural residents’ income in 2013. Selecting samples from 2013 and onwards can enhance the precision of the empirical examination.

In examining how agricultural innovation talents drive the upgrading of the industrial structure in rural counties, the use of the Chinese sample is representative. China, as the world’s largest agricultural country, faces the triple pressures of shrinking demand, supply shocks, and weakening expectations, but it has the advantage of systematic policy support and has conducted pilot projects of integrated industry–academia–research in multiple locations. Moreover, China has rich practices in food security, the modernization of small-scale farming, poverty reduction, and other areas, emphasizing the deep integration of technological applications with traditional industries and the nodal function of rural counties in urban–rural integrated development. This model not only addresses the dispersion of small-scale farming economies but also avoids the drawbacks of relying solely on capital-driven development. It provides a replicable model for global agricultural transformation and offers worldwide experience for rural counties to follow an upgrading path distinct from that of cities.

4.2. Variable Setting

Dependent Variable

In this study, the dependent variable is the upgrading of the industrial structure in rural counties. In existing research, there are two common approaches to representing this variable: The first method uses the degree of industrial sophistication, which reflects the development trend of the service sector, represented by the ratio of the tertiary to the secondary industry [30,31]. A larger value of this indicator implies a more advanced industrial structure, and its growth represents the upgrading of the industrial structure. The second method treats the primary, secondary, and tertiary industries as low-, medium-, and high-level industries, respectively, assigns their contributions to the upgrading of the industrial structure from 1 to 3 in ascending order, multiplies the value-added share of each industry by its respective contribution, and sums them up to obtain the indicator of industrial structure upgrading [32].

This study selects the first approach to measure the degree of industrial structure upgrading in rural counties. The rationale for this choice is as follows: The transformation path of agricultural strengths in rural counties is asymmetric. The essence of upgrading is to rely on agricultural modernization to generate the vertical extension of the industrial chain (such as agricultural product processing) and horizontal integration (such as agritourism services). The first method dynamically captures the coordinated development level of the secondary and tertiary industries related to agriculture through their ratio. In contrast, the second method rigidly treats agriculture as a low-level industry, ignoring the substantial contributions of agricultural technological innovation and value-added enhancement to the overall structural upgrading, which contradicts the core concept of this study that rural counties leverage their agricultural strengths to promote industrial structure upgrading. From the perspective of policy targeting differences, the first method pays attention to the relative evolution trends of “agricultural industrialization” and “agricultural servitization” to some extent. However, the linear synthesis of the second method simply equates the expansion of the tertiary industry with upgrading, which may mislead policies into excessively pursuing “de-agrarianization.” This is clearly less reasonable than the first method. Moreover, this study does not focus on the absolute comparison of the degree of industrial sophistication among different rural counties but rather on the evolution of the industrial structure within each county. The above considerations indicate that the indicator selected in this study is more scientific.

4.3. Independent Variable

The core independent variable in this study is agricultural innovation talent. The preceding sections have clearly elaborated on the rationality and representativeness of using the “World-Class Disciplines” construction project to estimate the impact of agricultural innovation talent. The measurement approach adopted here follows the general practice of employing the “World-Class Disciplines” construction project as a quasi-natural experimental shock [18,79], further refining this method for application to the agricultural domain. Specifically, counties located in cities with the “World-Class Agricultural Disciplines” construction project are designated as the treatment group, while the remaining counties serve as the control group. County samples that entered the treatment group and are from the year 2017 onward are identified as being influenced by agricultural innovation talent and are assigned a value of 1, whereas other samples are assigned a value of 0.

In the literature representing the impact of human capital, some studies have utilized the number of university graduates as a measure. However, this study argues that continuing this practice (i.e., using the number of graduates from the “World-Class Agricultural Disciplines” construction project) has potential flaws: First, it overlooks the integrity of the impact mechanism of agricultural innovation talent. The “World-Class Disciplines” construction not only directly supplies human capital through talent cultivation but also generates knowledge spillovers to counties through multiple channels such as research platform establishment, technology transfer, and the reconstruction of industry–academia–research networks. The number of graduates captures only a single dimension. Second, the theoretical logic behind this approach is not aligned with this study. Within the “talent–knowledge–industry” theoretical framework of this paper, variable design should conform to the logic of how the “supply-side reform” of agricultural innovation elements empowers the upgrading of the industrial structure in rural counties. The root role of the “World-Class Agricultural Disciplines” construction project is the transformation of weak links in county-level industries by the institutional supply of education, representing a disruptive change in the quality of agricultural talent. In contrast, the number of graduates reflects multiple factors such as market demand and social atmosphere, which reflect the qualitative transformation of agricultural talent less clearly.

4.4. Control Variables

The primary purpose of selecting control variables in this study is to avoid potential interference from confounding factors with the empirical results. After reviewing relevant studies on the impact of human capital on industrial structure upgrading [17,18,33] and integrating economic theories related to industrial development, we selected control variables from three dimensions. (1) Economic Foundation and Resource Allocation Dimension: The level of economic development determines the capital accumulation threshold for industrial structure upgrading. The fiscal self-sufficiency rate reflects the financial capacity boundary of local governments to intervene in industrial policies. The degree of financial development indicates the efficiency of market-based capital allocation. Together, these three variables control the foundational conditions for industrial development in rural counties. (2) Factor Endowment and Production Structure Dimension: The level of grain production, as a core indicator of agricultural strengths in rural counties, constrains the marginal output elasticity of agricultural innovation talent. The level of industrialization characterizes the existing path dependency of non-agricultural industries in rural counties. These two variables help separate the interactive effects of the benefits of agricultural innovation talent and historical industrial inertia on upgrading. (3) Education Popularization and Social Technical Conditions: The iteration of industrial models requires not only the impetus of innovative talents but also a workforce with a certain level of cultural literacy as the supportive infrastructure. Controlling for the degree of education popularization can mitigate differences in the adaptability of rural county industries to new technologies and models. The level of informatization can exclude the interference of information connectivity with the diffusion efficiency of industrial innovation knowledge.

4.5. Mechanism Variables

To empirically test Hypotheses 2 to 4, proposed in the theoretical analysis, this study selects the following mechanism variables at the county level. (1) Knowledge Creation and Industrial Integration: The first category of mechanisms involves promoting industrial integration through knowledge creation. This study uses the degree of tertiary industry integration (Transformation) and night-time light intensity (Light) to measure the status of industrial integration. (2) Knowledge Dissemination and Demand Responsiveness: The second category of mechanisms involves responding to demand information through knowledge dissemination. Retail density (Density) and per capita retail sales (Average) are used to measure demand vitality. (3) Knowledge Application and Digitally Intelligent Factor Integration: The third category of mechanisms involves guiding the integration of digital and intelligent factors through the application of knowledge. Robots and e-commerce are typical representatives of the integration of digital and intelligent factors with industries, but data on these are not readily available at the county level; as such, this study uses robot installation density (Robot) and the number of enterprises engaged in e-commerce transactions (E-commerce), both derived from provincial-level GDP data, to measure the accessibility of digital and intelligent factors.

4.6. Descriptive Statistic Analysis

Table 1. Definition and descriptive statistics variables.

Variables	Meaning and Assignment of Variables	Obs	Mean	S.D.
Dependent variable
Upgrade	The relative proportion of the value of output of the tertiary sector to that of the secondary sector at the county level.	15,939	1.364	1.421
Independent variable
Talent	Does the city to which the county belongs have a “World-Class Agricultural Disciplines” construction project, and is the year 2017 or later? 1 = Yes; 0 = No.	15,939	0.018	0.133
Control variables
Development	The ratio of county-level GDP to county-level total population.	15,939	4.545	4.371
Fiscal	The ratio of county-level fiscal revenue to county-level fiscal expenditure.	15,939	0.329	0.234
Finance	The ratio of the year-end balance of financial institution loans at the county level to county-level fiscal expenditure.	15,939	0.76	0.479
Education	The ratio of the total number of primary and secondary school students at the county level to the total population at the county level.	15,939	0.118	0.05
Inform	The ratio of the number of fixed telephone users at the county level to the total population at the county level.	15,939	0.099	0.097
Food	The ratio of the total grain output at the county level to the total population at the county level.	15,939	0.653	1.048
Industry	The ratio of the number of large-scale industrial enterprises at the county level to the administrative area of the county.	15,939	0.104	0.223
Mechanism variables
Transformation	The ratio of the total value of output of the secondary and tertiary sectors at the county level to the total value of output of the primary sector at the county level.	15,939	8.073	11.327
Light	The average nighttime light intensity at the county level.	15,831	5.861	8.433
Density	The ratio of the total retail sales of consumer goods at the county level to the administrative area of the county.	14,846	0.062	0.13
Avergae	The ratio of the total retail sales of consumer goods at the county level to the total population at the county level.	14,846	1.469	1.326
Robot	The density of robot installations at the county level.	15,939	0.011	0.023
E-commerce	The number of enterprises engaged in e-commerce transactions at the county level.	15,939	0.002	0.004

presents the descriptive statistics of the main variables. The mean value of Upgrade is 1.364, indicating that in all county-level samples, the value added of the tertiary industry is 1.364 times that of the secondary industry. This suggests that at the overall level, the tertiary industry in county-level economies shows signs of rising. However, it has not yet gained a dominant position in county-level economies. Further external support is needed for it to become the pillar of industrial takeoff in counties. Figure 2 illustrates the year-by-year evolution trend of the industrial structure in counties. It is evident that between 2013 and 2020, the share of the value added of the tertiary industry continued to grow relative to that of the secondary industry in counties. That is to say, during these eight years, thecontinuous structural upgrading of county-level industries took place, with the fastest pace of upgrading occurring between 2018 and 2019. However, between 2020 and 2021, the county-level industries experienced a downgrade instead. This was likely closely related to the stagnation of service industry development brought about by the COVID-19 pandemic. The Chinese government’s COVID-19 policy of advocating “no unnecessary going out and no unnecessary contact” ensured the safety and health of residents’ lives while also causing a temporary halt in the expansion of the service industry. The findings of this study on the status of industrial structure upgrading in Chinese counties are basically consistent with existing research [17,18].

4.7. Methods

Basic Models

To evaluate the impact of agricultural innovation talents on county-level industrial upgrading, this paper employs a difference-in-differences (DID) model to conduct a quasi-natural experiment. The two-way fixed effects model specification is presented in Equation (1), which controls for both county-specific and time-invariant heterogeneous effects:

$$Industr{y}_{it}={\beta }_0+{\beta }_{1}Talen{t}_{it}+\alpha X_{it}+{\eta }_i+{\mu }_t+{\epsilon }_{it}$$

(1)

In Equation (1), $Industr{y}_{it}$ represents the degree of industrial sophistication in county i in year t, which reflects the industrial structure of the county. $Talen{t}_{it}$ indicates whether county i is affected by agricultural innovative talents in year t. The determination is based on two dimensions: The first is whether county i belongs to the treatment group, which is identified by whether the city to which county i belongs has a “World-Class Agricultural Disciplines” construction project; if so, county i is considered to be in the treatment group. The second is whether the sample year is 2017 or later. If county i is in the treatment group and the sample year is 2017 or later, then $Talen{t}_{it}$ is assigned a value of 1, and the rest of the samples are assigned a value of 0. ${\eta }_i$ represents the county-level fixed effects, ${\mu }_t$ represents the year-level fixed effects, $X_{it}$ is the control variable, and ${\epsilon }_{it}$ is the random disturbance term. $\beta$ and $\alpha$ are the parameters to be estimated. We focus on ${\beta }_1$, which represents the net effect of agricultural innovative talents on inducing the upgrading of the county-level industrial structure. If this net effect is significant, then the coefficient of ${\beta }_1$ should be significantly positive.

4.8. Mechanism Verification Model

To explore the mechanisms through which agricultural innovative talents may influence the industrial structure upgrading at the county level, this paper further constructs the mechanism-testing model, as shown in Equation (2).

$$Mechanis{m}_{it}={\phi }_0+{\phi }_{1}Talen{t}_{it}+\vartheta X_{it}+{\gamma }_i+{\mu }_t+{\epsilon }_{it}$$

(2)

In Equation (2), $Mechanis{m}_{it}$ represents the mechanism variable for county i in year t and includes six variables: Transformation, Light, Density, Average, Robot, and E-commerce. ${\phi }_0$, ${\phi }_1$, and $\vartheta$ are the parameters to be estimated. This study focuses on the sign and significance of ${\phi }_1$, through which we can observe whether the three categories of mechanisms are empirically valid. This study does not include the mechanism variable $Mechanis{m}_{it}$ from Equation (2) in Equation (1) for re-estimation. The reason is that doing so would conceal serious endogeneity issues, which would be meaningless for verifying the mechanisms. Instead, this study prefers to explain the association between the mechanism variables and the dependent variable through economic logic, and only conducts regression analysis of the core explanatory variable on the mechanism variables at the empirical level.

5. Empirical Results and Discussion

5.1. Baseline Regression

Table 2. Results of basic regression.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Talent	0.2950 ***	0.3063 ***	0.2918 ***	0.2590 ***	0.2900 ***	0.2953 ***	0.3032 ***	0.2811 ***
	(0.0661)	(0.0649)	(0.0660)	(0.0669)	(0.0660)	(0.0661)	(0.0660)	(0.0652)
Development		−0.0440 ***						−0.0664 ***
		(0.0105)						(0.0121)
Fiscal			−0.1488 ***					−0.0759 *
			(0.0503)					(0.0445)
Finance				0.3189 ***				0.2332 ***
				(0.0489)				(0.0473)
Education					−0.5025 **			−1.7589 **
					(0.2122)			(0.6856)
Inform						0.0277		0.3014 ***
						(0.0712)		(0.1154)
Food							0.0934	0.2541 ***
							(0.0577)	(0.0625)
Industry								−0.4339 ***
								(0.1070)
Constant	1.3584 ***	1.5583 ***	1.4074 ***	1.1168 ***	1.4177 ***	1.3556 ***	1.2972 ***	1.5650 ***
	(0.0067)	(0.0485)	(0.0175)	(0.0379)	(0.0260)	(0.0098)	(0.0384)	(0.0833)
Year Effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
County Effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R2	0.6604	0.6643	0.6605	0.6635	0.6606	0.6604	0.6626	0.6754
Obs.	15939	15939	15939	15939	15939	15939	15939	15939

Note: ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. The data in parentheses are robust standard errors; This explanation applies to all the following tables.

presents the benchmark regression results of the impact of agricultural innovative talents on the industrial structure upgrading at the county level. In column (1), control variables are not included to observe the direct causal relationship between the core explanatory variable and the dependent variable. Columns (2) to (8) sequentially introduce control variables to further eliminate potential factors influencing farmers’ income growth and to minimize the interference of exogenous shocks in the quasi-natural experiment. All columns control for county-level and year-level fixed effects.

It can be seen that the regression coefficient of Talent is significantly positive at the 1% level. This indicates that after being influenced by agricultural innovative talents, the degree of industrial sophistication in the county increases, and the industrial structure of the county is upgraded. In terms of the magnitude of the promoting effect, as shown in column (8), the regression coefficient of Talent is 0.2811. This suggests that each unit of agricultural innovative talent can drive an increase of 28.11% in the degree of industrial sophistication. Hypothesis 1 is thus empirically verified, confirming the significant positive impact of agricultural innovative talents on the industrial structure upgrading at the county level.

Essentially, the research findings above reveal a virtuous interaction between agricultural innovative talents and county-level industrial development. This provides empirical evidence for understanding the intrinsic linkages between the cultivation of high-level disciplines and high-quality county-level development. This discovery not only corroborates the perspectives of prior studies but also extends them further.

Smith et al. [80] argued that the subjective initiative of individuals should be emphasized in regional development. Using counties in the UK as examples, they highlighted the significant role of talent specialization in driving industrial development. In the context of China, the empowering effect of talents cultivated through the “world-class discipline” construction projects on industrial structure upgrading is evident [18]. The findings in this part of our study are consistent with these conclusions. However, unlike Smith et al. [80], our study elucidates that the function of agricultural innovative talents is essentially the diffusion of benefits from high-level discipline construction in urban higher education institutions to county-level industries.

Moreover, while Zhao et al. [18] acknowledged the positive externalities of innovative talents, their study was conducted at the provincial level and overlooked the permeation of urban educational benefits to counties. It also failed to address the heterogeneous demands for innovative talents between cities and counties. Our study attempts to fill this research gap.

Our study also responds to two other sets of arguments. One is that urbanization and county-level industrial structure upgrading are complementary [39]. The other is that the benefits of higher education have spatial spillover characteristics [81]. Our findings provide additional evidence from a new perspective to support these two lines of research.

5.2. Robustness Checks

To ensure the robustness of the finding that agricultural innovative talents promote the industrial structure upgrading at the county level, this study employs the following six methods of robustness testing.

5.3. Parallel Trend Test

This study employs a quasi-natural experiment approach for empirical research. Following the methodology of Rambachan and Roth [82], it is necessary to ensure that there are no significant differences in the industrial structure between the treatment and control groups at the county level prior to the influence of agricultural innovative talents. That is, the evolution of the industrial structure in the treatment and control groups should follow a parallel trend before the intervention. If the parallel trend assumption is violated, it may lead to an overestimation of the effect of agricultural innovative talents on the upgrading of the county-level industrial structure.

To address this issue, this study utilizes an event study approach for verification. In the event study methodology, the year 2013 (the first year of the sample study period) is set as the reference point and assigned a value of zero to mitigate the adverse effects of multicollinearity. As shown in Figure 3, before the implementation of the “World-Class Agricultural Disciplines” construction project, there are no significant differences in the trend of industrial structure evolution between the treatment and control groups at the county level, indicating that the parallel trend assumption holds. After the first year of the project’s implementation, the differences in industrial structure between the treatment and control groups become significantly different from zero, suggesting that agricultural innovative talents played a positive role in promoting the upgrading of the county-level industrial structure. The above analysis indicates that the parallel trend test is passed, and the benchmark regression results based on the quasi-natural experiment are robust and reliable.

5.4. Placebo Test

To avoid the possibility that the effect of agricultural innovative talents on the upgrading of the county-level industrial structure observed is an unstable result influenced by random factors, this study employs two types of placebo tests to ensure the robustness of the research conclusions.

On the one hand, this study adopts a random selection method to only obtain a fictitious pseudo-policy implementation time and also to create a pseudo-treatment group with the same number of counties as the actual treatment group. This aims to observe whether the same research conclusions as the benchmark regression can still be obtained under a double-random scenario. If the results are still close to the empirical findings, it would indicate that the credibility of the research conclusions is questionable. Specifically, this study randomly selects the same number of counties as the actual treatment group to form a pseudo-treatment group. It then multiplies these with the pseudo-policy time to generate a pseudo-core explanatory variable. Five hundred regressions are conducted, and the estimated coefficients and p-values are recorded to produce the kernel density distribution shown in Figure 4a.

On the other hand, this study conducts a single-random placebo test. It retains the actual policy implementation year of 2017 and randomly selects the same number of counties as the actual treatment group to form a pseudo-treatment group. By multiplying these with the actual policy year, a “single-random” scenario is achieved. Five hundred regressions are performed, and the estimated coefficients and p-values are recorded to produce the kernel density distribution shown in Figure 4b.

The results indicate that, regardless of the placebo test strategy employed, the vast majority of the regression coefficients of the pseudo-core explanatory variables are significantly different from the actual coefficient of 0.2811, and the p-value distributions are mostly above the horizontal dashed line. This outcome confirms that the placebo tests were passed, and the research conclusions were robust and reliable.

5.5. Variable Replacement

Firstly, considering that talents cultivated by “World-Class Disciplines” possess mobility, which can influence the industrial structure upgrading at the provincial level [18], and given that higher education has positive externalities and knowledge dividends with significant spatial spillover characteristics [20,83], the intellectual resources of agricultural innovative talents may affect the industrial structure upgrading in all counties within the province where the city with a “World-Class Agricultural Disciplines” is located. Based on this consideration, this study replaces the measurement of the core explanatory variable by considering all counties in the province where the city with a “World-Class Agricultural Disciplines” is located as the treatment group. Samples in the treatment group from 2017 and later are assigned a value of 1, while all other samples are assigned a value of 0. The empirical results of this part are presented in columns (1) and (2) of

Table 3. Variable substitution test.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
	Core Explanatory Variables		Dependent Variables		Control Variables
Talent	0.075 ***	0.088 ***	0.166 ***	0.146 ***	0.307 ***	0.294 ***
	(0.026)	(0.026)	(0.034)	(0.033)	(0.067)	(0.067)
Constant	1.348 ***	1.549 ***	0.997 ***	1.116 ***	1.135 ***	1.666 ***
	(0.009)	(0.083)	(0.004)	(0.045)	(0.252)	(0.218)
Control Variables	No	Yes	No	Yes	Yes	Yes
Year Effect	Yes	Yes	Yes	Yes	Yes	Yes
County Effect	Yes	Yes	Yes	Yes	Yes	Yes
R2	0.660	0.675	0.690	0.700	0.685	0.685
Obs.	15,939	15,939	15,939	15,939	14,000	14,000

In columns (1) to (6) of Table 3, the regression coefficient of Talent remains significantly positive at the 1% level. This result is consistent with the findings in the benchmark regression, indicating that the research conclusions obtained from the benchmark regression are robust and reliable.

.

Secondly, this study changes the measurement of the dependent variable to the relative level of industrial structure upgrading at the county level, that is, the ratio of the degree of industrial sophistication in the county to its mean value in the same year. The purpose of this approach is to effectively eliminate the interference of macro-level factors on empirical results. The empirical results of this part are presented in columns (3) and (4) of Table 3.

Lastly, given that innovation is a vital driving force for industrial structure upgrading [29,33], the positive effect of the “world-class discipline” construction project on county-level industrial structure upgrading may be overestimated. To control for the impact of innovation quantity and quality on county-level industries, this study further adds the city-level patent authorization score and the number of invention patents to the benchmark regression model. The regression results are presented in columns (5) and (6) of Table 3, respectively. The data are sourced from the China Regional Innovation and Entrepreneurship Index (IRIEC) compiled by the Center for Enterprise Big Data Research at Peking University.

5.6. Sample Replacement

On the one hand, to mitigate the systematic differences between the treatment and control groups prior to the implementation of the “World-Class Agricultural Disciplines” construction project, this study first employs propensity score matching (PSM) with a 1:1 matching ratio. After eliminating the samples that fail to match successfully, the regression is re-conducted. Figure 5 illustrates the sample distribution before and after matching, showing that the differences in sample distribution between the treatment and control groups are significantly reduced after matching. Columns (1) and (2) of

Table 4. Sample replacement test.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	PSM-DID		Isolating the Treatment Effect from Other Empirical Confounders
Talent	0.2929 ***	0.3149 ***	0.3594 ***	0.3526 ***	0.3509 ***	0.3737 ***	0.1224 **	0.1146 **
	(0.0664)	(0.0641)	(0.0710)	(0.0698)	(0.0841)	(0.0825)	(0.0554)	(0.0564)
Constant	1.3278 ***	1.5059 ***	1.3621 ***	1.5559 ***	1.2149 ***	1.0910 ***	1.2633 ***	1.3704 ***
	(0.0067)	(0.0954)	(0.0067)	(0.0835)	(0.0067)	(0.0771)	(0.0062)	(0.0682)
Control Variables	No	Yes	No	Yes	No	Yes	No	Yes
Year Effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
County Effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R2	0.6730	0.6881	0.6606	0.6753	0.6948	0.7152	0.6589	0.6704
Obs.	14,864	14,864	15,777	15,777	11,095	11,095	14,514	14,514

present the estimation results of the PSM-DID approach. The regression coefficient of Talent remains significantly positive at the 1% level, and the estimates are close to those of benchmark regression, indicating that the conclusions of the benchmark regression are robust and reliable.

On the other hand, the “Reform Plan for National Urban-Rural Integration Development Experimental Zones” issued by China in 2019 announced a list of national urban–rural integration development experimental zones. This policy focuses on promoting the connection of urban and rural elements and may accelerate the process of county-level industrial transformation and upgrading. To avoid overestimating the effect of agricultural innovative talents, this study excludes the counties involved in this policy and re-conducts the regression. Meanwhile, the rise of e-commerce has had an impact on the evolution of county-level industrial structure and farmers’ employment [5,84]. This study also excludes the counties involved in the “Comprehensive Demonstration Project of E-commerce in Rural Areas” announced by China’s Ministry of Commerce and re-conducts the regression. Columns (3) to (6) of Table 4 present the relevant results after excluding the impact of the above two policies. Compared with the benchmark regression results, the sign and significance of the Talent regression coefficient remain unchanged, indicating that the research conclusions are not affected by other policies and are highly credible.

In addition, we further exclude the impact of agricultural fiscal support on the industrial structure upgrading in counties. One of our concerns is that a larger scale of fiscal support for agriculture can provide sufficient funds for the construction of agricultural infrastructure, the promotion of agricultural technology, and the training of rural talents, thereby improving the production conditions of agriculture, enhancing the production efficiency of agriculture, promoting agricultural modernization, and laying the foundation for the upgrading of the industrial structure. At the same time, fiscal support for agriculture can support new types of agricultural business entities and rural industrial integration projects, promote the integrated development of primary, secondary, and tertiary industries in rural areas, extend the agricultural industry chain, create more non-agricultural employment opportunities, attract the return of rural labor, optimize the allocation of rural labor resources, and promote the transformation of county-level industries from traditional agriculture to diversified and high value-added industries, thereby accelerating the upgrading of the industrial structure in counties. To eliminate this confounding factor of fiscal support for agriculture and to ensure that the results of the benchmark regression are not “spurious effects,” we use the proportion of agricultural-related expenditure in the total fiscal expenditure as a measure of the scale of fiscal support for agriculture. We divide the sample into deciles each year based on this proportion and assign values from 1 to 10. Our empirical strategy is to exclude samples with a fiscal support scale of 10 for agriculture and re-conduct the regression. This ensures the credibility of the empirical results. Columns (7)–(8) of Table 4 show that even after excluding samples with a large scale of fiscal support for agriculture, the coefficient of Talent in the regression remains unchanged in terms of its sign and significance. That is, the conclusion of the benchmark regression is robust and reliable.

5.7. Model Replacement—Double Machine Learning

To address the potential issues of “dimensionality curse” caused by redundant control variables and estimation bias due to linear model specification in causal identification, this study employs the double machine learning (DML) model proposed by Chernozhukov et al. [85] for estimation. This study utilizes Lasso and Gradboost as the machine learning algorithms for DML and sets the sample splitting ratios at 1:3 and 1:4. The rationale for selecting these two algorithms is that Lasso excels in feature selection and is adept at handling high-dimensional data, while Gradboost optimizes the model through gradient boosting, effectively reducing bias. Therefore, the chosen algorithms are representative.

Columns (1) to (4) of

Table 5. Double machine learning model test.

Vairables	(1) Lasso (kfolds = 4)	(2) Lasso (kfolds = 5)	(3) Gradboost (kfolds = 4)	(4) Gradboost (kfolds = 5)
Talent	0.2291 ***	0.2254 ***	0.5246 ***	0.5509 ***
	(0.0640)	(0.0684)	(0.0916)	(0.0956)
Constant	−0.0007	0.0012	0.0023	0.0046
	(0.0070)	(0.0072)	(0.0079)	(0.0079)
Control Variables	Yes	Yes	Yes	Yes
Year Effect	Yes	Yes	Yes	Yes
County Effect	Yes	Yes	Yes	Yes
Obs.	15,939	15,939	15,939	15,939

present the DML estimation results under the two algorithms. It is evident that regardless of the sample splitting ratio used, the regression coefficient of the core explanatory variable Talent remains significantly positive, and the coefficient size is consistent with the benchmark regression results. This indicates that the research conclusions of this study are robust and reliable.

5.8. Endogeneity Test

Employing two-stage least squares (2SLS) to further eliminate potential endogeneity bias is beneficial for ensuring the credibility of the research conclusions. This study selects the development level of the non-state-owned economy at the provincial level as an instrumental variable (IV_Non-sate) for agricultural innovative talents. The rationality of this choice is twofold: First, in terms of relevance, there is a close relationship between talent quality and enterprise development [19]. In China, small and medium-sized enterprises (SMEs) cannot compete with large enterprises across all stages of the industry. Their survival depends on providing upstream products to large enterprises. The specialized, refined, and unique business models of SMEs require various innovative talents for support. For example, digital agriculture is mostly driven by private enterprises [16]. Therefore, agricultural innovative talents are positively correlated with the development of the non-state-owned economy. Second, regarding exogeneity, the development of the non-state-owned economy at the provincial level is unlikely to be reversely affected by the industrial structure upgrading at the county level. It is also difficult for it to directly influence the evolution of the county-level industrial structure without the channel of agricultural innovative talents. This is because the comparative advantage of county-level economic development, compared to cities, lies in utilizing agriculture, and agricultural innovative talents are the special factors for leveraging this advantage. Thus, the selection of the instrumental variable is reasonable.

Table 6. Endogeneity test.

Varibales	(1) First Stage	(2) Second Stage	(3) First stage	(4) Second Stage
IV_Non-sate	0.0756 ***		0.0849 ***
	(0.0116)		(0.0118)
Talent		5.658 ***		5.273 ***
		(1.759)		(1.492)
Control	No	No	Yes	Yes
Year Effect	Yes	Yes	Yes	Yes
County Effect	Yes	Yes	Yes	Yes
Kleibergen–Paap rk Wald F	42.265 {16.38}		51.861 {16.38}
Kleibergen–Paap rk LM	48.382 ***		59.390 ***
Obs.	15,939		15,939

Note: The values enclosed in curly braces { } represent the critical values at the 10% significance level for the Stock–Yogo test.

reports on the results of the endogeneity test. The Kleibergen–Paap rk Wald F statistics in both types of endogeneity tests are greater than 10, and the Kleibergen–Paap rk LM is also significantly positive. These results empirically validate the appropriateness of the instrumental variable selection. Meanwhile, the regression coefficient of the core explanatory variable Talent remains significantly positive, indicating that the positive effect of agricultural innovative talents on county-level industrial structure upgrading is substantial and that the conclusions drawn from the benchmark regression are robust and reliable.

5.9. Mechanism

Creating Knowledge: Promoting Industrial Integration

Columns (1) to (4) of

Table 7. Mechanism test for promoting industrial convergence.

Variables	(1) Transformation	(2) Transformation	(3) Light	(4) Light
Talent	2.9735 ***	2.7006 ***	1.3986 ***	1.2072 ***
	(0.5360)	(0.5148)	(0.2178)	(0.2080)
Constant	8.0199 ***	5.5678 ***	5.8357 ***	5.2585 ***
	(0.0280)	(0.6069)	(0.0150)	(0.1964)
Control Variables	No	Yes	No	Yes
Year Effect	Yes	Yes	Yes	Yes
County Effect	Yes	Yes	Yes	Yes
R2	0.9090	0.9271	0.9546	0.9563
Obs.	15,939	15,939	15,831	15,831

present the results of the mechanism test for promoting industrial integration. The results show that the regression coefficient of the core explanatory variable Talent is significantly positive at the 1% level. When control variables are included, agricultural innovative talents increase the degree of tertiary industry integration in counties by 2.7006 and enhance the nighttime light intensity in counties by 1.2072. This indicates that agricultural innovative talents significantly promote industrial integration through knowledge creation, and the transmission path of “talent–knowledge creation–industry” is unobstructed. The robust existence of this mechanism validates the hypothesis derived from the theoretical analysis in the preceding sections, that is, Hypothesis 2 is confirmed.

The establishment of this mechanism reveals that the role of agricultural innovative talents is not merely to transfer knowledge but also to integrate and create new knowledge essential for industrial development through thoughtful synthesis. The creation of knowledge has profound positive impacts on regional development [81], and the integration of knowledge across fields is conducive to identifying development opportunities within industries [9]. Agricultural innovative talents excel at blending professional and practical knowledge from multiple fields to create new knowledge tailored to local conditions for county-level industrial integration, distinct from urban development models. This, in turn, provides an impetus for counties to pursue a path of specialized industrial structure upgrading.

5.10. Spreading Knowledge: Responding to Information on Needs

Columns (1) to (4) of

Table 8. Mechanism test for responding to information on needs.

Variables	(1) Density	(2) Density	(3) Average	(4) Average
Talent	0.0555 ***	0.0597 ***	0.5406 ***	0.4927 ***
	(0.0065)	(0.0065)	(0.0673)	(0.0570)
Constant	0.0611 ***	0.0369 ***	1.4610 ***	0.2524 ***
	(0.0004)	(0.0079)	(0.0060)	(0.0937)
Control Variables	No	Yes	No	Yes
Year Effect	Yes	Yes	Yes	Yes
County Effect	Yes	Yes	Yes	Yes
R2	0.8618	0.8747	0.7068	0.8424
Obs.	14,841	14,841	14,841	14,841

present the results of the mechanism test for responding to information on needs. The results show that the regression coefficient of the core explanatory variable Talent is significantly positive at the 1% level. When control variables are included, agricultural innovative talents increase the retail density in counties by 0.0597 and enhance the per capita retail sales in counties by 0.4927. This indicates that agricultural innovative talents significantly promote the response to demand information through knowledge dissemination, and the transmission path of “talent–knowledge dissemination–industry” is unobstructed. The robust existence of this mechanism validates the hypothesis derived from the theoretical analysis in the preceding sections, that is, Hypothesis 3 is confirmed.

The establishment of this mechanism reveals that agricultural innovative talents can transform demand into practical knowledge and disseminate it to drive the development of county-level industries. Responding to demand information is a complex endeavor, involving multiple aspects such as market prices, income, and consumer behavior [70]. However, the greater challenge lies in how to convert this demand information into valuable, transferable knowledge, link it with the comparative advantages of counties, transform it into actionable industrial projects, and ultimately find channels for market connection and sales. This is because the flow of knowledge can bring about innovative changes in industries [10], and this challenge is precisely the area where agricultural innovative talents excel.

5.11. Operating Knowledge: Guiding the Embedding of Digital Intelligence Elements

Columns (1) to (4) of

Table 9. Mechanism test for guiding the embedding of digital intelligence elements.

Variables	(1) Robot	(2) Robot	(3) E-commerce	(4) E-commerce
Talent	0.0108 ***	0.0121 ***	0.0010 ***	0.0011 ***
	(0.0018)	(0.0018)	(0.0002)	(0.0002)
Constant	0.0107 ***	−0.0014	0.0024 ***	0.0017 ***
	(0.0001)	(0.0022)	(0.0000)	(0.0002)
Control Variables	No	Yes	No	Yes
Year Effect	Yes	Yes	Yes	Yes
County Effect	Yes	Yes	Yes	Yes
R2	0.7179	0.7961	0.9284	0.9435
Obs.	15,939	15,939	15,939	15,939

present the results of the mechanism test for guiding the embedding of digital intelligence elements. The results show that the regression coefficient of the core explanatory variable Talent is significantly positive at the 1% level. When control variables are included, agricultural innovative talents increase the penetration rate of robot installations in counties by 0.0121 and the number of e-commerce enterprises by 0.0011. This indicates that agricultural innovative talents significantly facilitate the embedding of numint factors into county-level industries through the application of knowledge, and the transmission path of “talent–application of knowledge–industry” is unobstructed. The robust existence of this mechanism validates the hypothesis derived from the theoretical analysis in the preceding sections, that is, Hypothesis 4 is confirmed.

The establishment of this mechanism reveals that agricultural innovative talents are the facilitators for the embedding of numint factors into county-level industries. The utilization of numint factors faces certain barriers, leading to the emergence of digital divides [77]. In the agricultural sector, these issues are compounded by the potential for digital and intelligent tools to benefit only large agricultural enterprises [16]. In such a scenario, to ensure that numint factors benefit all market entities in counties equally, agricultural innovative talents need to optimize industrial layout and identify the convergence points between traditional and new-quality factors according to the actual situation. This process is also a transformation of county-level industries through the application of knowledge by agricultural innovative talents, with the aim of retaining only the positive externalities of digital tools in the process of upgrading the county-level industrial structure.

5.12. Heterogeneity Analysis

Within the research framework of this study, agricultural innovative talents can provide external impetus for the upgrading of the county-level industrial structure through knowledge creation, dissemination, and application. Based on this, the extent to which counties absorb this external impetus also depends on endogenous factors within the counties. For example, customer-centric service quality improvement is not only an important way to increase product added value [58] but also a reflection of the necessity for industrial integration, response to demand information, and guidance for the embedding of digital and intelligent factors. However, if there is a lack of sufficient purchasing power in the county, or if income inequality is too large, leading to a decrease in the county’s marginal propensity to consume, the demand for high-quality services in the county may be reduced. This situation, when fed back to the supply side, will weaken the motivation of agricultural innovative talents to transform county-level industries. This is because choosing not to upgrade industries may be the Nash equilibrium strategy for county-level industrial development in this context, which is a result of demand structure lock-in. Moreover, the existing industrial structure of the county also affects the functioning of agricultural innovative talents. The rigid allocation mechanism of factor resources squeezes the space for agricultural innovative talents to transform county-level industries, thereby leading to the emergence of a “Matthew effect.”

It is evident that incorporating income inequality, household savings, and the existing industrial structure into the research framework is necessary. The first two factors examine the conditions for knowledge to be transformed into value, and their heterogeneity test results are presented in

Table 10. Heterogeneity tests based on conditions for the conversion of knowledge into value.

Variables	(1) Gap	(2) Savings
Talent	0.4068 ***	0.1941 ***
	(0.0872)	(0.0663)
Talent × Gap	−0.4009 ***
	(0.1029)
Talent × Savings		0.4673 **
		(0.1941)
Constant	1.5754 ***	1.5169 ***
	(0.0891)	(0.0875)
Control Variables	Yes	Yes
Year Effect	Yes	Yes
County Effect	Yes	Yes
R2	0.6725	0.6759
Obs.	13,554	15,894

. The third factor reflects the heterogeneity of industrial endowment, and its heterogeneity test results are presented in

Table 11. Heterogeneity test based on industrial endowment conditions.

Variables	(1) 25% Quartile	(2) 50% Quartile	(3) 75% Quartile
Talent	0.1344 ***	0.1206 ***	0.3279 ***
	(0.0240)	(0.0399)	(0.1042)
Constant	0.6904 ***	1.0335 ***	1.5750 ***
	(0.0163)	(0.0267)	(0.0778)
Control Variables	Yes	Yes	Yes
Year Effect	Yes	Yes	Yes
County Effect	Yes	Yes	Yes
R2	0.1263	0.1919	0.1762
Obs.	15,939	15,939	15,939

. In terms of heterogeneity definition, this study considers samples with urban–rural income gaps and household savings above the 70th percentile in the county as samples with large income gaps and abundant household savings, respectively assigning them a value of 1, while other samples are assigned a value of 0, thereby constructing two categorical variables composed of 0 and 1. The heterogeneity of industrial endowment is reflected in the division of the degree of industrial sophistication according to the 25th, 50th, and 75th percentiles, followed by unconditional quantile regression. The data still come from the China County Statistical Yearbook.

From the heterogeneity test results in Table 10, it can be seen that the regression coefficient of the core explanatory variable Talent remains significantly positive, indicating that even with heterogeneity in the conditions for knowledge to be transformed into value, the promoting effect of agricultural innovative talents on the upgrading of the county-level industrial structure still exists significantly. Looking at the interaction terms, Talent × Gap is significantly negative, and Talent × Savings is significantly positive. This suggests that in counties with smaller income gaps and more abundant household savings, the promoting effect of agricultural innovative talents on the upgrading of the county-level industrial structure can be more fully realized. There are mainly two possible reasons for this: On the one hand, a smaller income gap means higher income equality among county residents, and a higher level of household savings usually indicates a stronger sense of well-being. Both income equality and well-being are positively correlated with trust [86]. Therefore, in counties with smaller income gaps and higher levels of household savings, the trust cost between agricultural innovative talents and upstream and downstream industrial chain entities is lower. This efficient connectivity of social networks allows knowledge for industrial integration to spread quickly [51], enabling the role of agricultural innovative talents to be more fully realized in such an environment. On the other hand, although agricultural innovative talents can respond to demand information and guide the embedding of digital and intelligent factors through knowledge dissemination and application, the transformation of these factors into real industrial scenarios requires the introduction of capital from outside as support. Once the conditions for knowledge to be transformed into value are missing, it is difficult to form industrial clusters and risk buffering mechanisms, and various types of capital will lack the motivation to enter county-level industries.

Table 11 presents the results of the heterogeneity test for industrial endowment. In samples at the 25th, 50th, and 75th percentiles of industrial sophistication, the regression coefficient of Talent is significantly positive at the 1% level, indicating that the promoting effect of agricultural innovative talents on the upgrading of the county-level industrial structure exists significantly across all three percentiles. Further observation reveals that the regression coefficient of Talent is smallest at the 50th percentile and largest at the 75th percentile, approximately 2.4 and 2.7 times the coefficients at the 25th and 50th percentiles, respectively. This result suggests that the promoting effect of agricultural innovative talents on the upgrading of the county-level industrial structure is strongest at the 75th percentile, far exceeding the other two percentiles, revealing a certain degree of “Matthew effect” in this promoting effect.

The possible reasons for this are as follows: First, counties with a higher degree of industrial sophistication provide convenience for agricultural innovative talents to use knowledge to transform industries. This is because the high-end links in the industrial chain necessarily require the infrastructure and production relations of the mid- and low-end links to be compatible with them, and counties with a certain industrial transformation foundation will show advantages in this regard. Second, the structural differences in county-level industries directly affect the willingness of talents to enter. The connection between talent resources and county-level industries is a process of mutual selection, and the effectiveness of knowledge transformation and application varies among talents at different levels. For example, the focus of industrial integration is opportunity identification, in which composite talents have a clear advantage [9]. However, counties with a low degree of industrial sophistication find it difficult to attract such talents. Moreover, some regions with a shortage of talents may misuse innovative talents, and the misallocation of talent resources can also have a negative impact on economic development [52]. These are the reasons for the differences in the magnitude of the effect of agricultural innovative talents across different percentiles. Third, the industrial structure shapes the inertia of decision-makers’ thinking. Counties with a low degree of industrial sophistication usually lack the motivation to implement discretionary policies. If standardized policies are used to solve the heterogeneous adoption of agricultural technologies, it may have a negative impact on agricultural development [56], thereby hindering counties from using their agricultural advantages to promote industrial structure upgrading.

5.13. Futher Analysis: Spatial Spillover Effect

In the context of increasingly close regional economic interactions, county-level industrial structure upgrading does not exhibit spatial isolation. Particularly, given the spillover effects of knowledge [20], the benefits of agricultural innovation talents can not only drive industrial structure upgrading within counties but also facilitate coordinated upgrading of industrial structures in neighboring counties. Clarifying this issue can not only eliminate the possibility of omitted variable bias caused by spatial factors but also help to examine whether the empowering effect of agricultural innovation talents on county-level industrial structure upgrading can transcend geographical constraints to achieve shared benefits with adjacent counties. To this end, this paper further investigates the spatial spillover effects.

We employ the Wald test, Lratio test, Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC) as the basis for model selection. The results indicate that the Spatial Durbin Model (SDM) is the most appropriate, leading to the construction of the model presented in Equation (3). Specifically, we jointly use the Wald test and Lratio test to assess the significance of spatial lag effects (SARs) and spatial error effects (SEMs). By comparing the goodness-of-fit of different models through AIC and BIC, we ultimately select the SDM because it can simultaneously capture the spatial autocorrelation of the dependent variable and the spatial spillover effects of the explanatory variables.

$$\begin{array}{c}\hfill Industr{y}_{it}={\varphi }_0+{\varphi }_{1}wIndustr{y}_{it}+{\varphi }_{2}Talen{t}_{it}+{\varphi }_{3}wTalen{t}_{it}\hfill \\ \hfill \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+{\vartheta }_1{X}_{it}+{\vartheta }_{2}w{X}_{it}+{\eta }_i+{\mu }_t+{\epsilon }_{it}\hfill \end{array}$$

(3)

In Equation (3), ${\varphi }_M$ and ${\vartheta }_N$ (M = 0, 1, 2, 3; N = 1, 2) are the parameters to be estimated. $wIndustr{y}_{it}$ represents the spatial lag term of $Industr{y}_{it}$, $wTalen{t}_{it}$ represents the spatial lag term of $Talen{t}_{it}$, and $w{X}_{it}$ represents the spatial lag terms of $X_{it}$. $w$ is the spatial inverse distance weight matrix, and the interpretations of the remaining variables are the same as those in the previous text.

Our empirical strategy employs the partial differentiation method to decompose the total effect into direct and indirect effects, with a particular focus on examining the cross-regional impacts of agricultural innovation talent through spatial spillover channels. The direct effect captures the influence of local agricultural innovation talent on industrial structure upgrading within their own counties, while the indirect effect measures the spillover impacts of local innovation talent on the industrial structure development of neighboring counties.

This study first examines the spatial correlation of the degree of industrial sophistication at the county level using Moran’s I index. As shown in Figure 6, the Moran’s I index was 0.048 in 2013 and 0.060 in 2021. The majority of counties are located in the first quadrant of high–high clustering and the third quadrant of low–low clustering. This indicates a relatively clear positive spatial correlation in the upgrading of the county-level industrial structure, indirectly suggesting that the factors driving the upgrading of the county-level industrial structure may have spatial spillover characteristics between counties.

The results of the spatial spillover effect decomposition are presented in

Table 12. Test for spatial spillover effect.

	(1) Direct	(2) Indirect	(3) Total	(4) Direct	(5) Indirect	(6) Total
Talent	0.2565 ***	37.6505 ***	37.9070 ***	0.1376 *	86.2270 ***	86.3646 ***
	(0.0813)	(7.6310)	(7.6284)	(0.0770)	(21.0062)	(21.0078)
rho	0.9203 ***			0.9468 ***
	(0.0178)			(0.0104)
sigma2_e	0.6951 ***			0.6606 ***
	(0.1054)			(0.1048)
Control Variables	No			Yes
Year Effect	Yes			Yes
County Effect	Yes			Yes
R2	0.0102			0.0459
Obs.	15,786			15,786

. For both the direct effect, indirect effect, and total effect, the regression coefficient of the core explanatory variable Talent is significantly positive. This indicates that the benefits of agricultural innovative talents for the upgrading of the county-level industrial structure are not only a boon to the local area but also have a radiating and driving effect on the industrial structure upgrading of neighboring counties. This result demonstrates that the spatial spillover effect of agricultural innovative talents is significant.

This finding is consistent with both theory and common understanding. Knowledge is fluid and can influence industrial innovation [10]. The mutual penetration of knowledge between counties bidirectionally increases the knowledge stock for industrial structure upgrading in both areas, including the practical knowledge accumulated through learning-by-doing [41]. This plays an important role in promoting the development of county-level industries by leveraging agricultural strengths. Therefore, the existence of this spatial spillover effect is expected.

6. Conclusions and Policy Implications

6.1. Main Conclusions

In China, the upgrading of the industrial structure at the county level is an important driver for high-quality economic development, comprehensive rural revitalization, and the common prosperity of all people. However, county-level industrial development should follow a path distinct from that of cities, requiring the reshaping of county-level industrial structure upgrading models supported by agricultural advantages through the power of knowledge. The important role of agricultural innovative talents in this process has yet to be fully explored.

This study primarily utilizes data from the China County Statistical Yearbook and employs a quasi-natural experiment approach to empirically examine the effect of agricultural innovative talents on the upgrading of the county-level industrial structure. It also investigates the underlying logic within the theoretical framework of “talent–knowledge–industry.” The findings indicate that agricultural innovative talents cultivated by the “World-Class Agricultural Disciplines” project significantly promote the upgrading of the county-level industrial structure. The core mechanism lies in knowledge empowerment. Agricultural innovative talents drive the integration of the tertiary industry, optimize the matching of supply and demand, and guide the embedding of digital and intelligent factors in counties through the creation, dissemination, and application of knowledge, all of which are important driving forces for the upgrading of the county-level industrial structure.

Further heterogeneity tests reveal that the conditions for the transformation of knowledge into value and the characteristics of county-level industrial endowment affect the magnitude of the promoting effect of agricultural innovative talents on the upgrading of the county-level industrial structure. Specifically, this promoting effect is stronger in counties with smaller urban–rural income gaps, more abundant household savings, and higher degrees of industrial sophistication. This finding indicates that the demand environment and the “Matthew effect” also influence the magnitude of the role of agricultural innovative talents. Additionally, the promoting effect of agricultural innovative talents on the upgrading of the county-level industrial structure has significant spatial spillover characteristics. The benefits of agricultural innovative talents not only affect the county itself but also spread to neighboring counties.

Future research can expand the sample scope, as the unique features of counties and agricultural development across different countries may limit the universal applicability of the conclusions. Future studies can also deepen the analysis of industrial linkage logic by investigating knowledge spillovers between and within industries to better understand how agricultural innovation talent influences the quality transformation of county-level industries.

6.2. Theoretical Contributions

Based on the above findings, this study makes three major theoretical contributions: First, previous studies have primarily focused on the impact of education and knowledge on industrial structure upgrading [17,25], while a few have examined how “world-class discipline” construction projects promote industrial structure upgrading [18]. The former overlooks the importance of high-level discipline construction, while the latter fails to recognize that the path for county-level industrial structure upgrading should differ from that of cities, with agricultural advantages serving as a comparative advantage for cultivating industrial competitiveness. In light of this, this study seizes the opportunity of the “World-Class Agricultural Disciplines” construction project and examines whether agricultural innovative talents can drive the upgrading of the county-level industrial structure and whether this effect has spatial spillover characteristics, aiming to fill the research gap.

Second, although the importance of knowledge for industrial development has been widely discussed [16,24], no systematic analytical framework has yet been established linking discipline construction and industrial development. This study constructs a theoretical framework of “talent–knowledge–industry,” highlighting the irreplaceable role of knowledge empowerment by agricultural innovative talents in driving industrial development based on agricultural advantages. It explores the concrete mechanisms of knowledge empowerment through knowledge creation, dissemination, and application. These mechanisms mainly reveal the role of knowledge empowerment in industrial integration, supply–demand alignment, and the effective utilization of digital and intelligent factors, providing a new perspective for understanding how the benefits of agricultural innovative talents permeate county-level industries.

Third, this study builds a “core–periphery” support structure for the upgrading of the county-level industrial structure. The “core” is agricultural innovative talents, while the “periphery” consists of the theoretical interactions among income inequality, savings, industrial endowment, and knowledge empowerment mechanisms. In this theoretical interaction, the effects of income inequality and savings can be seen as conditions for the transformation of knowledge into value, while the impact of industrial endowment is an important window for observing whether knowledge empowerment exhibits a “Matthew effect.” These conclusions not only have practical value for developing economies like China but also provide a theoretical basis for the in-depth integration of agricultural innovative talents in county-level industrial structure upgrading. They also offer enlightening insights for the integrated research of national economics, labor economics, and industrial economics within the same framework.

6.3. Managerial Implications

Based on the above research findings, this study proposes the following policy recommendations: First, policymakers should increase their emphasis on the cultivation of agricultural innovative talents. As revealed in this study, agricultural innovative talents play an important role in the upgrading of the county-level industrial structure, and it is feasible to incubate such talents through the “World-Class Agricultural Disciplines” construction project. Therefore, agricultural and forestry universities should focus on shaping advantageous agricultural disciplines, rather than being keen on comprehensive discipline optimization. Instead, they should pursue a specialized, high-level discipline construction path characterized by unique strengths. In cities that have not yet had the opportunity to participate in the “World-Class Agricultural Disciplines” construction project, it is encouraged to adopt a cross-university pairing and support model, guiding stronger universities to help weaker ones build advantageous disciplines. Alternatively, universities involved in the “World-Class Agricultural Disciplines” construction project could be encouraged to open branches in different locations to extend the benefits of agricultural innovative talents to more counties. Moreover, the government should shift away from the mindset of primarily using industrial innovative talents to drive county-level industrial development. It should re-evaluate the prominent advantages of agriculture in counties, actively invest financial resources, and use policy incentives to attract agricultural innovative talents, providing them with attractive salaries and enhancing their long-term willingness to leverage agricultural advantages to reshape the county-level industrial competitive edge.

Second, this study reveals that agricultural innovative talents play a crucial role in knowledge empowerment, which is the core driving force for the upgrading of the county-level industrial structure. Given this, on the one hand, the government should place agricultural innovative talents in key positions where old and new knowledge are integrated. In areas where the comparative advantage of county-level agriculture has not yet been fully explored, agricultural innovative talents should be given the space to experiment with combining their knowledge with practical experience to create new knowledge for industrial integration. Agricultural innovation talent think tanks should be established at the county level to design the best plans for county-level industrial layout based on local conditions and market demands. On the other hand, the government should support the introduction of digital and intelligent factors into county-level industries, guiding agricultural innovative talents to act as facilitators for the embedding of these factors. They should be encouraged to use their knowledge to address both explicit and implicit bottlenecks in the integration of digital and intelligent factors with county-level industries.

Third, enhance cross-county knowledge exchange oriented by agricultural advantages and provide external support for agricultural innovative talents to empower county-level industrial development. First, county-level industrial development experience exchange meetings should be organized, focusing on the integration, transformation of agricultural advantages, and their role in driving industrial structure upgrading. Promotable experiences should be included in a national case library. Second, efforts should be made to narrow the urban–rural income gap through stable employment, skill enhancement, and strengthened social security. Gradual exploration should also be conducted on how to implement a universal shareholding plan for state-owned enterprises to increase residents’ property income and ensure they have sufficient savings. Finally, given that the function of agricultural innovative talents exhibits a certain degree of “Matthew effect,” policymakers should adopt a coordinated policy design approach. They should not only focus on the role of agricultural innovative talents in county-level structural upgrading but also use a comprehensive set of policies covering multiple dimensions, including finance, innovation, and domestic and foreign capital linkage, to simultaneously and directionally promote county-level structural upgrading and induce county-level industries to enter a “spiral” path of structural upgrading.

For the world, compared to China, regions such as Africa, Latin America, and Eastern Europe have shown unique progress and potential in the cultivation and utilization of agricultural innovation talent, and China’s experience can provide valuable insights to support their development.

Taking Africa as an example, the continent is gradually strengthening the training and support of agricultural innovation talent in regions rich in distinctive agricultural resources. Africa possesses abundant biodiversity resources, such as Ethiopia’s coffee industry and Kenya’s flower industry. These regions are progressively establishing agricultural education and training systems tailored to their unique industries, cultivating professionals who understand local agricultural resources and market demands. These experts are dedicated to unlocking the market potential of specialty agricultural products such as coffee and flowers, developing high-value-added goods, and enhancing farmers’ production techniques and management capabilities through knowledge dissemination and skills training.

In Latin America, countries are increasingly drawing on China’s experience in optimizing industrial organization and market circulation during the process of agricultural industrialization. Brazil, a major agricultural country in the region, has achieved a high level of agricultural industrialization and boasts a well-developed network of agricultural cooperatives, which bears similarities to China’s agricultural industrialization models. Brazilian agricultural cooperatives play a crucial role in integrating agricultural resources and enhancing market competitiveness. However, they can further develop top-tier agricultural disciplines that align with their comparative advantages, led by universities and research institutions, with agricultural innovation talent serving as the foundation for ensuring long-term agricultural competitiveness. Latin America can also further learn from China’s experience by strengthening policy guidance to promote the integration of agriculture with processing industries, tourism, e-commerce, and other sectors, thereby expanding the agricultural industry chain and value chain. For instance, developing agricultural processing industries can increase the added value of agricultural products; promoting rural tourism can facilitate the integration of agriculture and tourism, boosting farmers’ incomes; and advancing e-commerce can broaden sales channels for agricultural products and improve market circulation efficiency—all of which require the guidance of agricultural innovation talent.

Poland, a key agricultural country in Eastern Europe, is steadily advancing its agricultural digital transformation. The Polish government has increased investment in agricultural information infrastructure, establishing a nationwide agricultural meteorological monitoring network and an agricultural resource management information system. These efforts have enhanced the scientific and precise nature of agricultural production decision-making. Eastern Europe can also further leverage the expertise of agricultural innovation talent to address bottlenecks in various segments of the industry chain, while maintaining existing strengths. By doing so, they can promote intelligent management and precision production in agriculture, improve production efficiency and sustainable development capabilities, and ensure that all links in the chain are interconnected rather than partially disconnected.

6.4. Limitations and Future Research

This study takes Chinese counties as a case and explores how agricultural innovative talents contribute to the upgrading of the county-level industrial structure within the framework of “talent–knowledge–industry.” However, the upgrading of the county-level industrial structure is a complex systemic project, and future research could focus on the following directions for expansion:

First, expanding the scope of samples. This study believes that using Chinese county samples to examine the role of agricultural innovative talents is representative. However, due to the particularities of counties and agricultural development in different countries, the conclusions drawn in this study may not be applicable to the needs of county-level industrial structure upgrading in all countries worldwide. Therefore, more empirical evidence and case facts are needed as support and supplementation. Future research will continue to track this topic in order to further expand the universality and transferability of the research conclusions.

Second, strengthening the in-depth interpretation of the logic of industrial linkages. The solution proposed in this study for the structural upgrading of county-level industries mainly relies on the knowledge empowerment of agricultural innovative talents, which essentially involves “people” transforming and integrating knowledge to reshape county-level industries. In fact, future research could further focus on the knowledge spillovers between different industries and between forward and backward linkages within the same industry, which is also within the functional positioning of agricultural innovative talents. This will refine the research to the specific links in which this knowledge spillover affects the quality transformation of county-level industries. Subsequent studies will also attempt to conduct research around this important idea, starting from input–output relationships to grasp the patterns of knowledge interaction within and between industries, and to reveal the implementation path for agricultural innovative talents to assist county-level industrial structure upgrading through industrial linkages.