Mechanisms for agricultural companies in enabling smallholders’ integration into modern agriculture Authors: Yakun Chen, Hongzhen Zhang, Yuchen Lu, Zhijie Song Published: 2025 ABSTRACT Smallholder articulation of modern agriculture is an important foundation for the realization of agricultural modernization, and in the context of the market playing a decisive role in resource allocation, can agricultural companies with market advantages promote smallholder articulation of modern agricultural production? By constructing a theoretical framework of agricultural companies and small farmers’ modern agricultural production, and on the basis of the micro dataset of the Third Agricultural Census, the role and influence mechanism of agricultural companies on small farmers’ articulation of modern agricultural production are empirically examined from the viewpoint of production factor allocation. The results of the study show that agricultural companies can significantly influence the resource allocation mode of small farmers and help small farmers connect to modern agricultural production by expanding the cultivation area, improving the mechanization level, and upgrading the specialized planting structure and technology level; however, the influence on the greening production of small farmers is not yet significant. Various robustness tests verified the robustness of the above results. However, heterogeneity analysis reveals differing responses among farmers of varying scales and crop types: small-scale farmers benefit more significantly from scale expansion, while larger-scale farmers and those growing cash crops gain greater advantages from mechanization and technology adoption. These findings provide empirical evidence for agricultural companies to assist smallholder farmers in integrating with modern agriculture and formulating differentiated policies. Introduction Smallholder farmers are the fundamental units of agricultural production systems in developing countries and play a pivotal role in safeguarding food security, promoting sustainable agricultural development, and achieving agricultural modernization. According to data from the Food and Agriculture Organization of the United Nations (FAO), smallholder farmers constitute over 80% of all farming households globally1 and contribute more than 70% of the world’s food production. Despite managing only about 12% of the world’s arable land, they provide livelihoods for over 2 billion people and produce approximately 80% of the food in sub-Saharan Africa and Asia2. Thus, smallholder farmers not only play an irreplaceable role as primary producers within the global food system but also hold profound institutional significance for social stability and ecological security. For China, this group constitutes both the foundational unit of agricultural production and the core supporting force for rural social structures and governance systems3. However, with the accelerated advancement of industrialization, urbanization, and agricultural modernization, the marginalization of smallholder farmers within the agricultural production system has become increasingly pronounced4. Intertwined issues such as land fragmentation5 extensive production methods, aging farming populations6 and constraints on capital and technology7 place smallholder farmers at a relative disadvantage within modern agricultural value chains, hindering their technological upgrading and organizational integration. This structural dilemma not only undermines the overall efficiency and resilience of agricultural production systems but also significantly hampers the institutional progress of agricultural modernization in developing countries, including China. Consequently, establishing effective linkage mechanisms between smallholder farmers and modern agricultural systems has become a core challenge requiring urgent resolution in agricultural economics research and policy practice. The integration of smallholder farmers into modern agricultural production serves as a foundational component of agricultural modernization. However, this process faces multiple structural constraints in practice. Extensive research indicates that smallholder development is constrained by a dual dilemma of insufficient internal endowments and external institutional failures, manifesting as issues such as an aging labor force, limited farm size, inadequate technology adoption, and weakened market position8–10. These intertwined constraints undermine the overall efficiency and resilience of agricultural production systems, forming a core bottleneck in agricultural modernization across developing countries, including China. From an endogenous perspective, the weakening labor force and skill structure among smallholder farmers constitute a key constraint on agricultural transformation. Studies in European countries reveal that the absence of farm successors and an aging workforce significantly reduce agricultural productivity and the pace of technological advancement11,12. In European countries like Ireland, the aging problem stemming from a lack of agricultural successors has already negatively impacted agricultural productivity growth13. Similar trends are equally prominent in rural China. Accelerated industrialization and urbanization have concurrently led to rural hollowing-out, aging smallholder farmers, and low cultural literacy, placing them in a passive position regarding agricultural technology adoption and management innovation14,15. Furthermore, the generally small scale of smallholder operations makes it difficult to develop the specialized and scaled production capabilities required for modern agriculture8,16. Research indicates that small-scale operations inherently face disadvantages in capital investment, mechanization levels, and risk resilience. They are particularly marginalized in high-value agricultural markets due to their inability to meet standardization and stable supply requirements9,10,17. Finally, low mechanization adoption among smallholder farmers, coupled with reliance on traditional tools, results in low production efficiency and high costs18. In Malawi, the average technical efficiency of small-scale maize farmers was only 46.23%, with 79% of farmers operating below 70% efficiency—partly attributable to insufficient mechanization penetration19. At the external environment level, smallholder farmers have long occupied a disadvantaged position within market systems. Inadequate access to information and insufficient bargaining power make it difficult for them to secure reasonable returns during transactions20. In many African countries, smallholders are forced to sell agricultural products at low prices due to their reliance on middlemen and informal trading channels, significantly reducing market returns and reinvestment capacity21. In the Chinese context, similar “market disconnect” phenomena persist. Smallholder farmers struggle to integrate directly into modern agricultural supply chains, further limiting the depth and breadth of their participation in agricultural modernization22,23. To address these challenges, both academia and policymakers have pursued multidimensional approaches, accumulating substantial research insights particularly in organizational and scaled operations. Among these, family farms and farmer cooperatives are widely recognized as key pathways for integrating smallholder farmers into modern agriculture. Research generally indicates that family farms enhance agricultural production organization through land transfers and scaled operations24,25 while cooperatives enhance smallholders’ market competitiveness through collaborative mechanisms and collective bargaining power26,27 In policy practice, models such as “cooperatives + farmers” and “family farms + farmers” have achieved significant results in multiple regions28–31. Academics widely acknowledge that advancing smallholder organization and scaling up agricultural socialized services constitute the core mechanisms for bridging smallholders with modern agriculture32. However, organizational advancement is often constrained by factors such as fragmented land ownership, capital shortages, and imperfect governance mechanisms. This has led to fragmented research outcomes—including studies on family farms, cooperatives, and agricultural corporations—lacking a systematic explanatory framework and unified logical thread. More importantly, despite the positive outcomes achieved by family farms and cooperatives in empowering smallholder farmers, their roles remain significantly constrained33. The family farm model is primarily suited for medium-sized farmers with adequate capital and management capabilities, making it difficult to broadly cover the vast number of dispersed smallholder farmers34. Meanwhile, cooperatives face challenges such as insufficient internal governance capacity and strong member heterogeneity, limiting their effectiveness in enhancing smallholder farmers’ market capabilities35. Furthermore, these models have relatively limited functions in promoting technology diffusion, factor integration, and risk sharing, making it difficult to meet the high-input, high-standard, and high-tech requirements of agricultural modernization. Thus, while existing models have partially alleviated the structural disconnect between smallholder farmers and modern agriculture, they have not fundamentally resolved issues of factor misallocation and inefficient resource allocation. This makes the introduction of a new type of economic entity—one capable of efficiently allocating resources under market logic, promoting factor mobility, and balancing equity with efficiency—an urgent topic requiring in-depth consideration. On basis of this, agricultural companies, as the most dynamic market-oriented entities, are increasingly recognized as a key force in effectively bridging smallholder farmers with modern agriculture36. Leveraging their advantages in capital, technology, branding, and information, agricultural companies can achieve full-chain integration across land transfers, agricultural input supply, technical services, and product sales, providing smallholder farmers with systematic support for modern agriculture37,38. Simultaneously, corporate management models can effectively reduce smallholder farmers’ market transaction costs and production risks through contractual relationships, contract farming, and benefit-sharing mechanisms, thereby enhancing their capacity to integrate into modern agricultural systems39–41. However, existing research has relatively limited attention on the role of agricultural companies, focusing more on issues such as profit distribution or market inequality in the linkage between leading enterprises and farmers42,43 with insufficient exploration of their resource allocation functions and operational mechanisms in facilitating the organic integration of smallholder farmers into modern agriculture. Can agricultural companies drive the modernization of smallholder farming through optimized allocation of production factors? Do their effects vary across different types of farmers and production structures? These questions remain without systematic theoretical elaboration or empirical validation. Against the backdrop of the era where “the market plays a decisive role in resource allocation,” this paper aims to systematically address the following core questions: Can agricultural companies facilitate the organic integration of smallholder farmers with modern agricultural development? If so, what are the underlying mechanisms? The answers to these questions not only help reveal the functional positioning of agricultural companies in the process of agricultural modernization but also hold significant theoretical and practical implications for optimizing the allocation of agricultural factors, improving the new agricultural management system, and promoting the sustainable development of smallholder farmers. Compared with existing research, the potential innovations of this paper are mainly reflected in the following three aspects. First, in terms of research perspective, this paper breaks away from the conventional analytical paradigm focused on organized models such as family farms and farmer cooperatives. It returns to the fundamental logic of agricultural production, re-examining the role of agricultural companies in bridging smallholder farmers to modern agriculture through the lens of production factor allocation efficiency. Existing research predominantly examines the profit distribution or contractual governance relationships between agricultural corporations and farmers (e.g., leading enterprises’ linkage mechanisms with farmers, contract farming models), while insufficiently recognizing their functions in facilitating factor mobility, reducing transaction costs, and improving resource allocation structures. This paper positions agricultural corporations as market-oriented entities for factor integration, thereby expanding a new analytical framework: “factor allocation—subject coordination—agricultural modernization.” Second, in terms of methodology, this study builds upon prior theoretical research and case analyses by incorporating microdata from China’s Third National Agricultural Census. It systematically examines the role of agricultural corporations in facilitating the modernization of smallholder farmers at the household level, thereby addressing existing literature gaps characterized by insufficient empirical evidence and unclear identification mechanisms. By constructing a multidimensional indicator system and conducting stratified heterogeneity tests, this study endeavors to reveal the multi-pathway mechanism through which agricultural companies influence smallholder modernization at the empirical level, providing robust data support for theoretical analysis. Third, in terms of research content, this paper synthesizes existing literature on the definition of agricultural modernization (including dimensions such as scale, mechanization, specialization, and greening) to construct a multidimensional framework for measuring the modernization level of smallholder farmers. It identifies five pathways through which agricultural companies facilitate smallholders’ integration into modern agriculture: scale of operations, mechanization level, technology adoption, specialized cropping, and green agriculture. Simultaneously, this study further examines heterogeneous effects across different operational scales and crop types to reveal the differential logic by which agricultural enterprises propel smallholder modernization. The findings not only enrich the empirical foundation for research on agricultural modernization and agricultural enterprises but also provide crucial policy implications for optimizing the “enterprise + farmer” linkage mechanism and improving the market-based allocation system of agricultural factors. Theoretical framework and research hypothesis Theoretical framework for the connection of small farmers to modernized agricultural production by agricultural companies From a theoretical perspective, the essence of the interface between small farmers and agricultural companies is that small farmers are no longer closed production and management units but rather participate in the specialized division of labor and cooperation in the industrial chain of agricultural companies with the production factors they own to avoid market risks and expand income returns, such as through expanding the scale of land management to participate in the production chain of agricultural companies, improving labor efficiency through the use of agricultural machinery, and improving the level of technology transfer through technology transfer or training in the company, and to obtain more market profits by providing standardized products that meet the requirements of agricultural companies. Therefore, the role of agricultural companies in this “linkage” is mainly to help small farmers choose modern agricultural production methods by influencing their factor allocation and production decisions. We have reviewed the specific requirements of the National Agricultural Modernization Plan (2015–2020) (hereinafter referred to as “the Plan”) for the modernization of agricultural production, as well as the guidelines in the Opinions on Promoting the Organic Convergence of Small Farmers and Modern Agricultural Development (hereinafter referred to as “the Opinions”) from the General Office of the Central Committee of the Communist Party of China (CPC), and the Opinions from the 14th Five-Year Plan “Plan for Promoting Modernization of Agriculture and Rural Areas” (hereinafter referred to as the “14th Five-Year Plan”), and drawing on the classification methodology of previous studies, and drawing on the classification methods of44 and other studies, this paper breaks down the modernization of the agricultural production of small farmers into five dimensions, which are detailed in Fig. 1, and constructs a theoretical framework for the assistance of agricultural companies to small farmers in the organic connection of modernized agricultural production through the analysis of the impacts of agricultural companies on the modernization of the agricultural production of farmers in different dimensions. theoretical framework of modernized agricultural production. Fig. 1Theoretical framework of agricultural companies helping small farmers articulate modernized production. Theoretical framework of agricultural companies helping small farmers articulate modernized production. Scale operation From a practical perspective, for small farmers, the main way to realize large-scale operation is still the transfer of agricultural land. However, in recent years, the efficiency of farmland transfer among small farmers has been declining, and there is still a large gap between small farmers and large-scale operations45. This is, on the one hand, due to the liberalization of land transfer, the rigid rise in transfer fees, resulting in a decline in the transfer profits of farmers, the increasing willingness of nonfarm workers, and therefore a decline in the willingness of rural land transfer; on the other hand, the corresponding market risks brought about by the expansion of the scale of operation, particularly the risk of fluctuations in the price of agricultural products and difficulties in the sale of agricultural products, have led to the unwillingness of farmers to expand the scale of operation46 noted that the scale of operation can be realized through the transfer of land by farmers to directly expand the scale of operation but also through the form of trusteeship or joining cooperatives and other socialized services. Therefore, effectively improving the transfer of farmers’ income and reducing the market risk of small farmers’ large-scale production can help farmers improve their willingness to transfer, expand their cultivation area to achieve large-scale operation, and play a positive role in the conversion of agricultural companies to small farmers, which is an important means to solve the above problems. Mechanized production How to enhance the mechanized production of farmers to improve the level of modern agriculture of small farmers is an urgent problem that needs to be solved. From a realistic perspective, in the process of farmers’ business practices, three modes of agricultural mechanization, namely, self-purchase, leasing and outsourcing, have been gradually developed47. The choice of different modes is a rational decision made by farmers in their own production and operations on the basis of the analysis of costs and benefits48, and there is no comparison of the advantages and disadvantages. Agricultural companies can increase the utilization rate of agricultural machinery among farmers, especially smallholders. On the one hand, existing research shows that the scale of land operation is an important factor for farmers to purchase their own agricultural machinery; otherwise, the scale of land operation is small, it is difficult for agricultural machinery to work at full capacity, and it is difficult to recover its purchase and maintenance costs, which seriously affects the willingness of farmers to purchase their own agricultural machinery. Agricultural companies can drive small farmers to expand their scale of operation, thus further increasing the willingness of farmers to purchase agricultural machinery. On the other hand, with the expanding market demand for agricultural mechanization services, it is difficult for small farmers to bear the costs and risks of purchasing agricultural machinery, and specialized agricultural companies can develop agricultural machinery outsourcing services49, especially the increasingly emerging practice of “cross-area operation service of agricultural machinery”. This vividly illustrates that cross-area socialized services have significantly increased the adoption rate of agricultural machinery by many small farmers, and with the lower-cost machinery outsourcing services of agricultural companies, the external benefits of the division of labor and economies of scale brought by socialized services have increased50. Therefore, the development of agricultural companies can significantly increase the willingness of farmers to purchase or adopt outsourcing services and increase the level of agricultural mechanization of farmers in modern agricultural production. Technology upgrading Agricultural companies are not only the central link between farmers’ modernized agricultural production and the consumption of agricultural products but also supporters of research and diffusers of new agricultural technologies. In the process of linking agriculture with agriculture, agricultural companies with technological advantages continuously teach farmers various advanced or standardized production techniques, management techniques or finishing techniques and promote the scientific and technological development of agricultural production to further ensure that the products produced by farmers have a larger scale, higher efficiency and better quality. Only with the continuous input of modern technology, modern equipment and modern management concepts can we continue to help, transform and sublimate small farmers, making them modern laborers with human capital from ordinary agricultural workers. Therefore, the development of agricultural companies helps farmers adopt and upgrade their production techniques. Specialized cultivation Agricultural modernization is also inevitably accompanied by the specialization of agricultural production. Agricultural specialization is embodied in multiple dimensions, including regional specialization, farmer specialization, and the specialization of the agricultural production process, of which the foundation is the specialization of farmers51. Constrained by the status quo of agricultural operation with a large number of people and little land and fragmentation of land, diversification and “small and complete” planting patterns have become the hallmarks of smallholder agricultural operation in China52, which, although effective in overcoming the risk of price fluctuations in monocultures, is prone to fall into the equilibrium trap of nonspecialized production53, which makes it difficult to obtain the efficiency-enhancing benefits of specialized production. The core of the problem still lies in the market risk itself; farmers are in an unfavorable market position due to their own conditions, as well as the lagging nature of agricultural production, and it is difficult for small farmers to effectively counteract the market risk, which leads to the self-insurance behavior of diversified planting54. Compared with farmers, agricultural companies, as the main body of the market economy, have greater sales ability and market grasping ability. Through the establishment of a community of interest, the “company + farmer” business model can effectively improve the farmers’ ability to predict the market and grasp the market to overcome the previous dilemma of asymmetric market information and enhance the confidence and willingness of farmers to specialize in production. On the other hand, the order of agriculture and other forms of cooperation greatly reduces the degree to which farmers are most worried about the risk of market price fluctuations, thus further enhancing their willingness to specialize in planting and helping agriculture achieve specialized planting to promote modern agricultural production. Greening production Emphasized that agricultural modernization needs to highlight ecological leadership and continuously promote the green development of agriculture55. Reported that the agricultural green cycle and low-carbon production system, agricultural input reduction and efficiency technology, the green transformation of agricultural product processing and the scientific and technological innovation system to support the green development of agriculture are crucial to agricultural modernization56. In agricultural production practice, agricultural companies play the roles of product quality inspectors and raw material providers at the same time. Therefore, agricultural companies can help farmers realize green production from both sides. During the period of cooperation between agricultural companies and farmers, on the basis of the signed green agriculture agreement, farmers actively reduce the use of chemical fertilizers and pesticides to ensure the quality of the final product and reduce the risk of contract breach. Therefore, rather than publicizing and encouraging farmers to reduce the use of fertilizers and pesticides, strengthening the supervision of agricultural companies’ products is more effective. This is because it is easier and less costly to monitor and manage agricultural companies than it is for farmers who are dispersed and numerous. Against the backdrop of increasing environmental regulations in China, agricultural companies have internalized the green monitoring of farmers’ production processes to ensure the quality of their own products, thus contributing to the greening of farmers’ production. Empirical design of agricultural firms affecting farmers’ modernized agricultural production Source of data The data used in this paper come from the micro-survey data of small farmers in the the Third National Agricultural Census conducted in 2016. The agricultural census has been conducted three times since 1997, and the project aims to comprehensively understand the development and changes of the “three rural areas”, provide statistical information services for agricultural production operators and the public, and provide a basis for research and formulation of rural economic and social development planning and new rural construction policies. Under the guidance of the State Council and local governments, the census team adopted a comprehensive survey methodology in which census takers visit households and units directly to register and collect comprehensive information on rural areas, agriculture and farmers, with the census targets mainly including ordinary farm households, large-scale agricultural households, agricultural business units, administrative villages and townships and township administrative districts. Notably, because the Third National Agricultural Census designed separate questionnaires for general agricultural households and large-scale agricultural households and registered them separately at the beginning of the survey, the “small-scale agricultural households” in this paper refer to those agricultural households with arable land sizes of less than or equal to 50 acres under the official statistical caliber of the Third National Agricultural Census. The “small farm households” are those with less than or equal to 50 mu of cultivated land according to the official statistics of the Third Agricultural Census44. By eliminating the null and extreme samples, a total of 10,270 samples were collected. The sample size distribution is relatively balanced, except for South China, where the sample size distribution fits better with the population size distribution of the region, and the sample group is well represented. Variable design and descriptive statistics Explained variable: modernized production of farm households As mentioned earlier, the modernized production of farm households includes five dimensions: large-scale operation, mechanized production, technological production, specialized cultivation and green agriculture. On the basis of the micro-survey data of small farmers in the 2016 Third National Agricultural Census, this paper draws on related research45 and adopts the multi-indicator comprehensive measurement method, which is currently the most widely used method in the literature, to construct an evaluation system for the modernized agricultural production of farm households, which consists of five second-level dimensions: large-scale operation, mechanized production, technologically based production, specialized planting and green agriculture. Considering that the above five dimensions are important manifestations of farmers’ modernized production, this paper sets the weight of each dimension to 0.2, and the indicators for each dimension are selected as follows: (1) Scale operation: This paper uses the actual farming area of farmers to indicate the scale production of farmers’ land. (2) Mechanized production: Regardless of purchasing machinery by themselves or purchasing machinery socialization outsourcing services, they are used in the process of agricultural operations. Therefore, this paper evaluates the utilization of farming machinery by farmers in the farming process from an outcome-oriented perspective, using the number of times and the selection of machine plowing, sowing and harvesting in the farming process. (3) Technology enhancement level: According to the relevant encouragement regulations in the National Agricultural Modernization Plan (2015–2020), and referring to the indicators of Ren and Yang44, this paper adopts the sum of the number of times farmers use yield-enhancing and efficiency-enhancing technologies, such as greenhouses, greenhouses, straw return to the field, and irrigation technologies in the actual cultivation process to measure the level of cultivation technologies used by farmers. (4) Specialized cultivation: This paper adopts the Herfindahl index and the maximum proportion of single crops, which are widely recognized in the academic world, to describe the cultivation structure of farmers. The higher the values of the Herfindahl index and the maximum proportion of single crops are, the higher the level of specialized cultivation of farmers. (5) Green production: Green production is a core dimension of sustainable agricultural development and a key indicator for measuring the quality of agricultural modernization. As an important symbol of the “Green Revolution”, fertilizer input is highly important for agricultural production. However, in reality, there is a clear phenomenon of excessive input in China’s agricultural production. And the current amount of fertilizer applied in China has exceeded the optimal amount of fertilizer applied in an economic sense and in a technical sense57. The Ministry of Agriculture’s “Zero-Growth Fertilizer Action Plan for 2020” shows that China’s average mu-per-mu fertilizer application is 21.9 kg (kg). The amount of fertilizer applied is 21.9 kg, which is much greater than the world average of 8 kg per mu. Therefore, this study employs the unit application rates of chemical fertilizers and pesticides to measure the level of green agricultural production based on data availability and indicator comparability (In existing research, fertilizer and pesticide application intensity serves as a commonly used proxy indicator for measuring the level of green agricultural production57(FAO, 2021), reflecting to some extent the resource utilization efficiency and environmental burden of agricultural production. However, this indicator system still has limitations: First, it fails to comprehensively cover the entire process characteristics of green production, such as water resource utilization efficiency, organic fertilizer substitution rate, and carbon emission intensity; Second, variations in crop structures and soil conditions across regions may affect the comparability of these indicators. Considering the availability and consistency requirements for micro-level individual data, this study employs fertilizer and pesticide application intensity as the core measure of green production within the scope of accessible data. Although simplified, this approach effectively captures the greening level of agricultural production and aligns with the operational feasibility of empirical analysis at this stage). Considering the differences between the above indicators, to accurately reflect the modern production level of farmers in different regions, this paper standardizes the above indicators, and the processing is shown in Eq. (1).1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C'=[C - \hbox{min} C]/[\hbox{max} C - \hbox{min} C]$$\end{document} where C denotes each of the above three-level dimension indicators; C’ represents the data after 0–1 standardization; and maxC and minC denote the maximum and minimum values in the original data of the indicators, respectively. After 0–1 standardization, the indicators can be effectively weighted and compared, better reflecting the level of modernized agricultural production of farmers and the gap in modernized agricultural production between regions. Core explanatory variable: agricultural companies linked with farmers To verify whether agricultural companies can help small farmers organically connect to modern production, this paper sets the independent variable as a dummy variable for whether farmers participate in the “company + farmer " service model and assigns a value of 1 if farmers participate in the model and 0 otherwise. Control variables To control for the influence of the individual characteristics of farmers, family characteristics and regional level status, this paper chooses to control other factors affecting the modernization of farmer production. Specifically, these variables include the age of the household head, gender, education level, number of household laborers, household per capita cultivated land area, farmland rights, land cultivation conditions (using the ratio of cultivated land that can be efficiently irrigated), and regional dummy variables (controlling for the impact of interregional differences) to achieve efficient estimation as much as possible. The descriptive statistics for each variable are shown in Table 1. Table 1Definition and assignment of model variables.Variable categoryVariable nameVariable meaning and assignmentMeanStandard deviationMin.Max.Dependent variableSmallholder Modern Agricultural DevelopmentScale operationActual household cultivation area4.215.831.0749.50Mechanized productionNumber of times of mechanized plowing, sowing and harvesting in actual cultivation0.511.0103Level of technologicalSum of items for yield-enhancing technologies in farmingSpecialized cultivation Herfindahl index3.111.4609Specialized cultivationHerfindahl index0.360.090.101Maximization index0.530.160.121Green productionFertilizer input53.2724.620191Independent variableCompany + farmerDummy variable0.0410.4901Control variableAge of household headActual age45.4616.551896Sex of household head0 = male and 1 = female0.480.4901Educational attainment of household head1 = no schooling, 2 = elementary school 3 = junior high school, 4 = high school or junior high school, and 5 = junior high school and above2.790.8915Part-time1 = pure farm and 0 = part-time farm household1.90.7113Total labor forceActual labor force2.331.81019Per capita cultivated land areaContracted area/total number of people (mu/person)1.682.71040.53Farmland rights status1 = yes and 0 = no0.860.3301Land cultivation conditionEffective irrigated area/total area0.610.1901 Table 1Definition and assignment of model variables.Variable categoryVariable nameVariable meaning and assignmentMeanStandard deviationMin.Max.Dependent variableSmallholder Modern Agricultural DevelopmentScale operationActual household cultivation area4.215.831.0749.50Mechanized productionNumber of times of mechanized plowing, sowing and harvesting in actual cultivation0.511.0103Level of technologicalSum of items for yield-enhancing technologies in farmingSpecialized cultivation Herfindahl index3.111.4609Specialized cultivationHerfindahl index0.360.090.101Maximization index0.530.160.121Green productionFertilizer input53.2724.620191Independent variableCompany + farmerDummy variable0.0410.4901Control variableAge of household headActual age45.4616.551896Sex of household head0 = male and 1 = female0.480.4901Educational attainment of household head1 = no schooling, 2 = elementary school 3 = junior high school, 4 = high school or junior high school, and 5 = junior high school and above2.790.8915Part-time1 = pure farm and 0 = part-time farm household1.90.7113Total labor forceActual labor force2.331.81019Per capita cultivated land areaContracted area/total number of people (mu/person)1.682.71040.53Farmland rights status1 = yes and 0 = no0.860.3301Land cultivation conditionEffective irrigated area/total area0.610.1901 Definition and assignment of model variables. Dependent variable Smallholder Modern Agricultural Development Sum of items for yield-enhancing technologies in farming Specialized cultivation Herfindahl index Model design To verify whether agricultural companies that link agriculture with agriculture help farmers achieve modern agricultural production, this paper designs the following OLS model to estimate the impact of agricultural companies on farmers’ large-scale operation, mechanized production, specialized planting, technology adoption and green planting to explore the impact of agricultural companies on farmers’ modernized production:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Y}_{i}={\alpha\:}_{0}+{\alpha\:}_{1}{Firm}_{i}+{\alpha\:}_{j}{Control}_{i}+{\tau\:}_{i}+{\epsilon\:}_{i}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Y}_{i}\:$$\end{document} represents the level of modernized production of farmers, including the five dimensions of scale, mechanization, specialization, technology and greening; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Firm}_{i}\:$$\end{document} represents the dummy variable of whether individual farmer i participates in “company + farmer”; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Control}_{i}\:$$\end{document} represents the set of control variables; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\tau\:}_{i}\:$$\end{document} represents the regional dummy variable; and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\epsilon\:}_{i}\:$$\end{document} represents the random disturbance term. This paper focuses on the significance and sign of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\:\alpha\:}_{1}$$\end{document}; if \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\:\alpha\:}_{1}$$\end{document} passes the significance test and is positive, it means that the agricultural company will help small farmers realize modernized agricultural production. Analysis of the empirical results of agricultural firms’ influence on farmers’ modernized production Analysis of benchmark results To guarantee the reliability of the estimation results, this paper performs a variance inflation factor (VIF) test for the multicollinearity problem of the variables, and the maximum value of the VIF of each variable and the model as a whole is 6.35, which is much lower than the empirical critical value of 10; thus, showing that there is no multicollinearity problem among the variables. In addition, To eliminate the heteroscedasticity of residuals and potential autocorrelation issues, robust standard errors are employed for estimation in this paper. The OLS estimation results for agricultural companies’ modernized production among farmers are shown in Tables 2 and 3. In the baseline model (Column 1), only the agricultural firm’s farmer engagement variable is included. Results show that, except for mechanized farming and green production, the coefficients for farmer engagement are positive at the 1% significance level. This indicates that, without controlling for household head and household characteristics, agricultural firm participation significantly enhances smallholder farmers’ modern production levels—including scale of operations, specialization, and technology adoption rates. Subsequently, columns (2) to (5) sequentially introduced regional fixed effects, household head characteristics (age, gender, education level, and whether engaged in multiple occupations), household characteristics (labor force size and per capita cultivated area), and regional agricultural production conditions (land rights certification and water conservancy conditions) as control variables. After controlling for these factors, the coefficient for agricultural companies’ linkage with farmers remained positive at the 1% significance level. This indicates that, even after controlling for potential confounding factors, agricultural companies’ linkage with farmers still significantly promotes modernized production among smallholder farmers. However, it is noteworthy that agricultural companies’ efforts to connect and support farmers have not significantly impacted mechanized farming or green production practices. This finding reflects the deep-seated constraints imposed by institutional and environmental factors on the modernization process of smallholder farmers. First, from the perspective of agricultural land systems, China’s farmland is highly fragmented and dispersed, with land contracting rights held by a vast number of smallholder farmers. This makes it difficult for individual farmers to achieve economies of scale. Even when participating in the “company + farmer” model, the introduction and efficient utilization of mechanized equipment remain constrained by factors such as small plot sizes, scattered fields, and transportation difficulties, resulting in limited improvements in mechanization levels. Second, from a market environment perspective, the promotion of green production practices (such as reduced fertilizer and pesticide use) relies on robust technical support, regulatory systems, and price incentive mechanisms. In China’s current agricultural market, smallholder farmers face pressure to meet high yield demands while securing income stability. Agricultural companies still have limited capabilities in field management and technical monitoring, making it difficult to comprehensively supervise farmers’ green production practices. Consequently, even when participating in corporate production models, farmers tend to maintain high input levels to ensure yields and economic returns, thereby inhibiting improvements in green production indicators. Furthermore, the developmental status of agricultural companies influences outcomes. While companies can enhance farmers’ scale, specialization, and technology adoption through technical training, contract farming, and resource integration, their direct promotion of mechanization and green production remains constrained by capital, technical, and regulatory limitations. Particularly among small and medium-sized agricultural companies, limited financial and technical capabilities hinder support for high-cost mechanized equipment promotion and green production oversight. Overall, the OLS estimation results indicate that agricultural companies’ role in linking and empowering farmers to modernize smallholder production is both significant and heterogeneous. On one hand, companies effectively enhance farmers’ scale, specialization, and technology adoption through contract farming, technical training, and production factor integration. On the other hand, institutional constraints, land characteristics, and insufficient market incentives partially limit improvements in mechanization rates and green production levels. This suggests that policymakers should concurrently improve agricultural land systems, strengthen green incentive mechanisms, and enhance agricultural companies’ support in technical services and regulatory capabilities when promoting the “company + farmer” model. Such measures will comprehensively advance the modernization and green transformation of smallholder farming. Table 2Benchmark regression results: agricultural companies help farmers modernize agricultural production.Variable categoryVariable nameFarmers’ level of modernized agricultural production(1)(2)(3)(4)(5)Core Explanatory VariablesCompany + farmer0.147***(0.021)0.132***(0.019)0.067**(0.017)0.060***(0.016)0.061***(0.016)Characteristics of household headAge of household head——−0.017(0.043)−0.019(0.040)−0.021(0.041)Sex of household head——−0.762***(0.091)−0.862***(0.087)−0.851***(0.87)Educational attainment of household head——0.097***(0.019)0.084***(0.020)0.080***(0.021)Part-time——0.020***(0.007)0.012**(0.006)0.011*(0.006)Household Characteristics VariablesTotal labor force———0.097***(0.017)0.093***(0.018)Per capita cultivated land area———0.008***(0.002)0.007***(0.002)Variables on production statusFarmland rights status————0.019(0.014)Land cultivation condition————0.041***(0.017)Constant6.175***(0.801)5.772***(0.502)8.093***(0.757)2.516***(0.982)3.351***(0.643)Area fixed effectsUncontrolledControlledControlledControlledControlledObservations10,27010,27010,27010,27010,270 Table 2Benchmark regression results: agricultural companies help farmers modernize agricultural production.Variable categoryVariable nameFarmers’ level of modernized agricultural production(1)(2)(3)(4)(5)Core Explanatory VariablesCompany + farmer0.147***(0.021)0.132***(0.019)0.067**(0.017)0.060***(0.016)0.061***(0.016)Characteristics of household headAge of household head——−0.017(0.043)−0.019(0.040)−0.021(0.041)Sex of household head——−0.762***(0.091)−0.862***(0.087)−0.851***(0.87)Educational attainment of household head——0.097***(0.019)0.084***(0.020)0.080***(0.021)Part-time——0.020***(0.007)0.012**(0.006)0.011*(0.006)Household Characteristics VariablesTotal labor force———0.097***(0.017)0.093***(0.018)Per capita cultivated land area———0.008***(0.002)0.007***(0.002)Variables on production statusFarmland rights status————0.019(0.014)Land cultivation condition————0.041***(0.017)Constant6.175***(0.801)5.772***(0.502)8.093***(0.757)2.516***(0.982)3.351***(0.643)Area fixed effectsUncontrolledControlledControlledControlledControlledObservations10,27010,27010,27010,27010,270 Benchmark regression results: agricultural companies help farmers modernize agricultural production. 0.147*** (0.021) 0.132*** (0.019) 0.067** (0.017) 0.060*** (0.016) 0.061*** (0.016) −0.017 (0.043) −0.019 (0.040) −0.021 (0.041) −0.762*** (0.091) −0.862*** (0.087) −0.851*** (0.87) 0.097*** (0.019) 0.084*** (0.020) 0.080*** (0.021) 0.020*** (0.007) 0.012** (0.006) 0.011* (0.006) 0.097*** (0.017) 0.093*** (0.018) 0.008*** (0.002) 0.007*** (0.002) 0.019 (0.014) 0.041*** (0.017) 6.175*** (0.801) 5.772*** (0.502) 8.093*** (0.757) 2.516*** (0.982) 3.351*** (0.643) Table 3Benchmark regression results: different dimensions of modernized agricultural production.Variable name(1)(2)(3)(4)(5)Panel A: large-scale operationCompany + farmer0.193***(0.016)0.168***(0.016)0.107**(0.014)0.096***(0.014)0.073***(0.014)Panel B: mechanised productionCompany + farmer0.023(0.016)0.018(0.015)0.013(0.016)0.013(0.015)0.013(0.015)Panel C: technological advancementCompany + farmer0.127***(0.032)0.118***(0.032)0.105***(0.030)0.104***(0.030)0.105***(0.030)Panel D: specialised cultivationCompany + farmer0.423***(0.017)0.409***(0.018)0.215***(0.018)0.210***(0.017)0.199***(0.017)Panel E: green productionCompany + farmer−0.136(0.137)−0.153(0.125)−0.117(0.123)−0.113(0.123)−0.108(0.124)Household Characteristics Control VariablesUncontrolledUncontrolledUncontrolledControlledControlledProduction Conditions Control VariablesUncontrolledUncontrolledUncontrolledUncontrolledControlledArea fixed effectsUncontrolledControlledControlledControlledControlledObservations10,27010,27010,27010,27010,270 Table 3Benchmark regression results: different dimensions of modernized agricultural production.Variable name(1)(2)(3)(4)(5)Panel A: large-scale operationCompany + farmer0.193***(0.016)0.168***(0.016)0.107**(0.014)0.096***(0.014)0.073***(0.014)Panel B: mechanised productionCompany + farmer0.023(0.016)0.018(0.015)0.013(0.016)0.013(0.015)0.013(0.015)Panel C: technological advancementCompany + farmer0.127***(0.032)0.118***(0.032)0.105***(0.030)0.104***(0.030)0.105***(0.030)Panel D: specialised cultivationCompany + farmer0.423***(0.017)0.409***(0.018)0.215***(0.018)0.210***(0.017)0.199***(0.017)Panel E: green productionCompany + farmer−0.136(0.137)−0.153(0.125)−0.117(0.123)−0.113(0.123)−0.108(0.124)Household Characteristics Control VariablesUncontrolledUncontrolledUncontrolledControlledControlledProduction Conditions Control VariablesUncontrolledUncontrolledUncontrolledUncontrolledControlledArea fixed effectsUncontrolledControlledControlledControlledControlledObservations10,27010,27010,27010,27010,270 Benchmark regression results: different dimensions of modernized agricultural production. 0.193*** (0.016) 0.168*** (0.016) 0.107** (0.014) 0.096*** (0.014) 0.073*** (0.014) 0.023 (0.016) 0.018 (0.015) 0.013 (0.016) 0.013 (0.015) 0.013 (0.015) 0.127*** (0.032) 0.118*** (0.032) 0.105*** (0.030) 0.104*** (0.030) 0.105*** (0.030) 0.423*** (0.017) 0.409*** (0.018) 0.215*** (0.018) 0.210*** (0.017) 0.199*** (0.017) −0.136 (0.137) −0.153 (0.125) −0.117 (0.123) −0.113 (0.123) −0.108 (0.124) Robustness test Replacement of proxy variables To weaken the interference of other new agricultural management main bodies, this paper further excludes the samples that join multiple new agricultural organizations at the same time and retains only the samples that participate in the “company + farmer” business model and do not join any new agricultural organization (form) to verify the robustness of the above estimation results. The results of the OLS estimation are shown in Table 4. Among them, the sign and significance of the estimation results of modernized production, large-scale operation, technology adoption, specialized cultivation and green production are more consistent with those of the baseline estimation results, confirming the robustness of the baseline results. Notably, the estimation result of mechanization operation is positive at the 10% significance level, indicating that agricultural companies can promote the mechanization operation level of farmers to a certain extent, and the estimation results show that after excluding the influence of other operation modes, the impacts of agricultural companies’ linkages with farmers on farmers’ modernized production are more comprehensive. Table 4Robustness test results: replacement variables.Variable name(1)(2)(3)(4)(5)(6)Modernized productionScale operationMechanized operationTechnology adoptionSpecialized cultivationGreen productionCompany + farmer0.174***(0.043)0.213***(0.020)0.037*(0.020)0.165**(0.035)0.462***(0.016)0.061(0.054)Control variables for household characteristicsControlledControlledControlledControlledControlledControlledControl variables for production conditionsControlledControlledControlledUncontrolledControlledControlledArea fixed effectsControlledControlledControlledControlledControlledControlledObservations785378537853785378537853 Table 4Robustness test results: replacement variables.Variable name(1)(2)(3)(4)(5)(6)Modernized productionScale operationMechanized operationTechnology adoptionSpecialized cultivationGreen productionCompany + farmer0.174***(0.043)0.213***(0.020)0.037*(0.020)0.165**(0.035)0.462***(0.016)0.061(0.054)Control variables for household characteristicsControlledControlledControlledControlledControlledControlledControl variables for production conditionsControlledControlledControlledUncontrolledControlledControlledArea fixed effectsControlledControlledControlledControlledControlledControlledObservations785378537853785378537853 Robustness test results: replacement variables. 0.174*** (0.043) 0.213*** (0.020) 0.037* (0.020) 0.165** (0.035) 0.462*** (0.016) 0.061 (0.054) Causal effect estimation method Empirical analysis of the benchmark regression indicates that agricultural companies’ collaboration with farmers significantly promotes farmers’ scaled operations, technology adoption, and specialized cultivation, thereby facilitating smallholder farmers’ seamless integration into modern agricultural production. Despite incorporating household-level control variables and regional fixed effects in the baseline model, potential selection bias may persist. This includes: - Selection bias [where differences between control and treatment groups extend beyond policy intervention, introducing individual variations that distort policy evaluation outcomes] Additionally, sample proportion issues [Only 4.1% of the total sample had a dummy variable value of 1 for agricultural companies linking and guiding farmers. PSM can reduce estimation bias through precise matching.] may compromise the robustness of the estimation results. Therefore, drawing on Cheng Hong’s (2018) research approach, this study employs counterfactual inference-based propensity score matching. By identifying a counterfactual control group similar to the treatment group, a counterfactual framework is constructed to compare differences between the treatment group (agricultural companies collaborating with smallholder farmers) and the control group (smallholder farmers not collaborating with agricultural companies). This isolates the causal effect of agricultural companies on smallholder farmers’ integration into modern agricultural production. To systematically identify the impact of agricultural companies on the modernization of smallholder farming, this study examines five specific dimensions of agricultural modernization in its empirical analysis to ensure the robustness of its conclusions. First, a Probit model is employed to estimate the propensity score for farm households to participate in the “company-plus-farmers” business model, controlling for household head characteristics, household attributes, and production conditions. Second, based on the estimated propensity scores, Propensity Score Matching (PSM) is applied to examine the impact of agricultural companies’ linkage and support mechanisms on smallholder modern agricultural production. To avoid model dependency bias and enhance estimation robustness, this study compares three commonly used matching algorithms: k-nearest neighbors matching (k = 3), caliper k-nearest neighbors matching (k = 3, caliper = 0.07), and kernel matching. While no single method is inherently superior, each emphasizes different aspects of matching quality and sample size, enabling multi-faceted validation of the consistency and robustness of estimation results. To ensure matching quality and estimation validity, this study further conducted tests for common support and covariate balance, with results presented in Table 5; Fig. 2. Findings indicate that the overlap in propensity score distributions significantly expanded after matching. Most samples from the treatment and control groups fell within the common support interval, showing no systematic differences in control variables but exhibiting significant distinctions in the outcome variable. This indicates that the matched samples satisfy the balance assumption and conditional independence hypothesis. The PSM model effectively reduces differences in the distribution of explanatory variables between the two groups, mitigates self-selection bias, and thereby ensures the robustness and reliability of the estimation results. Table 5Matching quality test.Sample groupingSample matchingPseudo R2Chi-squareDeviation meanB-value (%)R R-valueCompaniy + farmerPre-matching0.0431437.5410.545.71.31Nearest neighbor matching0.00120.371.66.41.08Caliper matching0.0005.020.64.10.99Kernel matching0.00117.741.66.91.02 Table 5Matching quality test.Sample groupingSample matchingPseudo R2Chi-squareDeviation meanB-value (%)R R-valueCompaniy + farmerPre-matching0.0431437.5410.545.71.31Nearest neighbor matching0.00120.371.66.41.08Caliper matching0.0005.020.64.10.99Kernel matching0.00117.741.66.91.02 Matching quality test. Fig. 2Probability density plot. Probability density plot. The PSM estimation results under the nearest neighbor matching (k-nearest neighbors) method are presented in Table 6(1). The findings indicate that after changing the matching method, the overall impact of corporate-farmer partnerships on smallholder modern agricultural production remained largely unchanged, except for the mechanization coefficient, which became positive at the 10% significance level. Specifically, the average treatment effects for scaled-up operations, mechanized planting, technology adoption, and specialized planting were 0.103, 0.167, 0.328, and 0.033 (standard error: 0.025), respectively, all positive at the 10% significance level. This indicates that after effectively eliminating self-selection bias and potential confounding factors, agricultural companies’ collaboration with smallholders still significantly enhances their modern production levels. To further test the robustness of estimation results to matching methods and parameter choices, this study conducted repeated analyses using propensity score matching and kernel matching. The results show that different matching methods and parameter settings did not substantially affect the sign or significance of the PSM estimation results, indicating the robustness of the estimation [Other matching results showed no significant changes. To avoid repetitive descriptions, they are not detailed here. Please contact the authors for further details if needed.]. Placebo test To further validate the robustness of empirical results, this study conducted a placebo test. Specifically, a virtual treatment group was constructed: 500 households were randomly selected from the sample of farmers who did not participate in any new business organization model, and it was assumed that they participated in the “company + farmer” model. PSM regression was then re-run. If significant results still emerged under this fictitious treatment group, it would indicate that the original estimates may have been affected by potential bias. The placebo test results are presented in Table 6(2). PSM estimates under all matching methods were insignificant, indicating no systematic effect from the virtual treatment group. This finding provides negative confirmation of the robustness of prior analyses, further demonstrating that agricultural companies’ role in linking and empowering farmers to modernize smallholder production is not an accidental outcome of sample selection or model specification. Instead, it represents a causally significant relationship with practical economic implications, thereby enhancing the credibility of the research conclusions. Table 6Robustness test and heterogeneity test.VariableRobustness testHeterogeneity testPSM testAuthenticity test≤ 10 acres10–50 acresGrain cropsEconomic crops(1)(2)(3)(4)(5)(6)Scale operation0.103**(0.047)0.005(0.016)0.187***(0.051)0.092*(0.053)0.017(0.034)0.132***(0.046)Mechanized production0.167*(0.094)0.003(0.005)0.142(0.251)0.223**(0.108)0.152*(0.089)0.201***(0.085)Level of technological0.328***(0.145)0.021(0.017)0.202*(0.113)0.472***(0.173)0.069*(0.038)0.841***(0.284)Specialized cultivation (MI)0.033**(0.016)0.003(0.002)0.014(0.012)0.037**(0.016)0.007(0.021)0.042**(0.020)Specialized cultivation (HI)0.025*(0.015)0.007(0.028)0.393(0.461)0.527(0.764)−0.507(0.431)0.120(0.113)Green production0.483(0.440)0.012(0.087)0.187***(0.051)0.092*(0.053)0.017(0.034)0.132***(0.046) Table 6Robustness test and heterogeneity test.VariableRobustness testHeterogeneity testPSM testAuthenticity test≤ 10 acres10–50 acresGrain cropsEconomic crops(1)(2)(3)(4)(5)(6)Scale operation0.103**(0.047)0.005(0.016)0.187***(0.051)0.092*(0.053)0.017(0.034)0.132***(0.046)Mechanized production0.167*(0.094)0.003(0.005)0.142(0.251)0.223**(0.108)0.152*(0.089)0.201***(0.085)Level of technological0.328***(0.145)0.021(0.017)0.202*(0.113)0.472***(0.173)0.069*(0.038)0.841***(0.284)Specialized cultivation (MI)0.033**(0.016)0.003(0.002)0.014(0.012)0.037**(0.016)0.007(0.021)0.042**(0.020)Specialized cultivation (HI)0.025*(0.015)0.007(0.028)0.393(0.461)0.527(0.764)−0.507(0.431)0.120(0.113)Green production0.483(0.440)0.012(0.087)0.187***(0.051)0.092*(0.053)0.017(0.034)0.132***(0.046) Robustness test and heterogeneity test. 0.103** (0.047) 0.005 (0.016) 0.187*** (0.051) 0.092* (0.053) 0.017 (0.034) 0.132*** (0.046) 0.167* (0.094) 0.003 (0.005) 0.142 (0.251) 0.223** (0.108) 0.152* (0.089) 0.201*** (0.085) 0.328*** (0.145) 0.021 (0.017) 0.202* (0.113) 0.472*** (0.173) 0.069* (0.038) 0.841*** (0.284) 0.033** (0.016) 0.003 (0.002) 0.014 (0.012) 0.037** (0.016) 0.007 (0.021) 0.042** (0.020) 0.025* (0.015) 0.007 (0.028) 0.393 (0.461) 0.527 (0.764) −0.507 (0.431) 0.120 (0.113) 0.483 (0.440) 0.012 (0.087) 0.187*** (0.051) 0.092* (0.053) 0.017 (0.034) 0.132*** (0.046) Heterogeneity analysis Given the significant variations among farmers in terms of production characteristics, scale, and crop types, the effectiveness of agricultural companies in linking and supporting farmers may exhibit heterogeneity. Existing research indicates that farmers of different operational scales exhibit distinct behaviors within modern agricultural management systems58,59. And there are also differences in the impacts of agricultural firms’ linkages with farmers on different types of crops. On the basis of the baseline results, this paper further examines the heterogeneity of the effects of agricultural firms’ joint farming with agriculture at two levels, namely, the scale of farmers’ operations and the type of crop species. Specifically, farmers are categorized into “small farmers” (less than 10 acres) and “large farmers” (10–50 acres) according to the actual land area under cultivation and are categorized into “food crops” and “cash crops” and “grain crops” according to the type of the single largest proportion of the crop cultivated by the farmer. The results of the PSM estimation are shown in columns (3)-(6) of Table 6, and there is significant heterogeneity in the role of agricultural firms in linking with farmers.From a scale perspective, after participating in the “company + farmer” model, micro-scale farmers primarily expand their cultivated area, while small-scale farmers more often enhance production efficiency by adjusting crop structures and adopting mechanization and technological methods. Overall, larger-scale farmers demonstrate greater responsiveness to agricultural companies in terms of mechanization, technology adoption, and specialized cultivation. Their modernization gains significantly outpace those of small-scale farmers. This reflects that small-scale farmers can more readily adjust operational strategies flexibly, while large-scale farmers better demonstrate the marginal benefits of corporate involvement in factor integration and technology adoption. From the perspective of crop types, agricultural companies’ collaboration with farmers has a more pronounced effect on enhancing modern production practices among households cultivating cash crops. This collaboration not only significantly boosts scale of operations, mechanization levels, technology adoption, and specialized cultivation but also yields substantial marginal effects. In contrast, grain crop farmers participating in the “company + farmer” model primarily show improvements in mechanization and technology adoption, with insignificant changes in other dimensions and smaller marginal effects. This disparity likely stems from the higher market value and greater technology intensity of cash crops, making them more reliant on the technical services, market information, and production inputs provided by agricultural companies—thus more susceptible to corporate intervention. Conversely, grain crops, characterized by longer production cycles, lower standardization requirements, and reduced vulnerability to market fluctuations, face constraints on modernization driven by institutional environments and technological infrastructure limitations. The mechanism through which agricultural companies drive modern agricultural production among farmers Empirical results indicate that agricultural companies can effectively help farmers connect with modern agricultural production, significantly improving their mechanization level, scale operation, technology adoption and specialized cultivation. However, its intrinsic influence mechanism still needs to be further tested. On the basis of the theoretical analysis framework, this paper further analyzes and tests the specific mechanism of modernized agricultural production by agricultural companies. Specifically, (1) the path of large-scale operations, i.e., land transfer or land reclamation, should be verified; (2) the level of mechanization, i.e., the purchase of outsourced farm machinery services, the purchase of farm machinery, and the hiring of laborers, should be verified; (3) the adoption of technology, i.e., technical training for farmers, should be verified; and (4) the specialization of planting, i.e., the tendency toward non-food cultivation, should be verified. Owing to data limitations, this paper is only able to test the mechanism of large-scale operation, mechanization level, technology adoption and specialized planting and is not yet able to empirically validate the mechanism of greening production, which needs to be considered in future research. In terms of the specific estimation method, this paper uses the PSM estimation method, and the data are divided into a control group and a treatment group on the basis of whether they cooperate with farmers or not, and the PSM model is used to study the impact of the driving role of agricultural companies on the behavior of farmers to transfer to the land, purchase machinery, etc., to reveal the role of the mechanism of the agribusiness linkage with agriculture. The results of the PSM estimation are shown in Table 7. To visualize the mechanism of the role of the path in this paper, the estimation results are shown in the form of graphs (Fig. 3), in which the shaded connections are the actual path of action and the other connections indicate that the validation of the impact mechanism has not been passed.This analysis provides direct evidence for understanding the functional role of agricultural companies in linking and empowering farmers to promote modernized production among smallholder farmers, and offers reference for policy design and optimization of new agricultural business models. Table 7Mechanisms of action tests.CategoryIndicatorOutcomeScale operationLand transfer0.104**(0.047)Reclamation0.044(0.037)Mechanized productionPurchase of farm machinery services0.169**(0.076)Number of owned farm machinery0.083**(0.045)Number of hired laborers−0.179(0.203)Level of technologicalTechnical training0.458*(0.257)Specialized cultivationPercentage of grain crops−0.023**(0.010) Table 7Mechanisms of action tests.CategoryIndicatorOutcomeScale operationLand transfer0.104**(0.047)Reclamation0.044(0.037)Mechanized productionPurchase of farm machinery services0.169**(0.076)Number of owned farm machinery0.083**(0.045)Number of hired laborers−0.179(0.203)Level of technologicalTechnical training0.458*(0.257)Specialized cultivationPercentage of grain crops−0.023**(0.010) Mechanisms of action tests. 0.104** (0.047) 0.044 (0.037) 0.169** (0.076) 0.083** (0.045) −0.179 (0.203) 0.458* (0.257) −0.023** (0.010) Fig. 3Company + farmer impact mechanisms. Company + farmer impact mechanisms. Table 7 presents the empirical test results of agricultural companies’ mechanisms for linking and empowering smallholder farmers in modern agricultural production. First, regarding scale of operations, the land transfer coefficient is 0.104, significantly positive at the 5% significance level. This indicates that agricultural company partnerships effectively promote smallholder farmers’ expansion of operational scale through land transfers. This result reflects that smallholders, constrained by fragmented land holdings and scale limitations, can reduce transaction costs, access market information and capital support, and thus participate more actively in land consolidation by collaborating with agricultural companies. In contrast, the coefficient for land reclamation was 0.044 and non-significant, indicating limited impact of agricultural companies’ farmer engagement initiatives on independent land reclamation. This may relate to China’s stable land contracting rights and policy constraints. Without clear property rights guarantees and capital investment, farmers adopt a cautious approach toward high-risk land reclamation. Consequently, the cooperative model more readily stimulates expansion on transferred land rather than newly reclaimed land. Second, regarding mechanization levels, agricultural companies significantly increased farmers’ purchases of agricultural machinery services (0.169, 5% significant) and the number of owned agricultural machines (0.083, 5% significant), while showing no significant impact on the number of hired laborers. This indicates that smallholder farmers prefer outsourcing machinery services over direct machinery purchases or labor substitution to achieve mechanized production. This behavior reflects smallholders’ risk aversion toward capital investment while also demonstrating the effectiveness of the current service-oriented trend in the agricultural machinery market. By providing machinery services, agricultural companies not only reduce smallholders’ initial investment costs but also enhance the convenience and efficiency of machinery use. However, the “machinery-labor” substitution effect has yet to fully materialize. This may be because, beyond tillage, seeding, and harvesting, other production stages still rely on manual labor. Additionally, the labor combinations and production methods of some smallholders limit the possibility of machinery fully replacing labor. In terms of technology adoption, agricultural companies’ collaboration with farmers significantly increased the frequency of farmer participation in technical training (0.458, 10% significant), further promoting the application of technologies in actual production processes. This finding indicates that the cooperative model not only provides production inputs and market access but also enhances farmers’ skills and management capabilities through knowledge transfer and technical training, thereby improving smallholders’ capacity to meet the demands of agricultural modernization. During training, farmers gain exposure to novel cultivation techniques, pest and disease control methods, and management expertise, leading to heightened enthusiasm for technology adoption. This provides crucial support for boosting production efficiency and advancing modernization levels. Finally, in terms of crop specialization, the cooperative model significantly reduced the proportion of food crops (−0.023, 5% significant), indicating a certain degree of non-food crop cultivation among smallholder farmers. Agricultural companies guide smallholders toward high-value cash crops through market intelligence, technical guidance, and production-marketing linkages, which helps increase farmer income and promotes agricultural structural optimization. However, this shift away from grain production also alerts policymakers to the need for balancing food security with cash crop development, while considering land quality, irrigation conditions, and market demand to avoid local declines in grain production capacity. Overall, empirical findings demonstrate that agricultural companies significantly elevate smallholder farmers’ modern agricultural production levels through multiple pathways, including land transfers, mechanization services, technical training, and crop structure adjustments. Simultaneously, the results reveal constraints on effectiveness imposed by institutional environments, land fragmentation levels, production hardware conditions, and smallholder behavioral preferences, providing theoretical and practical references for policy formulation and agricultural company operations. Conclusion and policy recommendations Conclusion Based on the micro dataset of the Third Agricultural Census, this paper empirically examines how agricultural companies help small farmers connect with modern agricultural production, specifically looking at the impact of agricultural companies’ linkage on farmers’ scale operation, mechanization, technology adoption, specialized planting, and green agriculture. First, agricultural companies significantly influence small farmers’ resource allocation by expanding cultivation areas, improving mechanization, and upgrading planting structures and technologies. However, their impact on small farmers’ greening production is not yet significant. Robustness tests confirm these results. Second, mechanism analysis indicates that agricultural companies primarily leverage four pathways to connect with and empower farmers: First, expanding cultivated acreage through land transfers to achieve economies of scale; second, enhancing mechanization levels by providing outsourced agricultural machinery services and encouraging farmers to purchase their own equipment, with smallholder farmers tending to rely more on outsourced services to reduce investment costs and usage risks; third, improving farmers’ technology adoption capabilities through technical training and guidance; Fourth, by optimizing crop structures to promote non-grain production, thereby increasing the proportion of cash crops. Third, heterogeneity analysis further reveals significant differences in how farmers of varying scales and crop types respond to agricultural companies’ engagement and support: Small-scale farmers demonstrate stronger responses in scale expansion, while larger-scale farmers benefit more significantly in mechanization and technology adoption. Farmers growing cash crops show more pronounced improvements in modern agricultural practices, whereas the marginal effects for grain crop farmers are relatively limited. This provides differentiated guidance for policy formulation.(ii) Recommendations. Recommendations These findings provide empirical evidence that agricultural companies help smallholder farmers connect to modern agriculture, with the following policy implications for modernized agricultural production. First, stimulate smallholder farmers’ endogenous motivation for modernization. Policies should lower barriers for smallholders to participate in corporate operations by providing technical training, information services, and production-marketing platforms. This enables smallholders to benefit from external corporate resources while simultaneously stimulating their own internal development momentum, achieving an organic integration of “external corporate impetus” and “smallholder endogenous modernization.” Second, strengthen the leading role of agricultural companies in modern agricultural development. Targeted incentive policies should be formulated to support agricultural companies in playing a leading role in land transfer, machinery service provision, technical training, and specialized production. Concurrently, a green production responsibility assessment mechanism should be established to encourage companies to promote energy conservation, emission reduction, rational fertilizer and pesticide application, and environmental protection technologies in their collaborations, thereby enhancing smallholder farmers’ green production capabilities. Third, implement differentiated development strategies. Policies should fully account for the heterogeneity of farm size, crop types, and regional characteristics. For small-scale farmers, support should focus on scale expansion and land transfer assistance. For larger-scale farmers and those growing cash crops, the emphasis should be on promoting mechanization and technology adoption. For grain-growing farmers, guidance should be provided toward appropriate specialization and greening while maintaining food security. This differentiated policy mix can enhance the overall efficiency of agricultural companies in helping smallholders connect with modernization. Fourth, lower participation barriers and enhance cooperation convenience. It is recommended to reduce institutional and transaction costs for smallholder participation in corporate production by improving agricultural financial services, providing low-cost technology and agricultural machinery leasing, and optimizing cooperative agreement terms. This will broaden participation and enhance the effectiveness of agricultural companies’ farmer-linkage models.