Impact of the Pilot Policy for Migrant Workers’ Return Entrepreneurship on High-Quality Agricultural Development in the Context of Rural Revitalization

Abstract

This study examines the effect of China’s Pilot Policy for Migrant Workers Returning Home for Entrepreneurship on High-Quality Agricultural Development (HQAD) at the county level. Despite extensive research on return migration and rural development, few studies have focused on how such entrepreneurship policies affect HQAD. We analyze county-level panel data from 2014 to 2021 using a difference-in-differences (DID) approach to assess the policy’s effectiveness and its regional variations across China. We further validate our findings through robustness tests, including parallel trend tests, placebo tests, and propensity score matching combined with difference-in-differences analysis (PSM–DID). The results indicate that the pilot policy significantly enhances HQAD (β = 0.216, p < 0.001), with more pronounced effects in the Eastern region (β = 0.401, p < 0.001) than in the Central region (β = 0.312, p < 0.001), and no significant effects in the Western region. Additionally, our analysis shows that population agglomeration (β = 0.306, p < 0.001) and technological innovation (β = 2.970, p < 0.001) positively moderate the policy’s impact, whereas e-commerce development (β = −0.257, p < 0.001) has a negative moderating effect. These insights highlight that the success of return entrepreneurship policies is heavily dependent on regional characteristics and supportive development factors, offering crucial implications for refining policies in rural development strategies.

1. Introduction

Over the past four decades, China’s rapid urbanization has served as a double-edged sword for rural areas. While fueling unprecedented economic growth, it has also led to significant rural labor outflow, resulting in the “hollowing out” of rural areas and creating profound imbalances in regional development [1]. This demographic shift poses a critical challenge to China’s modernization, especially in the agricultural sector, where labor shortages and an aging workforce hinder High-Quality Agricultural Development (HQAD) and rural economic vitality [2].

Following the 20th National Congress of the Communist Party of China, rural revitalization has gained prominence in the national development agenda, with the recognition that “the most arduous tasks in building a modern socialist country remain in rural areas”. This strategic shift moves beyond merely boosting agricultural output to a holistic strategy encompassing economic, social, cultural, ecological, and governance dimensions [3]. A pivotal element of this strategy is the Pilot Policy of Migrant Workers Returning Home for Entrepreneurship, initiated in 2015 by the National Development and Reform Commission (NDRC) to combat rural brain drain and foster HQAD through entrepreneurial efforts.

While recent literature has broadly explored China’s rural development policies, including the impact of rural entrepreneurship and the impact of return migration [4,5], little focus has been placed on how the Pilot Policy influences HQAD at the county level. This oversight is significant, as return migrants often contribute not only capital but also advanced knowledge, innovative ideas, and extensive social networks from their urban experiences.

The concept of HQAD, underscored in recent policy documents, marks a shift from quantity-driven to quality-focused agricultural production. This innovative approach prioritizes growth driven by innovation rather than reliance on resources, and integrates primary, secondary, and tertiary industries in rural settings. However, the mechanisms by which the Pilot Policy aids this transformation at the county level are not well understood.

This paper seeks to address this research gap by empirically evaluating the impact of the Pilot Policy of Migrant Workers Returning Home for Entrepreneurship on HQAD at the county level in China. We utilize county-level panel data from 2014 to 2021, applying a difference-in-differences (DID) approach to assess: (1) the effects of the pilot policy on county-level HQAD; (2) regional variations in policy impacts, especially comparing eastern regions to others; and (3) the moderating effects of population agglomeration, technological innovation, and e-commerce on the policy–HQAD relationship.

Our study contributes significantly to the existing literature in several ways. Firstly, it offers empirical evidence of the effectiveness of the Pilot Policy of Migrant Workers Returning Home for Entrepreneurship in promoting HQAD, thus enriching discussions on policy interventions in rural development. Secondly, through an examination of regional heterogeneity and moderating factors, we provide insights into the conditions that enhance the efficacy of such policies. Thirdly, our county-level analysis delivers detailed evidence for policy optimization and the promotion of experiences across pilot regions, while providing practical insights for supporting less-developed areas and leveraging human capital in China’s rural revitalization strategy.

2. Literature Review

2.1. Research on Migrant Workers Returning Home for Entrepreneurship

Recently, the return of migrant workers to their hometowns for entrepreneurship has become a key focus in academic research. Studies have shown that the influences on this entrepreneurship are complex. At the macro-level, the economic environment, legal framework, and policy support play essential roles [6]. The regional economic environment affects entrepreneurial outcomes through institutional, human, financial, and technical factors, leading to varied entrepreneurial activities among returnees across different regions of China [7,8]. A robust legal system and market-oriented policies have been shown to facilitate entrepreneurship in multiple countries [9,10]. In China, rural-favorable policies, particularly those related to finance, industry, and land, have encouraged this trend [11]. At the micro-level, the endowment of resources is critical. Economic, human, and social capital are key dimensions [12]. Economic capital aids in identifying entrepreneurial opportunities, while human capital, built through education and work experience, enhances decision-making and information processing [13]. Social capital, derived mainly from social networks, affects the discovery and utilization of opportunities [14,15].

Historically, research in this field predominantly applied the push–pull theory [16]. By utilizing this theory, researchers identified that factors such as the desire to reunite with family and local policy incentives, coupled with the identification of business opportunities in rural areas, motivated migrant workers to return [17,18]. According to human capital theory, the skills and knowledge acquired during urban employment are vital for entrepreneurship in their hometowns [19,20]. As research evolved, scholars have explored the impact of this phenomenon on rural economic development [1], with empirical studies showing that return entrepreneurship enhances rural industrial revitalization, job creation, and the transfer of advanced management and production techniques to rural areas [3,21,22]. Current research increasingly focuses on the detailed mechanisms and obstacles involved [6,23].

2.2. Research on HQAD

HQAD reflects a specific stage of economic development in China, emphasizing not only growth in speed and scale but also in quality and efficiency [24]. HQAD applies this concept to agriculture. While the term HQAD is primarily advocated by Chinese scholars, similar concepts globally, like sustainable agricultural development and smart agriculture, align with its essence [24]. Other developing countries in Asia, such as Indonesia, the Philippines, and Vietnam, have also increased investment in agricultural support measures, promoting industrialization and modernization of agriculture [25]. Martinez and Mlachila defined HQAD as robust, stable, and sustainable growth, highlighting its importance in improving livelihoods and reducing poverty [26]. HQAD is conceptualized as a multidimensional paradigm that prioritizes efficiency, innovation-driven transformation, ecological sustainability, structural optimization, and equitable welfare distribution in agricultural systems [27]. Some scholars have suggested dividing HQAD into four components: new drivers, efficient growth, integrated industrial systems, and sustainable development [28]. Additionally, some definitions of HQAD are based on its distinctive characteristics [29]. Currently, within the academic sphere, there is a lack of a unified and authoritative evaluation system for measuring HQAD indicators. In the Chinese context, research is developing multidimensional evaluation frameworks that incorporate efficiency metrics, sustainability indicators, and innovation parameters. Quantitative studies, particularly those using entropy-weighted methods, assess variations in agricultural development across regions, though studies on HQAD measurement are primarily focused at the provincial level [27,29,30]. Even in county-level analyses, the scope is often limited to counties within a single province [31].

2.3. Research Review

The existing literature has significantly advanced our understanding of both Migrant Workers Returning to Hometown for Entrepreneurship and HQAD independently. Research on return entrepreneurship has progressed from theoretical frameworks to practical applications, offering detailed insights into the entrepreneurial decisions of returned migrant workers [6,7]. These studies have analyzed how resource endowments, including economic, human, and social capital, impact entrepreneurial opportunities and outcomes [12,13]. Similarly, studies on HQAD have thoroughly explored conceptual frameworks and evaluation methodologies, particularly focusing on the multidimensional nature of agricultural development that prioritizes quality, efficiency, and sustainability over mere quantitative growth [24,27].

However, three critical research gaps persist. First, few studies have explored the connection between migrant workers returning for entrepreneurship and HQAD at the county level, where policy implementation directly interfaces agricultural development outcomes. Although research indicates that return entrepreneurship boosts rural industrial revitalization and job creation [3,22], its specific effects on agricultural quality transformation at this granular level are under-researched. Second, the mechanisms by which returned migrant entrepreneurs impact HQAD remain poorly understood. Despite evidence of these entrepreneurs bringing advanced management techniques to rural areas [21], the processes by which their accumulated human capital and urban–rural networks improve agricultural quality require further exploration [6,23]. Third, the discussion on regional heterogeneity in these dynamics is insufficient. Given that regional economic environments influence entrepreneurial activities variably [8], understanding how return entrepreneurship affects HQAD across different regions is essential for crafting targeted policies.

3. Policy Context and Theoretical Analysis

3.1. Policy Context

In June 2015, the General Office of the State Council of China launched a transformative policy framework with the Opinions on Supporting Migrant Workers and Others Returning to Hometown for Entrepreneurship (hereafter the Opinions). This pivotal policy aimed to foster entrepreneurial enthusiasm among returnees and establish a conducive environment that maximizes their initiative and creativity. The Opinions specified five key measures: easing entry barriers for returning entrepreneurs, offering targeted tax relief and universal fee reductions, enhancing fiscal subsidies, bolstering financial support for entrepreneurship, and improving policies for industrial parks catering to returning migrant workers. This comprehensive policy framework marked a significant shift in China’s approach to rural development, highlighting the government’s dedication to balanced urban–rural growth and addressing circular migration in developing economies.

The rollout of this policy framework was executed through a meticulously structured pilot program, initiated in February 2016 by the NDRC in collaboration with ten ministries including the Ministry of Industry and Information Technology. The program unfolded in three phases: the first in February 2016, covering 90 pilot regions, including Wei County in Hebei Province, the second in December 2016 adding 116 counties (cities, districts), and the third in October 2017 incorporating an additional 135 pilot areas, totaling 341 counties (cities, districts) by the end of 2017. Each phase was formalized through official notifications, such as the Notice on Approving 135 Counties (Cities, Districts) Including Daming County in Hebei Province to Conduct Pilot Programs Supporting Migrant Workers Returning to Hometown for Entrepreneurship in Conjunction with New-Type Urbanization. This phased approach allowed for continuous policy refinement and adjustment, integrating with contemporary urbanization theories and marking a substantial advancement in China’s rural revitalization strategy.

3.2. Theoretical Analysis

The pilot policy Migrant Workers Returning to Hometown for Entrepreneurship represents an innovative strategy for regional development, aiming to stimulate the return of talent to enhance the agricultural industry’s upgrading and transformation. In this context, returnee entrepreneurs apply their advanced knowledge and innovative capabilities to modernize traditional agriculture, thereby achieving HQAD.

Initially, these entrepreneurs introduce modern management knowledge and skills, infusing new ideas and methods into agriculture, which enhances production efficiency and product quality [32]. They typically exhibit strong market awareness, which aids in penetrating new markets for local agricultural products, thus increasing brand value and competitiveness [6]. Simultaneously, government support measures, such as tax incentives and loan facilities, provided under the pilot policy, reduce barriers to entrepreneurship, increase success rates, and offer robust policy support for HQAD [33]. Based on these observations, we propose the following hypothesis:

Hypothesis 1.

The implementation of the pilot policy can promote HQAD at the county level.

Despite China’s vast and diverse landscape, regional differences significantly affect the effectiveness of such policies [32,34].The eastern region, with its advantageous geographical location, abundant resources, high financial openness, balanced industrial structure, and rich talent reserves, consistently leads in national economic development. Entrepreneurs in the east can leverage these resources and develop market conditions for rapid development. The central region, supported by national policies and market opportunities, closely follows the east in economic strength, providing fertile ground for returnee entrepreneurship. In contrast, the western region, challenged by harsh natural conditions, limited resources, less developed financial markets, and talent drain, requires additional policy support and infrastructure development to boost entrepreneurial activity and potential. Reflecting on these regional disparities, we propose another hypothesis:

Hypothesis 2.

The effectiveness of the pilot policy exhibits regional heterogeneity.

Population density can significantly influence the effectiveness of policy implementation at the regional level. High population concentrations create a conducive environment for policy success by providing a larger market, improved infrastructure, and diverse human capital [35,36]. In areas with high population density, the potential for interaction and knowledge spillover between returnees and locals may amplify the impact of the policy on HQAD [37,38]. Additionally, densely populated areas often have better support services and resource allocation, enhancing the success rates of returnee entrepreneurs [39]. Based on this, we propose the following hypothesis:

Hypothesis 3.

Population density positively moderates the impact of the pilot policy on HQAD at the county level.

The potential for technological innovation within a region can determine how effectively new knowledge and skills are adopted and utilized. Counties with robust innovation capabilities, equipped with superior R&D facilities and a skilled workforce, are better positioned to capitalize on the expertise of returning entrepreneurs [40]. These areas can more effectively implement and spread the advanced technologies and management practices introduced by returnees. Conversely, regions with lower innovation potential may struggle to translate these experiences into practical outcomes [41]. Based on this, we propose the following hypothesis:

Hypothesis 4.

Technological innovation potential positively moderates the impact of the pilot policy on HQAD at the county level.

E-commerce’s penetration into agriculture has shown mixed effects on rural revitalization and development. While some studies highlight its role in facilitating economic development [42], others suggest a detrimental impact on agricultural growth [43,44]. Although e-commerce development has supported economic progress, its negative influence on traditional agriculture is becoming apparent. Some scholars argue that e-commerce could reduce transaction costs for producer e-tailers [45]. Nonetheless, contrasting viewpoints exist. As proposed by Chintagunta et al. [46], online marketing still requires the involvement of intermediaries, suggesting that the anticipated cost savings may not be realized. Current research shows that the vertical expansion of small farmers into e-commerce by becoming agricultural e-tailers does not reduce their subordination to commercial capital. Rather, small-farmer e-tailers are dominated by quasi-monopolistic agricultural e-commerce platforms in China, which exploit them for transaction fees, charge them for internet traffic, and prevent them from competing as platforms evolve into e-tailers themselves [43]. Moreover, the high profit potential of e-commerce encourages farmers to invest in operating e-commerce platforms instead of traditional agricultural production, leading to a significant increase in farmland abandonment, with implicit and explicit rates rising by 10.3% and 28.5%, respectively [44]. Based on this, we propose the following hypothesis:

Hypothesis 5.

The development of e-commerce negatively moderates the impact of the Migrant Workers Returning to Hometown for Entrepreneurship pilot policy on HQAD at the county level.

4. Data and Methods

4.1. Data Source

This study employs an unbalanced panel dataset spanning from 2014 to 2021. Given the special status and policy biases of municipalities directly under the central government and municipal districts, and the relatively small proportion of agricultural output value in these municipalities, combining their data with that of other regions for regression could skew the estimation of model parameters. To ensure the accuracy of our regression analysis, the data excludes the four directly administered municipalities of Beijing, Shanghai, Tianjin, and Chongqing, as well as municipal districts, due to their unique status and policy advantages that could introduce bias into the analysis [47]. Instead, the sample retains data from counties, county-level cities, autonomous counties, banners, and autonomous banners, covering 1850 county-level regions (see Figure 1).

Since 2015, the NDRC of China, in collaboration with various departments, has designated 341 counties (cities, districts) as pilot areas for the Migrant Workers Returning Home for Entrepreneurship policy in three successive phases. The list of pilot regions and corresponding years was obtained from the NDRC’s official website. Data on Taobao Villages were sourced from the AliResearch Institute, highlighting that their significant development began in 2015. As of 2021, these villages are mainly located in provinces including Zhejiang, Guangdong, Shandong, Jiangsu, Hebei, Fujian, Henan, Hubei, Jiangxi, Anhui, Shaanxi, and Sichuan. Information on patent authorizations is drawn from the Patent Publication and Announcement of China, issued by the National Intellectual Property Administration. Agricultural carbon emission data are sourced from the Emissions Database for Global Atmospheric Research (EDGAR) v2024, with a spatial resolution of 0.1 degrees [48]. Additional variables are taken from the China County Economic Statistical Yearbook and individual county statistical yearbooks.

4.2. Variable Definition

Independent variable: The Pilot Policy of Migrant Workers Returning Home for Entrepreneurship is denoted by a DID dummy variable. For selected counties, this variable is set to 1 in the year they are chosen and in subsequent years; otherwise, it remains at 0.

Dependent variable: After a thorough review of the literature and considering data availability at the county level, this study develops an index system to assess HQAD(as shown in

Table 1. Construction of HQAD Indicators.

HQAD Indicators	Formula	Attribute
Agricultural Mechanization Level	Agricultural Machinery Power	+
Harvest Mechanization Degree	Harvested Area/Total Sown Area of Crops	+
Grain Yield per Unit Area	Grain Output/Arable Land Area	+
Agricultural Proportion in Primary Industry	Agricultural Output/Total Output of Agriculture, Forestry, Animal Husbandry, and Fishery	+
Agricultural Economic Contribution	Agricultural Value Added	+
Urban-Rural Income Gap	Per Capita Disposable Income of Urban Residents/Per Capita Disposable Income of Rural Residents	−
Agricultural Carbon Emission Intensity	Agricultural Carbon Emissions/Agricultural Output	−

Note: “+”: Positive indicator, higher values contribute more to HQAD; “−”: Negative indicator, higher values contribute less to HQAD.

). We use an equal-weight composite index that includes seven key indicators: (1) agricultural mechanization level, which serves as a vital foundation for transforming agricultural development methods, enhancing rural productivity, and supporting the implementation of rural revitalization strategies; (2) harvest mechanization degree, reflecting the efficiency of technological application rather than scale; (3) grain yield per unit area, which is crucial for food security in the context of China’s specific policy environment and is a primary consideration in high-quality agricultural development; (4) agricultural proportion in primary industry, mainly referring to crop farming; (5) agricultural economic contribution, which focuses on increasing agricultural output value and transforming rural industries from quantity-driven to quality-driven growth; (6) urban–rural income gap, with the goal of continuously increasing farmers’ income levels to narrow the urban–rural income disparity; and (7) agricultural carbon emission intensity, serving as an indicator of environmental friendliness [27,30,47,49,50,51]. Since these indicators vary in units, they are standardized using an equal-weight method to create a composite index for measuring HQAD.

For positive indicators, the following formula is employed for standardization:

$$X_{ij}^{nor}={\displaystyle \frac{X_{ij}-{\min }_j}{{\max }_j-{\min }_j}},$$

(1)

For negative indicators, the following formula is used for standardization:

$$X_{ij}^{nor}={\displaystyle \frac{\max (X_j)-X_{ij}}{\max (X_j)-\min (X_j)}},$$

(2)

$X_{ij}^{nor}$ represent the standardized indicator, where i denotes the county, j denotes the measurement indicator, and ${\min }_j$ and ${\max }_j$ represent the minimum and maximum values of indicator j, respectively. To further define the weights of each indicator, this study adopts the equal-weight averaging method. This method is widely used due to its simplicity, transparency, and ability to avoid subjective bias in determining the relative importance of individual indicators. Given that each component indicator significantly contributes to the improvement of HQAD at the county level, the equal-weight averaging method provides a balanced and objective approach to weight calculation.

$$Q_i={\displaystyle \frac{1}{7}}{\displaystyle {\sum }_{j=1}^7{x}_{ij}^{nor}}\times 100,$$

(3)

Mechanism Variables: These include population agglomeration, technological innovation capacity, and e-commerce development level. Population agglomeration is expressed as the logarithm of population density. Technological innovation capacity is assessed by the number of patents granted per 10,000 people. E-commerce development level is evaluated based on the logarithm of the number of Taobao Villages.

Control Variables: This study examines several factors that influence e-commerce development and incorporates control variables to manage external impacts on county-level growth. These variables cover changes in capital investment intensity, enterprise development scale, telecommunication infrastructure, financial support intensity, household savings rate, consumer market size, foreign investment attractiveness, and social welfare level.

The descriptive statistics for all variables are presented in

Table 2. Definition of main variables and descriptive statistics results.

Variables	Variables	Definitions	Mean	Standard Deviation
Dependent Variable	HQAD	Equal Weight Average Method	6.35	2.80
Independent Variable	Returning Home for Entrepreneurship Policy	Pilot counties are assigned a value of 1 from the year of selection onward; others are assigned 0	0.08	0.27
Control Variables	Capital Investment Intensity	Total social fixed asset investment/Regional GDP	1.32	4.78
	Enterprise Development Scale	ln (Number of Large and Medium-sized Industrial Enterprises)	3.73	1.47
	Telecommunication Infrastructure	ln (Fixed-line Telephone Subscribers)	9.96	1.21
	Financial Support Intensity	Year-end Financial Institution Loans/Regional GDP	0.76	0.44
	Household Savings Rate	Household Savings Deposits/Regional GDP	0.85	0.43
	Consumer Market Size	ln (Retail Sales of Consumer Goods)	12.67	1.38
	Foreign Investment Attractiveness	ln (Actual Foreign Investment Amount)	7.50	1.15
	Social Welfare Level	ln (Number of Social Welfare Institution Beds)	6.67	1.25
Mechanism Variable	Population Agglomeration	ln (Year-end Total Population/Administrative Area Land Area)	2.56	1.58
	Technological Innovation	Number of patent inventions published/10000	0.01	0.03
	E-commerce Development	ln (Number of Taobao + 1)	0.15	0.57

. The dependent variable HQAD, constructed using an equal-weight average method, shows a mean value of 6.35 with a standard deviation of 2.80, indicating substantial variation in agricultural development quality across counties. For the independent variable, approximately 10% of the sampled counties were designated as pilot regions for the Returning Home for Entrepreneurship Policy.

Among the control variables, the capital investment intensity, measured as the ratio of total social fixed asset investment to GDP, averages 1.32. Enterprise development scale exhibits considerable variation (mean = 3.73, SD = 1.47), while telecommunication infrastructure, measured by the logarithm of fixed-line telephone subscribers, averages 9.96. The financial support intensity, represented by the ratio of year-end financial institution loans to GDP, shows a mean of 0.76. Household savings rate, indicated by the ratio of household savings deposits to GDP, averages 0.85, and consumer market size, measured by the logarithm of retail sales of consumer goods, centers around 12.67. Foreign investment attractiveness, measured by the logarithm of actual foreign investment amount, averages 7.50. Social welfare level, measured as the logarithm of the number of social welfare institution beds, averages 6.67. Regarding the mechanism variables, population agglomeration (mean = 2.56, SD = 1.58) shows significant regional differences, while both technological innovation (mean = 0.01, SD = 0.03) and e-commerce development (mean = 0.15, SD = 0.57) exhibit relatively low means but notable variation across counties, suggesting uneven development in these aspects.

4.3. Model Setup

The establishment of the pilot policy for migrant workers returning to their hometowns creates an almost natural experimental environment for the study. As an external driving force, this policy not only encourages labor migration back to rural areas for entrepreneurship but also fosters favorable conditions for agricultural transformation and upgrading, forming a real-world experimental context. This facilitates the use of the DID method to explore its effects on HQAD. Given the temporal differences in the policy’s implementation and its gradual rollout across regions, the study emphasizes the timeliness and regional scope of the analysis, effectively applying a multi-timepoint DID model to accurately quantify the impact mechanisms of the pilot policy on e-commerce development. The specific model is as follows:

$$Y_{i,t}={\beta }_0+{\beta }_{1}DI{D}_{i,t}+{\displaystyle \sum {\beta }_n}Contro{l}_{i,t}+{\mu }_i+{\delta }_t+{\epsilon }_{i,t},$$

(4)

$Y_{i,t}$ represents the per capita income level of farmers in county i during year t. $DI{D}_{i,t}$ indicates whether the return entrepreneurship policy is implemented in county i during year t. Specifically, $DI{D}_{i,t}=trea{t}_i\times pos{t}_t$, where $trea{t}_i$ represents whether county i is selected as a pilot area for the return entrepreneurship initiative (if selected, $trea{t}_i=1$; otherwise, $trea{t}_i=0$), and $pos{t}_t$ is a time dummy variable, which equals 1 for years following the selection of county i as a pilot area, and 0 for the years prior to its selection. $Contro{l}_{i,t}$ represents a series of control variables. ${\mu }_i$ denotes individual fixed effects, ${\delta }_t$ represents time fixed effects, and ${\epsilon }_{i,t}$ is the random disturbance term. The study focuses on the coefficient ${\beta }_1$ of the core explanatory variable $DI{D}_{i,t}$, which reflects the net effect of the return entrepreneurship pilot policy on the county-level e-commerce development.

All statistical analyses were conducted between July 2024 and January 2025, using Stata version 18.0, with statistical significance set as a two-sided p-value < 0.05.

5. Results

5.1. Baseline Regression Results

The baseline regression results in

Table 3. Results of regression analysis.

Variables	HQAD	HQAD
	(1)	(2)
Returning Home for Entrepreneurship Policy	0.229 *** (7.806)	0.216 *** (7.419)
Capital Investment Intensity		0.001 (0.084)
Enterprise Development Scale		−0.018 (−1.164)
Telecommunication Infrastructure		0.034 *** (3.430)
Financial Support Intensity		0.102 *** (4.809)
Household Savings Rate		−0.220 *** (−8.033)
Consumer Market Size		0.081 *** (4.512)
Foreign Investment Attractiveness		0.049 *** (5.283)
Social Welfare Level		0.012 (1.498)
_cons	6.328 *** (1441.154)	4.704 *** (17.612)
R2	0.976	0.976
AREA	YES	YES
YEAR	YES	YES
N	13,594	13,573

Note: *** p < 0.001. Robust standard errors in parentheses.

show the impact of the Returning Home for Entrepreneurship Policy on HQAD. Model (1), which only includes year and region fixed effects, reveals a positively and highly significant coefficient (β = 0.229, p < 0.001) for the policy variable. This relationship continues to be robust in Model (2) (β = 0.216, p < 0.001) after incorporating a comprehensive set of control variables, validating Hypothesis 1. The explanatory power of both models is notably high, with R-squared values of 0.976 and 0.976 respectively, demonstrating that the models explain approximately 98% of the variation in HQAD.

Among the control variables in Model (2), telecommunication infrastructure (β = 0.034, p < 0.001), financial support intensity (β = 0.102, p < 0.001), consumer market size (β = 0.081, p < 0.001) and foreign investment attractiveness (β = 0.049, p < 0.001) are positively correlated with HQAD. This suggests that advancements in telecommunication infrastructure, increased financial support, a larger consumer market, and greater foreign investment attractiveness all contribute positively to HQAD. Conversely, household savings rates (β = −0.220, p < 0.001) exhibit significant negative correlations with HQAD, suggesting complex trade-offs in regional development patterns.

5.2. Mechanism Test

Table 4. Mechanism test results.

Variables	HQAD
	(1)	(2)	(3)	(4)	(5)	(6)
Returning Home for Entrepreneurship Policy	0.160 *** (5.351)	0.149 ** (5.027)	0.240 *** (8.169)	0.221 *** (7.593)	0.215 *** (7.165)	0.201 *** (6.752)
Population Agglomeration	0.029 (0.686)	0.037 (0.889)
DID × Population Agglomeration	0.312 *** (11.269)	0.306 *** (11.133)
Technological Innovation			0.968 *** (3.362)	0.905 *** (3.169)
DID × Technological Innovation			3.119 *** (3.633)	2.970 *** (3.490)
E-commerce Development					0.080 *** (4.725)	0.079 *** (4.634)
DID×E-commerce Development					−0.246 *** (−3.462)	−0.257 *** (−3.629)
_cons	6.249 *** (57.964)	4.615 *** (15.950)	6.461 *** (1172.308)	4.720 *** (16.565)	6.316 *** (1219.092)	4.699 *** (17.623)
Control	NO	YES	NO	YES	NO	YES
R2	0.976	0.976	0.977	0.977	0.976	0.976
AREA	YES	YES	YES	YES	YES	YES
YEAR	YES	YES	YES	YES	YES	YES
N	13578	13557	11507	11491	13594	13573

Note: ** p < 0.05; *** p < 0.001. Robust standard errors in parentheses.

displays the results of mechanism tests exploring the moderating effects of population agglomeration, technological innovation, and e-commerce development on the relationship between the Returning Home for Entrepreneurship Policy and HQAD. For each mechanism, analyses are conducted both with and without control variables to ensure robustness.

The first mechanism, population agglomeration, shows a significant positive interaction effect with the policy (β = 0.312, p < 0.001 without controls; β = 0.306, p < 0.001 with controls), supporting Hypothesis 3. This suggests that regions with higher population densities enhance the positive impact of the policy on HQAD, likely due to better knowledge spillovers and resource allocation efficiency.

Technological innovation demonstrates the strongest moderating effect among the three mechanisms. The interaction term exhibits large positive coefficients (β = 3.119, p < 0.001 without controls; β = 2.970, p < 0.001 with controls), strongly supporting Hypothesis 4. This indicates that counties with greater innovative capacity are better positioned to leverage returnee entrepreneurs’ knowledge and experience for agricultural development.

Conversely, e-commerce development demonstrates a significant negative moderating effect (β = −0.246, p < 0.001 without controls; β = −0.257, p < 0.001 with controls), confirming Hypothesis 5. This negative interaction suggests that highly developed e-commerce regions may face market saturation and competitive challenges that could limit the policy’s effectiveness in promoting HQAD.

The consistency of results across models with and without control variables affirms the robustness of these mechanism effects. All models maintain high explanatory power with R-squared values above 0.98, indicating strong model reliability.

5.3. Robustness Tests

5.3.1. Parallel Trend Test

To verify the validity of the DID method, we checked that agricultural development trends in pilot and non-pilot regions aligned before policy implementation, in accordance with the parallel trends assumption. Tests conducted prior to the intervention confirm no significant differences in HQAD trajectories between the two regions, supporting the assumption’s validity. Figure 2 shows that the agricultural development trends were similar in the two years leading up to the policy launch, bolstering the credibility of the DID analysis. These findings also indicate that the pilot policy substantially enhanced HQAD, confirming the effectiveness of the return entrepreneurship initiative and the reliability of the findings.

5.3.2. Placebo Test

To further assess the impact of the return entrepreneurship pilot policy and rule out potential confounding factors, we conducted a randomized placebo test. This involved simulating a scenario where policy assignment was random, to reduce the influence of unobserved variables. A fictional “return entrepreneurship pilot” classification was created within the entire sample to establish a control group that had not actually experienced the policy intervention. Figure 3 displays the outcomes of 1000 random samplings, where the regression coefficients for HQAD in the fictitious pilot counties nearly approach zero, markedly different from the effects under the actual policy intervention. This test confirms that external random factors minimally affect HQAD, reinforcing the positive impact of the return entrepreneurship pilot policy on enhancing county-level agricultural quality. The observed positive effect is directly attributable to the policy itself, and not to other unobserved variables. Consequently, the baseline regression withstands the placebo test, further validating the stability of our research findings.

5.3.3. Other Robustness Tests

During the analysis, we considered the impact of the COVID-19 pandemic between 2021 and 2022. To ensure accuracy and reliability, data from these years were excluded from the regression analysis. Results in

Table 5. Robustness tests.

Variables	Reduced Sample Time	Weight Adjustment	PSM−DID
	(1)	(2)	(3)
Returning Home for Entrepreneurship Policy	0.205 *** (7.088)	0.002 *** (4.434)	0.202 *** (7.014)
Capital Investment Intensity	0.004 (1.145)	−0.001 (−0.350)	0.001 (0.605)
Enterprise Development Scale	0.049 *** (2.611)	−0.001 *** (−3.241)	−0.011 (−0.548)
Telecommunication Infrastructure	0.053 *** (4.671)	0.001 *** (3.286)	0.023 * (1.861)
Financial Support Intensity	0.113 *** (4.328)	0.001 ** (2.433)	0.137 *** (4.409)
Household Savings Rate	−0.255 *** (−7.580)	−0.002 *** (−3.426)	−0.268 *** (−7.947)
Consumer Market Size	0.103 *** (3.907)	0.003 *** (8.721)	0.104 *** (4.958)
Foreign Investment Attractiveness	0.056 *** (5.581)	0.001 ** (2.182)	0.050 *** (4.950)
Social Welfare Level	0.012 (1.326)	0.001 (1.017)	0.011 (1.117)
_cons	3.954 *** (10.796)	0.030 *** (6.249)	4.702 *** (14.485)
R2	0.979	0.973	0.978
AREA	Yes	Yes	Yes
YEAR	Yes	Yes	Yes
N	10234	7860	9792

Note: * p < 0.1; ** p < 0.05; *** p < 0.001. Robust standard errors in parentheses.

(1) demonstrate that, after adjusting for this period, the positive effect of the return entrepreneurship policy on HQAD remains significant (β = 0.205, p < 0.001), highlighting the robustness of our findings.

In constructing the core dependent variable, HQAD level, the equal-weight method was used to standardize the indicators and construct the composite index for e-commerce development level, Y. To test the robustness of the results, Table 5 (2) shows that, after adjusting the weights, the policy effect remains stable (β = 0.002, p < 0.001).

Although the initial analysis used a multi-timepoint DID model, and both parallel trend and placebo tests were completed, it is acknowledged that the selection of return entrepreneurship pilot areas might not be entirely random but based on specific criteria. This introduces a risk of sample selection bias, which could challenge the accuracy of the baseline regression conclusions. Therefore, this study further employs propensity score matching combined with difference-in-differences (PSM-DID) to enhance the robustness of the results. Data in Table 5 (3) show that, after implementing propensity score matching, the DID estimate remains positive and significant (β = 0.202, p < 0.001), confirming the policy’s positive effect and effectively addressing the selection bias issue, ensuring the reliability and stability of the research conclusions.

5.4. Heterogeneity Analysis

Given the regional diversity in China’s development, the impact of the return entrepreneurship pilot policy on agriculture may vary across regions. Following the National Bureau of Statistics’ classification standards and based on local natural endowments and economic conditions, the country is divided into three regions: Eastern, Central, and Western. This segmentation is designed to explore regional variations in the policy’s growth-promoting effects. Results are presented in

Table 6. Heterogeneity Analysis.

Variables	(1)	(2)	(3)
	East	Middle	West
Returning Home for Entrepreneurship Policy	0.401 *** (5.448)	0.312 *** (6.238)	0.050 (1.270)
Capital Investment Intensity	−0.006 (−1.205)	−0.050 *** (−2.635)	0.001 (0.589)
Enterprise Development Scale	−0.058 * (−1.648)	−0.065 * (−1.928)	−0.007 (−0.349)
Telecommunication Infrastructure	0.129 *** (3.979)	0.041 ** (2.127)	0.014 (1.143)
Financial Support Intensity	0.105 * (1.739)	−0.015 (−0.382)	0.118 *** (4.430)
Household Savings Rate	−0.764 *** (−13.915)	0.226 *** (4.104)	−0.160 *** (−3.936)
Consumer Market Size	−0.005 (−0.137)	0.291 *** (7.759)	0.052 ** (2.172)
Foreign Investment Attractiveness	0.0177 * (1.761)	0.582 *** (9.894)	0.036 (1.148)
Social Welfare Level	−0.010 (−0.651)	0.008 (0.502)	0.022 ** (1.988)
_cons	6.505 *** (10.278)	−1.355 * (−1.844)	4.279 *** (10.839)
R2	0.976	0.979	0.968
AREA	Yes	Yes	Yes
YEAR	Yes	Yes	Yes
N	3336	4079	6158

Note: * p < 0.1; ** p < 0.05; *** p < 0.001. Robust standard errors in parentheses.

.

Table 6 showed that the policy significantly promotes HQAD in the Eastern and Middle, with effects decreasing from east to west. Specifically, the Eastern region displays the most pronounced benefits (β = 0.401, p < 0.001), followed by the Central region (β = 0.312, p < 0.001). In the Western region, the policy’s effect is not statistically significant (β = 0.050, p > 0.05), indicating regional disparities in its effectiveness and confirming Hypothesis 2 on regional heterogeneity. The policy impact follows the general economic gradient from East to Central to West. The Eastern region’s well-established market system, infrastructure, and openness create favorable conditions for return entrepreneurship, thus amplifying the policy’s positive impact. The lack of a significant impact in the Western region may be due to policy implementation challenges, resource mismatches, or complex industrial restructuring.

6. Discussion

6.1. Summary of Research Conclusions

This study comprehensively investigated the impact of the Pilot Policy of Migrant Workers Returning Home for Entrepreneurship on HQAD. Our baseline regression results demonstrate a significant positive effect of the policy on HQAD, indicating that its implementation has effectively promoted high-quality development in the agricultural sector. Regarding regional heterogeneity, the policy’s effects varied across different regions: in the Eastern region, it demonstrated the most pronounced impact, while the Central region also experienced a positive effect. However, in the Western region, the policy did not show a statistically significant influence on HQAD. Regarding the moderating effects, population agglomeration played a positive role, suggesting that areas with higher population density better enhance the policy’s positive impact on HQAD. Technological innovation also showed a strong positive moderating effect, highlighting the importance of local innovation capacity in leveraging returnee-driven agricultural development. Conversely, e-commerce development exhibited a negative moderating effect.

6.2. Heterogeneity Analysis and Integration with Previous Literature

Our findings substantiate and extend research on return migration and rural development in several ways. First, the significant positive effect of the pilot policy on HQAD corroborates studies emphasizing the transformative potential of return migrants in rural development [21]. However, our county-level analysis reveals substantial regional variation in the effectiveness of return entrepreneurship. In terms of regional heterogeneity, previous research has highlighted that different regions vary in their resource endowments, technological levels, and policy support in agricultural development [34]. Our findings align with these observations: geographically, the Eastern region possesses superior transportation infrastructure, location conditions, and levels of industrialization and urbanization compared to the Central and Western regions [7]. As a result, returning entrepreneurial farmers in the East can utilize these resource advantages to extend the agricultural industry chain and develop modern agriculture, thus creating a favorable environment for high-quality rural development. Our heterogeneity analysis further reveals this pattern, showing that the policy’s positive impact is strongest in the Eastern region, slightly lower but still significant in the Central region, while the effect in the Western region, though positive, lacks statistical significance. This gradient of effectiveness aligns with China’s established economic geography and echoes recent findings by Shen and Wang [32] on agricultural labor productivity, suggesting that regional differences in market systems, industrial integration, and human capital capabilities fundamentally shape how effectively such policies can drive agricultural transformation.

6.3. Mechanism Analysis and Theoretical Implications

In this study, we identified three core mechanisms that influence the impact of migrant workers’ return entrepreneurship on HQAD. Prior research has long recognized the importance of population density in regional economic development. Studies in regional economics indicate that higher population density can lead to economies of scale in various industries [52]. In the context of our study, in densely populated areas, there is an abundance of agricultural production factors that effectively allocate resources among different agricultural activities [32], thus positively moderating the transition towards HQAD. Previous studies have also underscored the role of technological innovation in rural development [53].Consistent with previous findings, our research shows that technological innovation serves as a positive moderator, likely because high-level technological innovation potential creates significant knowledge and technology spillover effects. These effects enable local labor to swiftly adapt to advanced agricultural technologies and equipment, providing returned entrepreneurs with more efficient means to enhance HDAQ. Some prior research has explored the impact of e-commerce on rural economies, primarily focusing on its benefits, such as expanding market access for rural products [42]. However, in regions with developed e-commerce, as our research shows, the direct promotional effect of the return entrepreneurship policy on HQAD may diminish. This is because the high-profit potential of e-commerce entices returned entrepreneurs to invest resources (including capital, human resources, and time) into e-commerce platform operations rather than traditional agricultural production [44]. In addition, most agricultural producers in China are small farmers who struggle to negotiate effectively with large commercial capital and their agents, leading to unequal exchange [54]. Our research thus introduces a new perspective to the existing literature by illuminating this potential drawback of e-commerce development in the context of return entrepreneurship and HQAD.

6.4. Policy Recommendations

First, the government should continue to implement and expand the scope of the returning-home-for-entrepreneurship pilot policy. A comprehensive support mechanism, including financial subsidies, technical guidance, and business training, should be established to further enhance high-quality rural development. Second, the critical roles of population agglomeration and technological innovation in advancing HQAD through returning entrepreneurship must be optimized. To encourage talent return, varied incentives for returnees’ families can be provided. In terms of technological innovation, sustained support for the R&D and application of practical agricultural technologies is crucial. Third, due to the regional variance in the policy’s impact on HQAD, differentiated policies are essential. For the economically advanced Eastern region, policy guidance should emphasize encouraging returning entrepreneurs to explore the multifunctional value of agriculture. In the Central and Western regions with relatively lower economic development levels, more resources should be dedicated to improving infrastructure related to returning entrepreneurship, including transportation, communication, and power supply. Additionally, inter-regional cooperation should be enhanced. The Eastern region can play a leading role in sharing its experience and resources, promoting the flow of factors like labor, capital, and technology between regions, and jointly promoting HQAD across the country.

6.5. Limitations and Future Research Directions

This study presents several limitations that point to valuable directions for future research. Although our 2014–2021 dataset is extensive, it may not fully capture the long-term effects of the policy. Extended longitudinal studies could determine whether the policy’s impact either strengthens or weakens over time. Additionally, the DID model, predicated on a linear assumption, might not faithfully represent the potentially non-linear effects of the pilot policy on HQAD. Future research could adopt non-linear econometric models to more effectively depict these relationships. Further, conducting detailed case studies in selected counties would enable a qualitative analysis of the intricate policy influence processes. Moreover, while our county-level analysis offers strong evidence of the policy’s impact, integrating mixed methods in future research could clarify the specific mechanisms through which returnee entrepreneurs affect local agricultural practices. Exploring potential spatial spillover effects between pilot and non-pilot counties would also enhance understanding of how successful practices are transferred across administrative boundaries. Cross-national comparative studies could shed light on how different institutional contexts impact the effectiveness of return migration policies in rural development. Finally, in light of our findings on e-commerce’s negative moderating effect, future research should focus on how returnee entrepreneurs navigate and potentially reshape digital agricultural markets.

7. Conclusions

This study provides a comprehensive analysis of the Pilot Policy of Migrant Workers Returning Home for Entrepreneurship and its influence on HQAD at the county level in China. Our empirical findings lead to three main conclusions:

First, the pilot policy has a significant positive effect on HQAD, affirming its role as an effective rural development strategy. This strong positive correlation is consistent across various model specifications, indicating that returnee entrepreneurs play a crucial role in driving agricultural modernization through their accumulated human capital and innovative capabilities.

Second, our findings indicate considerable regional differences in the effectiveness of the policy, with the most substantial impacts observed in the Eastern region, followed by the Central region. Conversely, the Western region shows minimal statistical significance. This variation emphasizes the importance of regional resources and developmental stages in shaping policy outcomes, underscoring the need for customized implementation strategies.

Third, our analysis of the mechanisms involved demonstrates complex interactions between the policy and local conditions. Factors such as population concentration and technological innovation beneficially influence the policy’s effectiveness, whereas the growth of e-commerce exerts a detrimental moderating effect. These findings highlight the importance of considering local contextual factors in policy implementation and suggest that the success of return entrepreneurship initiatives relies heavily on aligning returnee capabilities with local development needs.

These conclusions broaden our understanding of return migration policies and rural development, offering practical insights for policymakers aimed at promoting agricultural modernization through returnee entrepreneurship. The findings advocate for future rural development policies to employ more nuanced, region-specific approaches while accounting for the complex interplay between human capital, innovation capacity, and digital market development.