Rural Industrial Integration and Economic Performance of Leading Agricultural Enterprises: Firm-Level Heterogeneity in Jiangxi, China

Abstract

Rural industrial integration has become a key strategic direction for promoting agricultural transformation and rural revitalization, with the economic performance of leading agricultural enterprises serving as a critical metric for evaluating its success. However, empirical evidence on the characteristics and boundary conditions of how rural industrial integration affects the economic performance of these enterprises remains limited. This study investigates this relationship using panel data from 821 leading agricultural enterprises in Jiangxi Province from 2021 to 2023, analyzed with a random-effects model. The results reveal a significant positive statistical association between the degree of rural industrial integration and firm economic performance, which remains robust across multiple reliability checks. Further analysis indicates that this positive association varies considerably across different firm characteristics, exhibiting four forms of contingent heterogeneity: “size-reverse,” “leverage-positive,” “age-U-shaped,” and “subsidy-dampening.” Based on these findings, two policy recommendations are proposed: first, implement a differentiated support strategy that prioritizes guiding small-sized enterprises and moderately leveraged firms into rural industrial integration activities, together with phase-specific support measures for firms at different life-cycle stages; second, improve the allocation efficiency of fiscal subsidy funds, alongside the supporting systems and long-term mechanisms for rural industrial integration.

1. Introduction

The integrated development of primary, secondary, and tertiary industries in rural areas has been widely recognized as a key strategy for promoting rural revitalization, increasing farmer incomes, and achieving agricultural modernization [1,2,3]. Leading agricultural enterprises act as key agents driving rural industrial integration, playing a central role in connecting smallholder farmers to markets, facilitating technological innovation, and upgrading value chains [4,5]. Operating on a for-profit basis, these firms’ economic performance directly affects their willingness and capacity to engage in rural industrial integration, thereby providing the foundation for sustainable industrial integration. In 2025, the Chinese government issued the policy document “Building an Agricultural Powerhouse: A Plan for 2024–2035,” which explicitly calls for promoting integrated rural industrial development, upgrading production, processing, and distribution across all segments, and expanding value-added opportunities for the sector. Leading agricultural enterprises are positioned within this document as the core vehicles of rural industrial integration. National policy has therefore placed strong emphasis on both rural industrial integration and the development of these enterprises. According to the 2024 “Development Analysis Report on New Agricultural Business Entities in China,” there are 2285 nationally recognized leading agricultural enterprises and approximately 90,000 at or above the county level. These enterprises have become a central force linking smallholder farmers to broader markets and spearheading agricultural industrialization. A review of the existing literature organizes previous research into three main strands.

From the perspective of rural industrial integration, a large body of literature has focused on its macroeconomic effects in rural areas. Evidence confirms that rural industrial integration significantly raises farmer incomes by extending agricultural value chains, lowering transaction costs, and creating off-farm employment opportunities [1]. It also enhances agricultural total factor productivity through technological progress and efficiency gains, although this effect exhibits considerable regional heterogeneity [6,7]. In addition, rural industrial integration strengthens the economic resilience of the agricultural sector and improves firms’ adaptive capacity [3,8]. By raising green total factor productivity, rural industrial integration helps narrow the urban–rural income gap [9,10]. Other studies have examined the drivers of rural industrial integration. The digital economy is considered a core driving force, operating through technological innovation and human capital as intermediary channels, with its influence taking nonlinear forms—such as an inverted U-shape—or exhibiting threshold effects [11,12]. Fiscal support for agriculture, digital inclusive finance, and transportation infrastructure have also been identified as important drivers [13,14,15]. That said, the macro-level studies described above treat rural industrial integration as an aggregate regional phenomenon. They focus primarily on its overall effects on farmers, the agricultural sector, or the urban–rural divide. Very little attention has been paid to how rural industrial integration affects individual enterprises, especially the leading agricultural enterprises that serve as the main agents of this process.

Turning to the functional positioning and mechanisms of leading agricultural enterprises within rural industrial integration, existing research largely portrays these firms as institutional bridges coordinating vertical collaboration along agricultural value chains. Some studies point to the incomplete nature of traditional “leading enterprise + smallholder” contracts, which tend to invite opportunistic behavior. Introducing cooperatives or moving toward vertical integration can markedly improve collaboration stability [5]. By providing services such as technology, capital, and machinery, leading enterprises attract farmers into vertical chain coordination, with a mix of factors driving this process [16]. Leading enterprises also play an important role in guiding farmers toward pre-production quality and safety control, realized mainly through organizational linkage mechanisms and production factor linkages tied to new quality productive forces [17]. An evolutionary game model suggests that cooperative behavior among governments, leading agricultural enterprises, and farmers evolves toward the desired equilibrium only under specific conditions relating to subsidies, penalties, and resource endowments [18]. While these studies clarify that leading agricultural enterprises act as facilitators of rural industrial integration, very few have quantified the economic returns—such as return on assets or sales profit margin—that these enterprises gain from participating in rural industrial integration. Looking at the factors that shape the economic performance of leading agricultural enterprises, digital transformation has been identified as a key driver of high-quality development and green transition, although its effect operates through a threshold mechanism: a certain level of digitalization must be crossed before positive effects emerge [19,20]. Firm-level endowments, technological innovation, and entrepreneurial traits serve as core drivers of sustainable development [21]. On the policy front, China’s Leading Agricultural Enterprise Program (ALFP) has significantly boosted productivity, yet this effect only appears among non-state-owned firms; state-owned enterprises remain unaffected. The mechanism behind this difference lies in credit support flowing primarily to state-owned firms, whereas non-state-owned firms receive tax benefits and subsidies [22]. Spatial factors also merit attention: headquarters of leading enterprises tend to be city-oriented, while their branches follow raw material locations, with each type influenced by different locational factors [23]. That said, no study has directly tested whether the level of rural industrial integration at the regional level is associated with the economic performance of individual leading enterprises. Nor has any research systematically examined how this association varies across different firm characteristics—such as firm size, debt level, life-cycle stage, or fiscal support.

Synthesizing the three strands of literature reviewed above, macro-level studies confirm that rural industrial integration generates aggregate economic benefits but treat this phenomenon solely as a regional phenomenon, leaving micro-level firm analysis largely underdeveloped. Functional studies position leading enterprises as facilitators of rural industrial integration but fail to measure the economic returns these firms obtain from the convergence process. The micro-level empirical link between rural industrial integration and the economic performance of leading agricultural enterprises remains untested, and firm-level heterogeneity in terms of size, debt level, life-cycle stage, and fiscal support has not been systematically explored. The lack of micro-level evidence prevents policymakers from determining whether rural industrial integration truly benefits leading agricultural enterprises, and makes it difficult to avoid uniform policies that may prove ineffective or even counterproductive for certain types of firms. To address this gap, this study employs a random-effects model to examine the statistical association between rural industrial integration and firm performance, and further analyzes the heterogeneous nature of this association along four dimensions: firm size, debt level, firm age, and the amount of fiscal support received. Jiangxi Province provides an ideal setting for this investigation. As a major agricultural province actively implementing rural industrial integration policies under China’s rural revitalization strategy, it offers both policy relevance and sufficient variation in firm characteristics. Moreover, focusing on a single province helps control for cross-provincial institutional differences, thereby enhancing the internal validity of the findings. This paper makes three main contributions. First, it reveals a positive association between rural industrial integration and the economic performance of leading agricultural enterprises at the micro level. Second, it identifies four forms of heterogeneity—“size-reverse,” “leverage-positive,” “age-U-shaped,” and “subsidy-dampening”—providing empirical reference points for understanding the boundary conditions of the rural industrial integration–performance link. Third, by focusing on Jiangxi Province, it enhances the applicability of the research findings within a provincial context.

2. Theoretical Analysis and Research Hypotheses

2.1. The Logic Underlying the Relationship Between Rural Industrial Integration and the Economic Performance of Leading Agricultural Enterprises

Rural industrial integration led by leading agricultural enterprises represents a process through which firms optimize resource allocation and realize value appreciation. From the perspective of the relationship’s characteristics, a multi-dimensional positive link exists between rural industrial integration and the economic performance of these enterprises. First, as firms push forward with rural industrial integration, strategic adjustments and organizational optimization can improve their efficiency in resource allocation, thereby lowering transaction costs between industries and generating a positive relationship with economic growth [24]. Agricultural enterprises that promote the convergence of primary, secondary, and tertiary industries through activities such as deep processing of agricultural products, warehousing and logistics, and agritourism can extend value chains, expand their operational scale, and raise product value added [25], further strengthening this positive link. Second, driven by the digital economy, leading agricultural enterprises that adopt smart equipment, digital platforms, and e-commerce technologies are able to significantly improve production efficiency and management standards [26]. This technology-driven pathway to rural industrial integration also exhibits a positive association with firm economic performance. By collaborating with technology-oriented firms on agricultural technology projects, leading agricultural enterprises can increase the efficiency of technology transfer and generate notable technology spillovers [27], thereby improving economic performance and reinforcing the positive relationship. Third, participation in rural industrial integration broadens the business scope of leading agricultural enterprises, enriches their product and service portfolios, and facilitates the sharing of brand, channel, and customer resources across different business segments. This resource sharing can generate synergies among business units. which in turn creates a positive link with firm economic performance [28,29]. Fourth, leading agricultural enterprises engaged in rural industrial integration are more likely to receive policy support in the form of government subsidies, tax benefits, and financial assistance [30,31]. Such support lowers the costs associated with participation in rural industrial integration and contributes to a positive relationship with firm economic performance. Based on the above reasoning, the first hypothesis is proposed:

Hypothesis 1 (H1):

Rural industrial integration exhibits a significant positive association with the economic performance of leading agricultural enterprises.

2.2. Heterogeneity in the Association Between Rural Industrial Integration and Firm Economic Performance Across Different Firm Characteristics

The positive association between rural industrial integration and the economic performance of leading agricultural enterprises is unlikely to be uniform across different firm characteristics—consider firm size first. Small and medium-sized enterprises tend to be organizationally more flexible and faster in decision-making, advantages that may help them reap greater benefits from rural industrial integration; adding to this, their lower performance baseline leaves more room for marginal improvement, so the positive association is stronger among these smaller firms. Large enterprises, by contrast, despite their resource abundance, often struggle with organizational inertia and slower decision-making processes, which tend to limit their marginal gains and may even introduce a negative moderating effect [32,33], potentially weakening the positive association. In terms of debt level, firms with moderate to high debt generally possess stronger financing capacity and a greater willingness to expand; with sufficient funds at hand, these firms are better positioned to engage in rural industrial integration projects, thereby driving economic performance. For this reason, the positive association is more pronounced among moderately to highly indebted firms, whereas low-debt firms face financing constraints or lack incentives for expansion—both of which hinder their participation in rural industrial integration and limit performance growth [34]—making the association insignificant for this group. Firm age tells a more nuanced story: young firms are relatively agile and highly motivated to innovate; mature firms command abundant resources and rich experience; both types appear capable of translating rural industrial integration into performance gains quickly. Middle-aged firms, however, fall into an awkward position—they lack the innovation edge of young firms while not Yet possessing the resource endowments of mature ones, and worse still, path dependence and organizational inertia may drag their performance down. Thus, whether the positive association holds across different life-cycle stages remains an open question [35]. Finally, fiscal support cuts both ways: on the one hand, it eases financing constraints and lowers the costs of participating in rural industrial integration; on the other hand, excessive fiscal support risks creating subsidy dependence and crowding-out effects, thereby eroding firms’ endogenous motivation [36,37]. As a result, the positive association between rural industrial integration and firm economic performance varies depending on the level of fiscal support received. Based on the above reasoning, we propose the second hypothesis:

Hypothesis 2 (H2):

The positive association between rural industrial integration and the economic performance of leading agricultural enterprises varies significantly across firm size, debt level, firm age, and fiscal support.

Figure 1 summarizes the conceptual model of this study, illustrating the characteristics of the association between rural industrial integration and the performance of leading agricultural enterprises, as well as the heterogeneous effects.

3. Data Sources, Variable Descriptions, and Model Specification

3.1. Data Sources

This study used panel data on leading agricultural enterprises in Jiangxi Province from 2021 to 2023. The data used in this study are derived from a monitoring survey of leading agricultural enterprises conducted by the agricultural authorities of Jiangxi Province, China. In compliance with a data use agreement, all enterprise-identifying information (e.g., firm names, registration numbers) was anonymized prior to analysis, and the data are used solely for academic research purposes in accordance with the Statistics Law of the People’s Republic of China. The systematic monitoring work has only been established for a relatively short period. and the data content became relatively complete starting from 2021; accordingly, this study selects the 2021–2023 period as the observation window, which represents the longest continuous three-year panel data currently available. This time window offers two additional advantages. First, 2021 marks the beginning of China’s 14th Five-Year Plan period, during which Jiangxi Province introduced a series of policies to promote the integrated development of rural primary, secondary, and tertiary industries, and taking this year as the starting point allows the analysis to capture the implementation effects of these rural industrial integration policies within a relevant policy context. Second, employing three consecutive years of panel data, rather than a single cross-section, enables better control of firm-level heterogeneity and facilitates the identification of dynamic associations between rural industrial integration and firm performance. The sample covers all 11 prefecture-level cities in Jiangxi Province and closely matches the spatial distribution of the province’s agricultural industry. It should be noted that the “leading agricultural enterprises” in our sample refer to firms whose production and operation activities are rooted in rural areas. These enterprises form close benefit-sharing linkages with local farming households through mechanisms such as contract farming and cooperative partnerships. According to the classification criteria of Jiangxi Province’s agricultural industrialization authorities, even for enterprises that maintain administrative or sales offices in urban areas, their primary production bases and processing facilities remain located in rural areas. Therefore, the leading agricultural enterprises analyzed in this study are essentially rural-based business entities. the business activities of the sampled leading agricultural enterprises span a variety of agricultural products—including grain, fruit, vegetables, and livestock—and cover multiple stages such as planting, processing, storage, transportation, and sales. The indicators include core elements such as rural industrial integration practices, firm assets, operational efficiency, and farmer-driven benefits, which align well with the objectives of this study and provide strong representativeness. To improve the accuracy of the empirical analysis, enterprises not continuously monitored throughout the 2021–2023 period were excluded. Observations with missing or abnormal values were also removed. All data cleaning and statistical analyses were performed using Stata 18.0. After these cleaning steps, a balanced panel dataset of 821 leading agricultural enterprises in Jiangxi Province was obtained, yielding 2463 firm-year observations.

3.2. Variable Descriptions

3.2.1. Dependent Variable

The dependent variable in this study is the economic performance of leading agricultural enterprises. Following established practices from Tao et al. [38], Zhou and Chen [39], Wang and Deng [40], return on assets (ROA) was employed as the proxy variable for economic performance, as it adequately reflects firms’ asset returns and profitability.

3.2.2. Core Independent Variable

The core independent variable is rural industrial integration. Drawing on Li and Ran [41], Zhao et al. [42], Sun and Yang [43], and taking into account the specific characteristics of leading agricultural enterprises in Jiangxi Province, rural industrial integration behaviors led by these enterprises were categorized into four core dimensions: value chain extension, multi-functionality expansion, technology penetration, and benefit linkage. A rural industrial integration measurement index system was constructed based on these dimensions (see

Table 1. Measurement index system for rural industrial integration.

Primary Indicator	Secondary Indicator	Variable Measurement and Definition	Value Range
Rural industrial integration	Industrial Chain Extension (X1)	1. Whether the firm involves primary industry (agricultural product planting); 0 = No; 1 = Yes 2. Whether the firm involves secondary industry (agricultural product processing); 0 = No; 1 = Yes 3. Whether the firm involves tertiary industry (agricultural product sales); 0 = No; 1 = Yes (Each firm involves at least one industry, so the starting value is 1)	1–3
	Multi-functional Expansion (X2)	1. Whether the firm involves production function; 0 = No; 1 = Yes (All firms involve production function, so the starting value is 1) 2. Whether the firm involves living function (leisure agriculture, rural tourism firms); 0 = No; 1 = Yes 3. Whether the firm involves ecological function (firms adopting green production technologies or obtaining green product certification); 0 = No; 1 = Yes	1–3
	New Technology Penetration (X3)	1. Whether the firm has established an R&D institution; 0 = No; 1 = Yes 2. Whether the firm has R&D personnel; 0 = No; 1 = Yes 3. Whether the firm has invested in R&D funds; 0 = No; 1 = Yes	0–3
	Benefit Linkage Mechanism (X4)	1. Whether the firm links with and drives agricultural industrialization consortia; 0 = No; 1 = Yes 2. Whether the firm links with and drives family farms; 0 = No; 1 = Yes 3. Whether the firm links with and drives farmer cooperatives; 0 = No; 1 = Yes	0–3

). The entropy method was applied to assign weights to each dimension, and the composite score of the four dimensions was used as the proxy variable for the level of rural industrial integration. However, it is important to note that: The industrial chain extension dimension is measured by the number of primary, secondary, and tertiary industries a firm engages in. Because every firm engages in at least one industry, this indicator starts at 1 and ranges from 1 to 3. The multi-functional expansion dimension is measured by the number of production, living, and ecological functions a firm engages in. Since all firms possess the production function, this indicator also starts at 1 and ranges from 1 to 3. In contrast, the technology penetration and benefit linkage dimensions do not share this feature—firms may have no R&D institutions, personnel, or funding, or may drive no consortia, family farms, or cooperatives. Thus, these two dimensions range from 0 to 3.

This study uses the entropy method to measure the level of rural industrial integration. The specific calculation steps are as follows:

(1)

Data standardization.

For the j-th indicator ${\mathrm{x}}_{\mathrm{i}\mathrm{j}}(\mathrm{i}=1,\dots ,n)$, data standardization is performed as follows:

$$\begin{gathered} {\mathrm{y}}_{\mathrm{i}\mathrm{j}}={\displaystyle \frac{{\mathrm{x}}_{\mathrm{i}\mathrm{j}}-\mathrm{m}\mathrm{i}\mathrm{n}\left({\mathrm{x}}_{\mathrm{j}}\right)}{\mathrm{m}\mathrm{a}\mathrm{x}\left({\mathrm{x}}_{\mathrm{j}}\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left({\mathrm{x}}_{\mathrm{j}}\right)}} \\ {\mathrm{y}}_{\mathrm{i}\mathrm{j}}\in [0, 1] \end{gathered}$$

(2)

Calculate the proportion (weight) of the i-th sample under the j-th indicator.

$${\mathrm{p}}_{\mathrm{i}\mathrm{j}}={\displaystyle \frac{{\mathrm{y}}_{\mathrm{i}\mathrm{j}}}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{y}}_{\mathrm{i}\mathrm{j}}}}$$

(3)

Calculate the information entropy of the j-th indicator.

$${\mathrm{e}}_{\mathrm{j}}=-{\displaystyle \frac{1}{\ln \mathrm{n}}}{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}}{\mathrm{p}}_{\mathrm{i}\mathrm{j}}\ln \left({\mathrm{p}}_{\mathrm{i}\mathrm{j}}\right)$$

(4)

Calculate the coefficient of variation.

$${\mathrm{d}}_{\mathrm{j}}=1-{\mathrm{e}}_{\mathrm{j}}$$

(5)

Calculate the weight of the j-th indicator.

$${\mathrm{w}}_{\mathrm{j}}={\displaystyle \frac{{\mathrm{d}}_{\mathrm{j}}}{{\displaystyle {\sum }_{\mathrm{k}=1}^{\mathrm{m}}}{\mathrm{d}}_{\mathrm{k}}}}$$

(6)

Calculate the comprehensive score of the i-th sample.

$$\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}={\displaystyle \sum _{\mathrm{j}=1}^{\mathrm{m}}}{\mathrm{w}}_{\mathrm{j}}{\cdot \mathrm{y}}_{\mathrm{i}\mathrm{j}}$$

All selected indicators were drawn from the monitoring data of leading agricultural enterprises from 2021 to 2023, allowing for a reasonably objective characterization of the basic forms of agricultural rural industrial integration. Figure 2 presents the kernel density distribution of the rural industrial integration index for the 821 sample firms. The index ranges from approximately 0.1 to 0.8, with the peak located near the interval of 0.2–0.3. The overall distribution exhibits a slight rightward skew. Despite this skewness, the empirical distribution is broadly consistent with the theoretical normal distribution and may be treated as approximately normal. This distributional feature provides a viable empirical basis for subsequent parameter estimation and heterogeneity analysis.

3.2.3. Control Variables and Grouping Variables

Drawing on the studies of Zhang et al. [44], Xie et al. [45], and Jin et al. [46], this study identifies a set of variables that may influence firms’ resource acquisition capacity, production efficiency, market bargaining power, and risk resilience. However, due to differences in analytical objectives, we adopt three variables—ownership type, Quality management capacity, and regional economic development level—as baseline control variables, The aforementioned literature consistently identifies them as key determinants of both industrial convergence adoption and firm performance. The remaining four variables—firm size, debt level, firm age, and fiscal support—are treated as heterogeneity analysis variables, as they are theoretically conceptualized as contextual factors shaping the efficiency of industrial convergence transformation (e.g., through differential resource allocation, risk-bearing capacity, or policy responsiveness). It should be noted that the concern remains that these grouping variables may act as omitted confounders. Therefore, we will conduct two validation procedures in the robustness checks section (Section 4.3). First, we will include all four grouping variables simultaneously into the baseline regression model (i.e., the “full model”) and compare the coefficient of the core independent variable between the parsimonious model (with three control variables) and the full model (with all seven variables). This comparison primarily serves to demonstrate that the four grouping variables do not function as significant controls that substantially alter the overall model results. Second, following the framework of Oster [47], we focus on the stability of the core coefficient across models. If the magnitude of the coefficient change falls within a reasonable range (typically considered as less than 20%), this suggests that omitted variable bias due to excluding these four variables is unlikely to be severe. The definitions and measurements of all variables are presented in

Table 2. Variable definitions and value assignments.

Variable Name	Definition and Value Assignment
Economic performance	Return on assets (ROA) = after-tax profit/total assets
Rural industrial integration	Entropy value of the four core dimensions: value chain extension, multi-functionality expansion, technology penetration, and benefit linkage
Firm size	Number of employees
Ownership type	Ownership type: 0 = state-owned; 1 = non-state-owned
Debt level	Debt-to-asset ratio (%)
Firm age	Firm age = survey year − year of establishment
Quality management capacity	Whether the firm has passed ISO 9001 [48] or other quality certification: 0 = no; 1 = yes
Fiscal support	Total amount of fiscal subsidy funds received (10,000 RMB)
Regional economic development level	Gross domestic product (GDP) at the prefecture-level city level (100 million RMB)

.

3.3. Model Specification

The primary objective of this study is to examine the association between rural industrial integration and the economic performance of leading agricultural enterprises, rather than to establish a causal relationship between the two. To determine the appropriate estimator for panel data analysis, the Hausman specification test was conducted [49]. The test yielded a chi-squared statistic of 2.83 with 6 degrees of freedom, corresponding to a p-value of 0.8296, failing to reject the null hypothesis that the individual-level random effects are uncorrelated with the regressors. Therefore, compared to the fixed effects estimator, the random effects estimator is more efficient and does not introduce inconsistency. Accordingly, the random effects model was adopted for this study. In addition, time-fixed effects were included in the model to account for common time trends. The specific model specification is as follows:

$${FP}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{IC}_{\mathrm{it}}+\sum {\alpha }_{\mathrm{k}}{\mathit{Control}}_{\mathrm{kit}}+{\mu }_{\mathrm{i}}+{\lambda }_{\mathrm{t}}+{\epsilon }_{\mathrm{it}}$$

where ${FP}_{\mathrm{it}}$ denotes the economic performance variable of firm i in year t, and ${IC}_{\mathrm{it}}$ represents the rural industrial integration variable of firm i in year t. ${\mathsf{\alpha }}_0$ is the intercept term, and ${\mathsf{\alpha }}_1$ is the coefficient of the rural industrial integration variable. ${\mathit{Control}}_{\mathrm{kit}}$ is the value of the k-th control variable for firm i in year t. with ${\mathsf{\alpha }}_{\mathrm{k}}$ being its corresponding coefficient ${\mu }_i$ denotes the firm-specific random effect, which captures unobserved time-invariant heterogeneity across firms and is assumed to be uncorrelated with the explanatory variables. ${\mathsf{\lambda }}_{\mathrm{t}}$ denotes time-fixed effects, and ${\epsilon }_{it}$ is the idiosyncratic error term.

4. Empirical Analysis

4.1. Descriptive Statistics

This study employs Stata 18 for data processing and analysis. To prevent extreme observations from biasing the baseline regression results, a 1% Winsorization procedure was applied uniformly to economic performance, firm size, debt level, and the alternative indicator of economic performance. To address potential heteroscedasticity and improve the normality of data distributions, natural logarithmic transformations were applied to the original variables. Specifically, economic performance, firm debt level, and government support were transformed using ln(x + 1) due to the presence of zero values in the original measurements. Firm size, firm age, and regional economic development level were directly transformed using ln(x). After transformation, all variables exhibited approximately normal distributions, rendering them suitable for subsequent regression analyses.

Table 3. Summary statistics of key variables.

Variable Type	Variable Name	Obs.	Mean	Std. Dev.	Min	Max
Dependent Variable	Economic performance	2463	0.088	0.071	−0.612	0.879
Independent Variable	Rural industrial integration	2463	0.183	0.126	0	0.638
Control Variables	Ownership type	2463	0.958	0.201	0	1
	Quality management capacity	2463	0.476	0.5	0	1
	Regional economic development level	2463	6646.34	1442.231	1102.31	7324.46
Grouping Variables	Firm size	2463	375.642	1304.312	10	30,000
	Debt level	2463	26.581	15.782	0	148.41
	Firm age	2463	12.613	5.895	1	39
	Fiscal support	2463	139.978	524.631	0	12,126.94

presents the descriptive statistics of the variables after Winsorization.

4.2. Baseline Regression Analysis

A random-effects model was employed, with control variables added sequentially, and the results are presented in

Table 4. Baseline regression results.

Economic Performance	(1)	(2)	(3)	(4)
Rural industrial integration	0.036 *** (0.013)	0.039 *** (0.013)	0.042 *** (0.013)	0.042 *** (0.013)
Ownership type	/	0.031 *** (0.008)	0.029 *** (0.007)	0.030 *** (0.007)
Quality management capacity	/	/	−0.007 *** (0.003)	−0.008 *** (0.003)
Regional economic development level	/	/	/	0.003 (0.004)
Year 2022	−0.005 *** (0.001)	−0.005 *** (0.001)	−0.005 *** (0.001)	−0.006 *** (0.001)
Year 2023	−0.011 *** (0.002)	−0.011 *** (0.002)	−0.011 *** (0.002)	−0.011 *** (0.002)
Constant	0.082 *** (0.003)	0.052 *** (0.008)	0.056 *** (0.008)	0.028 (0.034)
N	2463	2463	2463	2463
Overall R2	0.014	0.028	0.036	0.037

Notes: Clustered standard errors in parentheses (clustered at the individual level). *** p < 0.01.

. Model (1) includes only the core independent variable and year dummies. Models (2) through (4) add control variables in sequence. The core independent variable remains positive and statistically significant at the 1% level across all specifications, confirming a positive statistical association between rural industrial integration and the economic performance of leading agricultural enterprises. Hypothesis H1 is therefore supported. After the stepwise inclusion of control variables, both the magnitude of the coefficient on the core independent variable and its significance level remain stable. The year fixed effects show that, relative to the baseline year 2021, the coefficients on the year dummies for 2022 and 2023 are significantly negative at the 1% level, a pattern that aligns closely with the macroeconomic shock of the COVID-19 pandemic.

Among the control variables, ownership type is significantly positive at the 1% level, indicating that ownership type is associated with firm performance. Quality management capacity is significantly negative at the 1% level, suggesting that stricter quality management standards may temporarily affect performance conversion. The coefficient on regional economic development level is not significant, implying that this external factor has little influence on the positive association between rural industrial integration and firm performance. As control variables are gradually incorporated, the overall R² of the model increases from 0.014 to 0.037, indicating a progressive improvement in the model’s explanatory power. Although this R² value falls within an acceptable range for micro-level firm studies, it remains relatively low, suggesting that firm economic performance is influenced by many other factors not captured by our model, such as managerial capability, organizational culture, market competition, and exogenous shocks. Therefore, caution should be exercised when interpreting the findings of this study.

4.3. Robustness Checks

To verify the reliability of the above findings, several robustness tests were conducted, including alternative model estimation, sample substitution, 5% Winsorization, and a one-year lag of the core independent variable.

Table 5. Robustness Check Results.

Variable	(1)	(2)	(3)	(4)	(5)
Rural industrial integration	0.042 *** (0.013)	0.044 *** (0.012)	0.040 *** (0.014)	0.041 *** (0.011)	0.041 *** (0.012)
Controls	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Observations	2463	2463	1642	2463	1642
R2/Overall R2	0.037	0.038	0.033	0.039	0.033

Notes: Clustered standard errors in parentheses (clustered at the individual level). *** p < 0.01. Column (1) reports R² from pooled OLS; Columns (2)–(5) report overall R² from random-effects models.

presents the results of these robustness checks, with column (1) serving as the baseline reference.

4.3.1. Alternative Model Estimation

A pooled cross-sectional model with clustering at the firm level was employed to re-estimate the relationship, testing whether the baseline results are sensitive to model specification. The results are presented in Table 5, column (2). The coefficient of the core independent variable (rural industrial integration) is 0.044, significant at the 1% level, with its sign and significance largely consistent with the baseline results. This finding suggests that the baseline regression results are not sensitive to model specification and demonstrate reasonable robustness.

4.3.2. Sample Substitution

Data for the year 2021 were excluded, and the regression was re-estimated using only the 2022–2023 sample. The results are shown in Table 5, column (3). The coefficient of rural industrial integration is 0.040, significant at the 1% level, remaining largely unchanged in both direction and significance compared to the baseline model. This provides further evidence supporting the robustness of the baseline findings.

4.3.3. Five-Percent Winsorization

A 5% Winsorization procedure was applied to economic performance, firm size, debt level, and related alternative indicators, followed by re-estimation of the regression model. The results are reported in Table 5, column (4). The coefficient of rural industrial integration is 0.041, significant at the 1% level, which is highly consistent with the baseline regression results. This indicates that the baseline results are not driven by extreme values, further confirming the robustness of the estimates.

4.3.4. Omitted Variable Bias Test

To examine whether excluding firm size, debt level, firm age, and fiscal support from the baseline model leads to severe omitted variable bias, we estimated a “full model” including all seven control variables and compared it with the “parsimonious model” (containing only three control variables) used in the main analysis.

Table 6. Omitted Variable Bias Test—Model Comparison.

Model	Core EV Coefficient	N	R2
Parsimonious Model (3 controls)	0.042 *** (0.013)	2463	0.037
Full Model (7 controls)	0.035 *** (0.013)	2463	0.090

Notes: Clustered standard errors in parentheses (clustered at the individual level). *** p < 0.01.

presents the regression results for both models. The coefficient of the core independent variable (rural industrial convergence,) is 0.042 (p < 0.01) in the parsimonious model and 0.035 (p < 0.01) in the full model. In both specifications, the coefficient of the core independent variable remains positive and highly significant, demonstrating strong consistency. Following the framework of Oster [49], we focus on the magnitude of the coefficient change. The coefficients are similar across the two models, with a difference of approximately 16.7%, which falls within a reasonable range (typically considered as less than 20%). Therefore, the possibility that these four variables lead to severe omitted variable bias is low. In conclusion, treating these four variables as grouping variables rather than baseline controls is justified.

4.3.5. Mitigating Concerns of Reverse Causality

The baseline regression results indicate a positive statistical association between rural industrial integration and the economic performance of leading agricultural enterprises. Nevertheless, reverse causality remains a concern—better-performing firms may be more willing or better positioned to engage in rural industrial integration, potentially confounding the estimated association. To preliminarily assess this possibility, a one-period lag of the core independent variable was introduced as an auxiliary test. The logic underlying this test is straightforward: current economic performance cannot influence past levels of rural industrial integration. Thus, if a significant positive association still holds between the lagged integration measure and current performance, confidence in the robustness of the baseline findings would be strengthened to some extent. As shown in Table 5, column (5), the coefficient of the one-period lagged rural industrial integration is 0.041 and statistically significant at the 1% level, broadly mirroring the baseline estimates. This finding suggests that earlier rural industrial integration tends to be positively correlated with firm economic performance in subsequent years. In addition, year fixed effects were included in the model to absorb macro-level time trends. These measures jointly help support the robustness of the results. While these analyses offer several key pieces of evidence for the association between rural industrial integration and economic performance, rigorous causal claims about their relationship are not feasible. Accordingly, this study does not completely exclude the possibility of reverse causality and restricts its conclusions to the associational level.

4.4. Heterogeneity Analysis

To further examine the heterogeneous effects of rural industrial integration on the economic performance of leading agricultural enterprises, the sample was divided into low, medium, and high groups based on firm size, debt level, firm age, and fiscal support, with separate regression analyses conducted for each group. A tertile grouping approach was adopted for the following reasons. First, these variables represent key firm characteristics that influence resource endowments, financing capacity, managerial experience, and the intensity of policy support, and their moderating effects on the association between rural industrial integration and economic performance may be particularly pronounced. Second, the sample size of 2463 observations meets the requirements for tertile grouping. Third, tertile grouping allows for a more intuitive presentation of coefficient differences across groups, helping to reveal potential “interval-dependent” patterns of these variables. Therefore, the tertile grouping approach balances between-group comparability and within-group homogeneity, preserving sample information while effectively capturing how the benefits of rural industrial integration evolve as each variable moves from low to high levels. The results of the tertile group regressions are presented in

Table 7. Heterogeneity analysis results.

Variable	Low Group	Medium Group	High Group
Firm size	0.089 *** (0.020)	0.050 ** (0.023)	−0.014 (0.022)
N	841	802	820
Debt level	0.001 (0.022)	0.060 *** (0.019)	0.055 *** (0.019)
N	821	822	820
Firm age	0.051 ** (0.020)	0.008 (0.022)	0.060 *** (0.021)
N	899	776	788
Fiscal support	0.092 *** (0.024)	0.037 ** (0.019)	0.032 * (0.017)
N	841	801	821

Notes: Clustered standard errors are reported in parentheses. *, **, *** indicate significance at the 10%, 5%, and 1% levels, respectively.

.

4.4.1. Heterogeneity by Firm Size

The positive association between rural industrial integration and firm economic performance is most pronounced among small-sized enterprises (low group: β = 0.089, p < 0.01), followed by medium-sized enterprises (medium group: β = 0.050, p < 0.05), but is not statistically significant among large-sized enterprises (high group: β = −0.014, p > 0.10), exhibiting a clear “size-reverse” characteristic. A possible interpretation is that small and medium-sized leading agricultural enterprises start from a lower performance baseline, leaving greater room for marginal improvement, and their organizational structures are relatively flexible with higher decision-making efficiency, making them better able to translate rural industrial integration opportunities into actual performance gains. In contrast, large-scale enterprises may face organizational inertia and departmental silos—so-called “big firm disease”—which limit the marginal contribution of industrial convergence, potentially explaining the lack of a significant association for this group.

4.4.2. Heterogeneity by Debt Level

The positive association between rural industrial integration and firm economic performance is not significant in the low-debt group (β = 0.001, p > 0.10) but is significant in both the medium-debt group (β = 0.060, p < 0.01) and the high-debt group (β = 0.055, p < 0.01), displaying a clear “leverage-positive” characteristic. A possible interpretation is that firms with moderate to high debt levels tend to have stronger financing capabilities and a greater willingness to expand, thereby facilitating the implementation of rural industrial integration projects and their translation into performance gains. By comparison, firms with very low debt levels may face binding financing constraints that make it more difficult to convert convergence opportunities into performance improvements. That said, although the high-debt group exhibits a significant positive association, the potential financial risks associated with high leverage should not be overlooked.

4.4.3. Heterogeneity by Firm Age

The positive association between rural industrial integration and firm economic performance is significant in both the young firm group (β = 0.051, p < 0.05) and the mature firm group (β = 0.060, p < 0.01), but is not significant in the middle-aged group (β = 0.008, p > 0.10), exhibiting a clear “age U-shaped” characteristic. A possible interpretation is that young firms are organizationally flexible, efficient in decision-making, and highly motivated to innovate, making them better able to translate rural industrial integration opportunities into performance gains. Mature firms have accumulated considerable managerial experience and industrial chain resources, equipping them with the capacity to sustain convergence activities, which allows the positive association to emerge. Middle-aged firms, in contrast, may face stronger organizational inertia and transformation bottlenecks, which prevent the positive association between rural industrial integration and performance from materializing.

4.4.4. Heterogeneity by Fiscal Support

The positive association between rural industrial integration and firm economic performance is statistically significant across all three fiscal support groups: the low support group (β = 0.092, p < 0.01), the medium support group (β = 0.037, p < 0.05), and the high support group (β = 0.032, p < 0.10), exhibiting a clear “subsidy-dampening” characteristic. This pattern suggests that the positive association is more pronounced among firms receiving relatively less fiscal support. A possible interpretation is that limited—rather than substantial—fiscal support may more effectively stimulate firm growth momentum. Conversely, when firms receive substantial fiscal support, this positive association may become attenuated due to factors such as subsidy dependence, market crowding-out effects, or weakened incentives for autonomous innovation. However, the present data do not allow for direct testing of these mechanisms; therefore, alternative explanations cannot be ruled out. For instance, the observed pattern may also reflect sample selection effects or non-linear threshold effects in the relationship between fiscal support and firm performance. These underlying mechanisms warrant further investigation.

In summary, the group-wise regressions by firm size, debt level, firm age, and fiscal support consistently reveal significant between-group differences in the positive association between rural industrial integration and firm performance. These patterns are characterized respectively by “size-reverse” heterogeneity (i.e., the association weakens as firm size increases), “leverage-positive” heterogeneity (i.e., the association is stronger among firms with higher debt levels), “age-U-shaped” heterogeneity (i.e., the association is significant for both young and old firms but not for middle-aged firms), and “subsidy-dampening” heterogeneity (i.e., the association is significant only among firms with low fiscal support). These findings provide empirical support for Hypothesis H2.

4.5. Summary of Empirical Findings

To provide a concise overview of the empirical analyses,

Table 8. Summary of empirical Results.

Analysis Category (Corresponding Section)	Research Content	Key Findings	Statistical Result
Main effect (Section 4.2)	Relationship between rural industrial integration and firm economic performance	Positive association	Statistically significant
Robustness check (Section 4.3.1, Section 4.3.2, Section 4.3.3 and Section 4.3.4)	Sensitivity to model specification	Finding remains robust	Supported
	Sensitivity to sample period	Finding remains robust	Supported
	Sensitivity to extreme values	Finding remains robust	Supported
	Omitted Variable Bias Test	Finding remains robust	Supported
Endogeneity mitigation (Section 4.3.5)	Mitigating Concerns of Reverse Causality	Concerns of reverse causality are partially mitigated	Statistically significant
Heterogeneity analysis (Section 4.4.1, Section 4.4.2, Section 4.4.3 and Section 4.4.4)	Moderating effect of firm size	Size-reverse effect (smaller firms benefit more)	Statistically significant
	Moderating effect of debt level	Leverage-positive effect (moderately to highly leveraged firms benefit more)	Statistically significant
	Moderating effect of firm age	Age-U-shaped effect (benefits decline then rise over the firm life cycle)	Statistically significant
	Moderating effect of fiscal subsidy	Subsidy-dampening effect (excessive subsidies weaken the positive association)	Statistically significant

Note: “Statistically significant” indicates p < 0.1; “Supported” indicates that the main finding remains unchanged under alternative specifications. For the endogeneity mitigation analysis, statistical significance refers to the coefficient of the lagged independent variable remaining significant, suggesting that reverse causality does not fully explain the main finding.

summarizes the main findings, including the baseline positive association, robustness checks across alternative specifications, endogeneity mitigation using lagged variables, and four distinct heterogeneity patterns. As shown in the table, the positive association between rural industrial integration and firm economic performance remains consistent across multiple specifications, while the four forms of heterogeneity—Size-reverse, leverage-positive, age-U-shaped, and subsidy-dampening—are empirically supported.

5. Discussion

5.1. The Heterogeneous Characteristics of the Integration–Performance Association

This study reveals a significant positive association between rural industrial integration and the economic performance of agricultural leading enterprises. More importantly, this association exhibits four distinct heterogeneous characteristics—scale-reverse, leverage-positive, age-U-shaped, and subsidy-dampening—indicating that the benefits of rural industrial integration are not uniformly distributed across firms. Smaller firms reap larger gains due to organizational flexibility and greater marginal improvement space; moderately to highly indebted firms leverage external financing to fund convergence-related investments; young and mature firms translate convergence into performance gains through innovation drive and resource endowments respectively, while middle-aged firms lag behind due to path dependence; and fiscal support follows an inverted-U pattern, where moderate subsidies help but excessive support crowds out endogenous motivation. These findings challenge the conventional “one-size-fits-all” assumption underlying government policy implementation and demonstrate that firm-level heterogeneity partially determines the returns to rural industrial integration, thereby providing a reference for improving government policy approaches.

5.2. Theoretical Contributions and Practical Implications

Theoretically, this study shifts the level of analysis from macro-regional to micro-firm, bridging a critical gap between aggregate policy discourse and individual firm outcomes. It further demonstrates that even when facing the same external opportunity, the extent to which firms benefit depends critically on internal characteristics. The four heterogeneous characteristics are each supported by or complement existing evidence. The scale-reverse characteristic aligns with Chen [32], Dai et al. [33], who found that firm size negatively moderates innovation–performance and digital transformation–performance relationships. The leverage-positive characteristic is consistent with Liu and Huang [34], who documented that credit constraints significantly suppress rural enterprise performance, implying that access to debt can be performance-enhancing. The age-U-shaped finding complements Wang and Liu [35] by extending their lifecycle analysis from diversification to rural industrial integration. The subsidy-dampening characteristic echoes Xu et al. [36] and Wu et al. [37], who identified crowding-out effects and inverted-U relationships between subsidies and firm investment. Together, these four heterogeneous characteristics offer a novel framework for understanding policy response heterogeneity. Practically, policymakers should adopt differentiated strategies: lower barriers for SMEs, improve credit access for moderately indebted firms, provide targeted support for middle-aged firms to overcome transitional disadvantages, and carefully calibrate fiscal support to avoid subsidy dependence. For firm managers, the findings suggest that participation in rural industrial integration should be aligned with their firm’s specific characteristics.

5.3. Limitations and Future Research Directions

Several limitations warrant acknowledgment. The data are limited to one province in China, the panel covers only three years, The explanatory power of our model is limited. and there may be other unobserved confounding factors, such as firm-level innovation capacity and digitalization level. Consequently, the interpretation of these associative patterns should be approached with caution. Future research should extend the analysis to other regions, employ longer panel data or quasi-experimental designs to strengthen causal claims, improve model fit by collecting richer micro-level data and incorporating more refined variable measurements, and broaden the outcome space to include environmental and social performance dimensions. Despite these limitations, this study provides robust evidence that rural industrial integration is positively associated with firm economic performance in a heterogeneous manner, offering actionable insights for more targeted and effective policy interventions.

6. Conclusions

6.1. Research Conclusions

Based on panel data from 821 leading agricultural enterprises in Jiangxi Province spanning 2021–2023, this study employed a random-effects model to examine the statistical association between rural industrial integration and the economic performance of leading agricultural enterprises, as well as the heterogeneous patterns underlying this association. The baseline regression results confirm a significant positive statistical association between rural industrial integration and the economic performance of leading agricultural enterprises. This finding remains robust across a series of checks, including alternative model estimation, sample substitution, 5% Winsorization, and one-year lag analysis.

Further heterogeneity analysis reveals several distinct characteristics. “Size-reverse” heterogeneity emerges: the positive association is most pronounced among small-sized enterprises, followed by medium-sized enterprises, and is not significant for large-sized enterprises. “Leverage-positive” heterogeneity is observed: the positive association is significant for firms with moderate to high debt levels but not significant for low-debt firms. “Age-U-shaped” heterogeneity is identified: the positive association is significant for both young and mature firms but not significant for middle-aged firms. “Subsidy-dampening” heterogeneity is also found: the positive association is most pronounced in the low fiscal support group and is not significant in the medium or high support groups.

6.2. Policy Implications

These findings offer useful references for other Chinese provinces at similar development stages or for other middle-income countries promoting rural industrial integration. Accordingly, the following policy recommendations are proposed.

6.2.1. Implement Differentiated Support Strategies to Enhance Policy Precision

Priority should be given to guiding small and medium-sized leading agricultural enterprises into rural industrial integration activities, where value chain extension, multi-functionality expansion, technology penetration, and benefit linkage can jointly improve both the level of rural industrial integration and firm economic performance. Attention should also be paid to firms’ debt-to-asset ratios, guiding enterprises to employ moderate leverage in pursuing rural industrial integration: firms with excessively low leverage may be encouraged to appropriately expand their financing scale, while those with excessively high leverage require close monitoring and careful financial risk assessment to prevent unplanned or indiscriminate expansion. A life-cycle stage-based support strategy should be implemented, such that young firms may benefit primarily from financial, technical, and platform support; mature firms could be provided with specialized support for digital transformation and technology research and development to further unlock their economic performance potential; and middle-aged firms facing organizational inertia and development bottlenecks should be encouraged to pursue management, technological, and product innovation. These targeted approaches can enable more precise support for leading agricultural enterprises in advancing rural industrial integration.

6.2.2. Optimize the Allocation Efficiency of Fiscal Funds and Improve Supporting Systems and Long-Term Mechanisms

The distribution mechanism of fiscal subsidy funds should be adjusted by prioritizing competitive subsidies over universal subsidies, directing funds toward enterprises that demonstrate stronger willingness and better outcomes in rural industrial integration, thereby stimulating their motivation to pursue convergence activities. Funding delivery mechanisms should also be innovated: alongside fiscal support, social capital should be guided to participate, and joint rural industrial integration development funds should be established to maximize the leverage of social capital in supporting agricultural rural industrial integration. Fund supervision and performance evaluation need to be strengthened, with subsidies periodically reviewed and gradually withdrawn from enterprises that become dependent on such support, so as to enhance their endogenous motivation. A risk prevention and control system for rural industrial integration should be progressively improved, accompanied by enhanced risk monitoring and early warning for highly leveraged firms to prevent unplanned or excessive pursuit of rural industrial integration. Finally, the identification and promotion of exemplary rural industrial integration cases should be intensified, and through experience sharing and collaborative assistance among enterprises, the demonstration effect of these cases can be fully utilized.