The Impact of Agricultural Cooperatives on Farmers’ Agricultural Revenue: Evidence from Rural China

Abstract

Farmer’s incentive is a core issue in achieving sustainable agricultural development. In many developing countries, smallholder farming is predominant in agricultural production, potentially limiting improvements in agricultural sustainability. Promoting agricultural cooperatives is a widely adopted strategy to help resource-poor farmers obtain higher agricultural revenue. In China, these organizations have expanded rapidly since the early 21st century, reaching 2.22 million by September 2023 and providing services to nearly half of farming households. However, their effectiveness and impact on enhancing agricultural revenue remain subjects of ongoing debate. To provide more empirical evidence on this topic, this paper constructs an agricultural cooperatives database based on the national commercial registration enterprise dataset and matches it with the National Fixed Point Rural Survey (NFP). The findings reveal that the development of agricultural cooperatives in China significantly helps farmers enhance their production revenue, leading to an increase in household income. Furthermore, the paper identifies strong heterogeneity in the positive effects of cooperative development at both the village and household levels. In the mechanism analysis, it is shown that agricultural cooperatives in China facilitate increased investment in capital, intermediate inputs, and technology, optimizing the allocation of production factors in agricultural processes, thereby improving land productivity and ultimately increasing agricultural revenue.

1. Introduction

Sustainable agricultural development ensures food security and benefits for poverty reduction, serving as an important foundation for economic growth and social stability [1,2,3]. Farmer’s incentive is a core issue facing sustainable agricultural development, as it affects the allocation of land, capital, labor, technology, and other resources [4]. When the return to agricultural production is much lower than other industries, farmers tend to shift resources away from agriculture and reduce related investment, thus negatively impacting agricultural productivity and sustainability [5,6,7].

Worldwide, most farms operate on a small scale and are managed by individual families, a system known as smallholder farming [8]. Since smallholder farmers typically have limited access to financial capital, quality inputs, and modern technologies, they are at a disadvantage in market competition, which hinders their ability to improve the return on agricultural production [9,10,11,12]. The Food and Agriculture Organization of the United Nations (FAO) reports that farms of less than 1 hectare account for 72 percent of all farms, while only 1 percent of all farms are larger than 50 hectares [8]. Especially in many developing countries, smallholder farming is predominant in agricultural production due to factors such as limited arable land, large populations, and the constraints imposed by institutional arrangements [13,14]. As for China, according to the third agricultural census in 2016, the average scale of farmland is approximately 0.45 hectares per household, and over 91% of households manage farmland smaller than 0.67 hectares. Most smallholder farmers, disadvantaged in market competition, frequently encounter challenges such as weak bargaining power, asymmetric market information, and vulnerability to market volatility [15,16,17]. These barriers limit their capacity to increase agricultural revenue and, in turn, compromise agricultural sustainability.

Promoting agricultural cooperatives is a common approach to helping resource-poor producers improve farming activities and achieve higher agricultural revenue [18,19,20]. Generally, a cooperative is an autonomous association of persons united voluntarily to meet their common economic, social, and cultural needs and aspirations through a jointly owned and democratically controlled enterprise [21]. In some Western countries, the presence of agricultural cooperatives can be traced back to the 19th century, and this organization plays a critical role in sustainable agricultural development. As of now, the average market share of all agricultural cooperatives in the EU is about 40% [22]. In the U.S., approximately 44% of the agricultural input products used by farmers are supplied by cooperatives, and 80% of processed agricultural products come from cooperatives [23].

Compared with Western countries, agricultural cooperatives in China started relatively late but have grown rapidly in recent years, as they exhibit distinct structural attributes such as strong government support and a notable heterogeneity between core members and common members [24,25]. In this context, an intense debate has arisen regarding the role of agricultural cooperatives in China. Based on instances of the failure of cooperative movements in developing countries, some scholars argue that government intervention leads rural elites to seek rents by establishing “distorted” agricultural cooperatives [26,27]. In these cooperatives, most common members are excluded from decision-making processes, resulting in minimal economic benefits for smallholder farmers. However, some scholars hold a different view. Since labor, land, capital, and other factor markets are underdeveloped in rural China, government support is necessary to incentivize some farmers to become cooperative entrepreneurs. This can help to establish a cooperative equilibrium, effectively avoiding the “collective action dilemma” and actively contributing to the improvement of the income and livelihoods of smallholder farmers. Therefore, the unique attributes of agricultural cooperatives in China are a form of adaptability to the existing reality [26,28].

To empirically verify the role of agricultural cooperatives in China, several studies have been conducted. On one hand, some studies have found that the development of agricultural cooperatives contributes to increased income and reduced poverty among farmers. For instance, Ma and Abdulai (2017) used household survey data from 481 apple farmers in Gansu, Shaanxi, and Shandong provinces, finding that cooperative membership, on average, increased apple prices by 8.82%, gross income by 1.83%, farm profit by 2.18%, and return on investment by 14% [29]. Similarly, Hoken and Su (2018) used the survey data from 354 rice-producing households in Jiangsu province and found that the average net rice income of cooperative participants was 12.3% higher than that of non-participants, with this effect being more significant for smallholder farmers [30]. On the other hand, some studies have found that agricultural cooperatives have a limited economic impact. Li et al. (2016), using survey data from 701 rural households in Hainan province, concluded that while cooperatives have enhanced farmers’ bargaining power, they have also weakened farmers’ ability to access market information, and these two opposing effects tend to offset each other, resulting in little overall impact [31]. Zhu et al. (2016) analyzed the survey data of 242 agricultural cooperatives in Jiangsu province and found that the scale of most cooperatives is not large, which limits their ability to significantly improve farmers’ agricultural revenue [32].

Based on the existing empirical literature, there is no consensus on whether agricultural cooperatives can effectively increase agricultural revenue in China. This may be due to two reasons. Firstly, China is vast with significant heterogeneity in agricultural production conditions across different regions. Most studies rely on survey data from localized areas for empirical analysis, which makes it difficult to draw conclusions which are representative of the national level. Secondly, most research evaluates the economic effects of agricultural cooperatives by comparing the income of two types of farmers: those who have joined cooperatives and those who have not. However, self-selection bias exists when farmers decide to join cooperatives, and theoretically, cooperatives can generate “market yardstick effect”, producing spillover effects on non-participants. These factors could not be captured by the research design mentioned above, which may lead to divergent conclusions. To alleviate these problems, this paper constructs a nationwide agricultural cooperatives database by conducting extensive data cleaning on the national commercial registration enterprise dataset and then matches it with the National Fixed Point Rural Survey (NFP) to conduct empirical research, aimed at providing more empirical evidence on the impact of agricultural cooperatives on farmers’ agricultural revenue in China.

2. Theoretical Analysis

In rural China, after decades of agricultural collectivization, the Household Responsibility System (HRS) was adopted in 1978, allowing cultivated land to be contracted out to individual households while land ownership remained under collective control. This shift means that agricultural production has been primarily managed by households, with rural collective economic organizations providing some unified services, such as irrigation, purchasing inputs, and selling outputs [33]. This two-tier management system once led to a significant increase in farmers’ incentives and agricultural productivity [34]. However, as the market economy develops and the social environment changes, most rural collective economic organizations have tended to weaken, leading to a reduced willingness and capacity to provide services [35]. Consequently, farming households must rely on their own efforts to engage in pre-production, production, and post-production activities. Given China’s reality of having a large population and limited land resources, farming households have small and fragmented plots of land; therefore, smallholder farmers constitute the main driving force behind agricultural development in China [36].

As in many developing countries, smallholder farmers in China often face limitations both internally and externally. Internally, they lack sufficient capital, technical knowledge, and social networks. Externally, they contend with limited access to market, information, and financing opportunities. In this context, firstly, farmers have weak bargaining power and are in a disadvantaged position when dealing with upstream and downstream enterprises. This often results in “buying high and selling low” scenarios, where farmers purchase inputs at relatively high prices but sell their outputs at relatively low prices, ultimately squeezing agricultural revenue [37]. Secondly, when scattered farmers and enterprises engage in transactions, the high supervision costs can lead to unstable contractual relationships. From the farmers’ perspective, agricultural production cycles are long and characterized by high levels of uncertainty, making it challenging for enterprises to conduct thorough quality inspections throughout the entire production process. As a result, farmers may be motivated by profit-seeking behavior to sell substandard products to enterprises at the agreed-upon price. From the enterprises’ perspective, the significant imbalance of market power allows them to suddenly push prices down during the purchasing process, eroding farmers’ profits, while farmers find it difficult to resort to legal recourse. The opportunistic motives of both parties may result in the breakdown of contractual relationships, hindering the creation of effective incentives for both sides and thereby constraining agricultural revenue [38]. Thirdly, as the early provision of productive public goods by rural collective economic organizations depreciates, farmers are unable to bear the costs of repairs or new construction on their own. This situation compels them to revert to traditional cultivation methods, which restricts agricultural production efficiency and leads to a decline in agricultural revenue [39].

The development of agricultural cooperatives may help farmers overcome the bottlenecks that restrict their ability to increase agricultural revenue. First of all, by forming alliances through cooperatives, farmers can alleviate the imbalance of bargaining power in the market through both direct and indirect effects [19,40,41]. In terms of direct effect, when farmers join cooperatives and engage in unified purchasing and selling, they share the costs of market expansion, gain better access to information and social networks, improve market accessibility, and may even bypass certain intermediaries to transact directly with input producers or agricultural product consumers. This, in turn, strengthens their price negotiation power with upstream suppliers and downstream buyers. Regarding the indirect effect, also known as the “market yardstick effect”, the presence of cooperatives provides farmers with more options. Non-participants can either choose to transact individually with enterprises or join cooperatives for collective purchasing and selling. This forces enterprises to offer more favorable terms to retain market share, indirectly enhancing farmers’ bargaining power.

Subsequently, the development of agricultural cooperatives can help reduce transaction costs in agricultural market transactions and stabilize contractual relationships between farmers and enterprises [42,43,44]. For farmers, informal institutions like social relationships and personal reputation still play an important role in rural China, facilitating internal supervision among farmers after joining cooperatives. This means that if one farmer within the cooperative engages in “cutting corners”, affecting the quantity or quality of their output, the enterprise may lower the overall purchase price, thereby impacting the earnings of other farmers in the cooperative. This shared risk motivates farmers to monitor each other and discourage opportunistic behavior. For enterprises, dealing with farmer cooperatives means transacting with a collective of farmers rather than dealing with scattered and small-scale farmers. External supervision forces contribute to restraining opportunistic behavior by enterprises. In other words, if an enterprise unjustifiably lowers the purchase price or refuses to fulfill contractual obligations, the agricultural cooperative can seek local government intervention or resort to legal action, thus reducing the likelihood of the enterprise defaulting.

In addition, as farmers’ market position improves and a mutually beneficial relationship with enterprises is established, the uncertainties they face are reduced. This strengthens farmers’ financial capacity and willingness to invest in agricultural machinery, construct productive public infrastructures, and adopt advanced production technologies [45,46]. As a result, agricultural production efficiency is enhanced, leading to increased agricultural revenues. Therefore, the paper proposes the following hypothesis:

Hypothesis 1. 

The development of agricultural cooperatives can improve farmers’ agricultural revenue in China.

Based on the existing literature, agricultural revenue can be decomposed into three key components: the selling price of the product, productivity, and production costs [47]. The impact of agricultural cooperatives on farmers’ agricultural revenue may therefore arise from the following sources: (1) Price-enhancing effect. Cooperatives may enhance farmers’ bargaining power, enabling them to secure higher sale prices for products through collective selling or by bypassing intermediaries; (2) Productivity-enhancing effect. By facilitating access to shared resources, information exchange, and expanded market channels, cooperatives may help farmers obtain cheaper inputs, better production technologies, and more financial support. This encourages farmers to increase investment in production, improving efficiency and boosting agricultural productivity; (3) Cost-saving effect. Through the collective purchasing of inputs such as seeds, fertilizers, and machinery, cooperatives may help farmers reduce production costs, thereby increasing overall profitability. To verify the effects mentioned above, the paper proposes the following hypotheses, which will be tested individually in the subsequent sections.

Hypothesis 2A. 

The development of agricultural cooperatives in China helps to enhance the sale price of agricultural products, thereby improving farmers’ agricultural revenue.

Hypothesis 2B. 

The development of agricultural cooperatives in China helps to enhance agricultural productivity, thereby improving farmers’ agricultural revenue.

Hypothesis 2C. 

The development of agricultural cooperatives in China helps to save agricultural production costs, thereby improving farmers’ agricultural revenue.

3. Research Design

3.1. Data Sources

This paper primarily uses two datasets for empirical analysis. The first is an agricultural cooperatives database constructed from the national commercial registration enterprise dataset. As the establishment of agricultural cooperatives in China requires registration with the administration for industry and commerce, we construct the agricultural cooperatives database based on the existing literature [33,48]. (1) According to registration name, type, and industry classification, we extract the agricultural cooperatives sample from the national commercial registration enterprise dataset. (2) Considering that the sample may contain “shell cooperatives,” which exist only in name, we utilize enterprise penalty information, abnormal business information, and records of severe violations and dishonesty to identify cooperatives involved in situations such as “not in normal operation”, “not engaged in actual business activities”, “false operations”, or “unable to be contacted through the registered address”. We subsequently remove these cooperatives from the sample. (3) We classify the sample into three types based on enterprise shareholder information. Cooperatives with corporate shareholders are designated as “corporate-led cooperatives”. Those with shares held by rural collective economic organizations or those containing keywords such as “village committee” or “village collective” in their registered addresses are categorized as “village committee-led cooperatives”. The remaining cooperatives are classified as “farmer-led cooperatives”. (4) We use the Gaode Map geocoding API to extract the latitude and longitude of the registered addresses of the sample cooperatives. This information helps determine the province, city, county, and village in which each cooperative is located. (5) Based on the registration and exit years of each cooperative, we generate a panel database of nationwide agricultural cooperatives. This allows us to calculate the number of actively operating agricultural cooperatives in each village annually, thus laying the foundation for constructing the explanatory variable.

The second dataset this paper used is the National Fixed Point Rural Survey (NFP) conducted by the Research Center of Rural Economy (RCRE) of the Ministry of Agriculture and Rural Affairs. This survey was established in 1986 and has been in operation ever since, except in 1992 and 1994. The sample villages in this dataset are distributed across 31 provinces (municipalities and autonomous regions, hereafter referred to as provinces) in mainland China. Counties with different income levels were selected from each province, and sample villages were chosen within each county based on consistent selection criteria. Representative households from these villages were then sampled for long-term follow-up surveys. Benjamin et al. (2005) demonstrate that this dataset is of high quality and provide a detailed overview of it [49]. Chari et al. (2022) highlight its key advantages, including its panel structure and detailed household agricultural production information [50]. This dataset is then matched with the number of actively operating agricultural cooperatives at the village level to carry out the empirical analysis.

In selecting the sample period, we considered three key factors. First, the implementation of the Agricultural Cooperatives Law of China in 2007 required the establishment of agricultural cooperatives to be registered with the commercial registration authority. Since our agricultural cooperatives database is constructed based on the national commercial registration enterprise dataset, the sample period should be after 2007. Second, the NFP survey, our primary data source, underwent significant adjustments in 2009 and 2016, particularly in the collection of household-level agricultural production information. To ensure consistency in measurement, we defined the sample period as spanning from 2009 to 2015. Third, as shown in the data description (Figure 1), agricultural cooperatives in China experienced rapid development between 2009 and 2015. Moreover, this period exhibits significant regional disparities in growth trends (Figure 2), providing rich variation for the empirical analysis in our study.

3.2. Identification Strategy

This paper employs a two-way fixed effects regression model to evaluate the impact of the development of agricultural cooperatives on farmers’ agricultural revenue. Considering the potential endogeneity issues, we adopt several robustness checks to ensure the validity of the results.

$$Agrirevenu{e}_{ivt}=\alpha +\beta Cooperativ{e}_{vt}+\gamma X_{ivt}+\phi Z_{vt}+{\eta }_{iv}+{\lambda }_t+{\xi }_{ivt}$$

(1)

In this model, i represents the surveyed household, v represents the village where the household is located, and t represents the year. The dependent variable, Agrirevenue, represents the net profit per acre from agricultural production for each surveyed household. To reduce the impact of price fluctuations on purchasing power, the dependent variable is adjusted using the rural resident retail price index for each province, with constant prices set to the base year of 2009. The key explanatory variable, Cooperative, represents the number of actively operating agricultural cooperatives in the village where the household is located. To enhance estimation accuracy, we control a set of household characteristics (X) that are likely to affect the agricultural revenue, including household size, whether the household head is a cadre, age of household head, gender of household head, educational experience of household head, and health status of household head. The previous literature suggests that village characteristics, such as the scale of the agricultural industry, economic development, and government support policies, can influence the formation of local agricultural cooperatives as well as the dependent variable [26,51,52]. Therefore, we control for a set of village characteristics (Z) to mitigate endogeneity issues arising from omitted variables. Specifically, the village characteristics include total population, total land area, total labor force, the proportion of village income derived from agricultural production, whether the village is an impoverished village recognized by the local civil administration, and the total amount of agricultural subsidies the village received from the government during the year. To further mitigate endogeneity concerns, we incorporate household and year fixed effects. Household fixed effects encompass time-invariant characteristics that vary across villages, such as geographical endowments, while year fixed effects account for factors that change over time but remain constant across villages, such as fluctuations in the global agricultural market. To address potential autocorrelation issues among error terms, we cluster the standard errors at the village level.

It is worth noting the rationale behind the identification strategy. Unlike most related empirical studies, our baseline regression uses the development of agricultural cooperatives in a village as the key explanatory variable, rather than household-level cooperative membership. This choice is motivated by two main reasons. First, data limitations. Given China’s vast territory and the significant heterogeneity in agricultural production across regions, we try to use a nationwide rural household survey to analyze the average impact of agricultural cooperatives and explore regional heterogeneity. However, the NFP survey, which serves as our primary data source, does not record the information on household-level cooperative membership. Therefore, we constructed an agricultural cooperatives database and calculated the number of actively operating agricultural cooperatives in each village annually, providing a proxy for the development of agricultural cooperatives in the villages sampled by the NFP survey. Second, spillover effects. Existing studies suggest that the development of agricultural cooperatives may generate spillover effects on non-members, such as the “market yardstick effect” [19]. In our regression using Model (1), the coefficient captures the average impact of the direct effects on cooperative members and the indirect effects on non-members. To address the potential endogeneity caused by self-selection issues, we conduct a series of robustness checks, detailed in the following sections.

For detailed information regarding the names, explanations, and descriptive statistics of each variable in the baseline model, refer to

Table 1. Variable measurement and descriptive statistics.

Variables	Symbol	Measurement	Mean	Std.
I. Dependent Variable:
Agricultural revenue	Agrirevenue	ln ((agricultural income − agricultural costs)/farmland scale + 1)	6.98	1.29
II. Explanatory Variable:
The development of agricultural cooperatives	Cooperative	ln (number of actively operating agricultural cooperatives in the village + 1)	0.56	0.69
III. Control Variables
Household size	Hhsize	Total household population	3.88	1.60
Cadre household	Cadre	Whether the household head is a cadre	0.07	0.26
Household head’s age	Age	Age of the household head	55.19	11.73
Household head’s gender	Male	Whether the household head is male	0.94	0.23
Household head’s education level	Education	Educated years of the household head	6.77	2.57
Household head’s health	Healthy	Whether the household head is healthy	0.50	0.50
Total population	Villpop	ln (number of permanent residents in the village)	7.36	0.69
Total land area	Villland	ln (total land area of the village)	8.55	1.21
Total labor force	Villlabor	ln (number of labor force in the village)	6.77	0.69
Agricultural income percentage	Villind	Village’s income from agriculture/village’s total annual operating income	0.50	0.27
Impoverished village	Villeco	Whether the village is an impoverished village recognized by the local civil administration	0.12	0.32
Agricultural subsidies	Villsub	ln (total amount of agricultural subsidies received from government/100 + 1)	7.21	2.52

.

4. Empirical Results

4.1. Baseline Regression

Based on Equation (1),

Table 2. Baseline regression results.

	(1)	(2)	(3)	(4)	(5)
Dependent Variable	Agrirevenue	Agrirevenue	Agrirevenue	Agriinc	Netinc_pc
Cooperative	0.119 ***	0.121 ***	0.124 ***	0.089 ***	0.012 *
	(0.013)	(0.013)	(0.013)	(0.030)	(0.007)
Hhsize		0.004	0.004	0.080 ***	−0.168 ***
		(0.005)	(0.005)	(0.009)	(0.003)
Cadre		−0.004	−0.011	0.061	0.063 ***
		(0.021)	(0.021)	(0.046)	(0.011)
Age		0.004 *	0.005 *	0.015 **	0.005 ***
		(0.003)	(0.003)	(0.006)	(0.001)
Age_square		−0.000	−0.000	−0.000 **	−0.000 ***
		(0.000)	(0.000)	(0.000)	(0.000)
Male		−0.025	−0.018	0.576 ***	0.051 **
		(0.060)	(0.062)	(0.126)	(0.026)
Education		0.005	0.006	0.002	0.009 ***
		(0.006)	(0.006)	(0.011)	(0.002)
Healthy		0.022	0.019	0.080 **	0.045 ***
		(0.018)	(0.018)	(0.036)	(0.008)
Villpop			−0.065	0.144 *	0.080 ***
			(0.046)	(0.083)	(0.027)
Villland			0.001	0.029 **	0.006 *
			(0.009)	(0.014)	(0.003)
Villlabor			0.062 ***	0.188 ***	0.015 **
			(0.016)	(0.029)	(0.006)
Villind			0.148 ***	0.038	0.008
			(0.032)	(0.059)	(0.013)
Villeco			−0.126 ***	−0.049	−0.032 **
			(0.035)	(0.076)	(0.014)
Villsub			0.003	0.039 ***	0.001
			(0.002)	(0.005)	(0.001)
Individual fixed effects	Y	Y	Y	Y	Y
Year fixed effects	Y	Y	Y	Y	Y
N	66,537	64,462	63,107	72,063	76,125
adj. R-sq	0.5733	0.5730	0.5752	0.5636	0.6317

Note: (1) ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively; (2) Standard errors clustered at the village level are reported in parentheses.

reports the impact of the development of agricultural cooperatives on farmers’ agricultural revenue. In column (1), only individual and year fixed effects are controlled. In columns (2) to (3), household and village characteristics are added sequentially. The coefficient of the explanatory variable is significant at the 1% level, indicating that the development of agricultural cooperatives in the village significantly increases the net profit obtained by households in agricultural production, validating Hypothesis 1. Additionally, this paper replaces the dependent variable with agricultural net income and total net income per capita of each surveyed household. The coefficients of the explanatory variable remain significantly positive, indicating that the current development of agricultural cooperatives in China is not merely a tool for rural elites to seek external policy support. Rather, on average, these cooperatives can help farmers overcome the bottleneck of low agricultural production income and further increase household income, which contributes to narrowing the urban–rural income gap.

4.2. Robustness Check

4.2.1. Replace Dependent Variable

This paper uses agricultural net profit per acre to measure farmers’ agricultural revenue in the baseline regression. Considering potential measurement error issues, we replace the dependent variable with the net profit per acre from grain cultivation in the robustness check. This choice is based on the following considerations. Firstly, the production revenue from growing grain crops is typically lower than that from cash crop planting, livestock breeding, or fishery farming, making it a relatively weak area of agricultural production in terms of returns. Secondly, improving the revenue from grain cultivation is crucial for enhancing farmers’ motives to grow grain crops, which lays the foundation for ensuring food security. The construction steps of this new indicator are as follows. We first calculate the net profit per acre of each grain crop for each surveyed household, and then weigh the net profit per acre of each grain crop by their respective planted areas to obtain the household’s net profit per acre from grain cultivation. The result in column (1) of

Table 3. Robustness check.

	(1)	(2)	(3)	(4)	(5)
	Replace Dependent Variable	Replace Explanatory Variable			Excluding Impact of Other Policies
Dependent Variables	Grainrevenue	Agrirevenue			Agrirevenue
Cooperative	0.074 ***				0.108 ***
	(0.009)				(0.016)
Large and medium cooperative		0.104 ***
		(0.013)
Cooperative× average registered capital			0.020 ***
			(0.003)
Farmer-led cooperative				0.105 ***
				(0.012)
Control variables	Y	Y	Y	Y	Y
Individual fixed effects	Y	Y	Y	Y	Y
Year fixed effects	Y	Y	Y	Y	Y
N	59,667	63,107	63,107	63,107	46,973
adj. R-sq	0.4481	0.5748	0.5748	0.5749	0.6122

Note: (1) *** indicate statistical significance at the 1% levels; (2) Standard errors clustered at the village level are reported in parentheses.

indicates that the development of agricultural cooperatives significantly promotes the revenue from grain cultivation at the 1% significance level, further supporting Hypothesis 1.

4.2.2. Replace Explanatory Variable

In the baseline regression, this paper measures the development of local agricultural cooperatives from a quantitative perspective, using the number of actively operating agricultural cooperatives in the village as an indicator. This approach is informed by the strong regional characteristics of agricultural cooperatives in China, where members mainly come from within the village, with an average membership size of about 30 to 70 people [48]. However, this indicator also has its drawbacks. Firstly, existing research has found that some agricultural cooperatives are established to obtain external policy support, lacking actual business activities, thus functioning as “shell cooperatives” [53]. Although efforts were made in the preliminary data cleaning to exclude cooperatives that do not engage in genuine production activities, measurement error issues regarding the explanatory variable may still exist. Secondly, the indicator does not sufficiently account for the operational scale and different types of cooperatives.

To address these issues, this paper replaces the explanatory variable using the following methods. First of all, since the registered capital represents the amount that all initiators have committed to contribute to the cooperative, it reflects the actual willingness of cooperative members to collaborate. Following Xie et al. (2018), agricultural cooperatives are categorized into small, medium, and large based on their registered capital. The number of actively operating medium and large cooperatives in each village for each year is calculated and used as the explanatory variable in the regression model [54]. The estimation result in column (2) of Table 3 is statistically significant at 1%, and the coefficient size is close to that of the baseline regression, which verifies the robustness of Hypothesis 1. Subsequently, we replace the explanatory variable with the interaction term between the number of agricultural cooperatives in the village and the average registered capital of local cooperatives. This adjustment allows for the consideration of both quantity growth and operational scale differences when assessing the development level of local agricultural cooperatives. The result in column (3) of Table 3 is statistically significant at 1%, which aligns with and supports the baseline outcome. Additionally, given that agricultural cooperatives can be divided into three categories based on their leading entities: farmer-led cooperatives, corporate-led cooperatives, and village-committee-led cooperatives, among which the farmer-led cooperatives account for over 70% of the total and can reflect the cooperative behaviors of farmers most effectively, we use the number of farmer-led cooperatives in each village as the explanatory variable in the regression model. The result in column (4) of Table 3 remains significantly positive, indicating that the development of agricultural cooperatives improves farmers’ agricultural revenue in China.

4.2.3. Exclude the Impact of Other Policies

A number of studies have examined how the new round of land titling in China has affected farmland transfer [55], farmers’ entrepreneurship [56], and agricultural production efficiency [7]. Since the sample interval in this paper overlaps with the land titling reform, the implementation of this policy may affect both the entrepreneurial behavior of farmers, which could influence the emergence and development of agricultural cooperatives, as well as the efficiency of agricultural production, which could impact agricultural production returns, thus potentially introducing an omitted variable bias. According to Bu and Liao (2022), China’s new round of land titling began in 2009, starting with pilot projects in eight villages [56]. The scope of policy implementation was expanded to six counties in 2012 and has been gradually rolled out across the country in 2014. To reduce the potential interference of the land titling reform on this paper’s identification strategy, we use the 2009–2013 sub-sample in the baseline regression. The estimated result in column (5) of Table 3 shows that the coefficient is slightly lower than in the baseline results, but the decline is not significant, supporting the robustness of the baseline findings.

4.2.4. Change Identification Strategy

To mitigate the endogeneity problem, referring to Li and Qin (2022), this paper treats the year of establishment of the first agricultural cooperative in each sample village as the point of shock, and then uses the staggered difference-in-differences (DID) regression model to identify the causal effects of agricultural cooperatives on farmers’ agricultural revenue [57]. The specific model is set up as follows:

$$Agrirevenu{e}_{ivt}=\alpha +\beta Treatpos{t}_{vt}+\gamma X_{ivt}+\phi Z_{vt}+{\eta }_{iv}+{\lambda }_t+{\xi }_{ivt}$$

(2)

In this model, the dependent variable and control variables are consistent with the baseline model. The explanatory variable (Treatpost) is a dummy variable indicating whether the first agricultural cooperative of the sample village has been established, which takes the value of 1 if it has been established, and 0 if it has not. Regarding the sample used, this section excludes villages that had already established agricultural cooperatives in 2009 or earlier. The remaining sub-sample is then regressed according to Equation (2), where β measures the change in agricultural revenue of farmers in the experimental group compared to those in the control group after the establishment of the first agricultural cooperative in the village. As shown in column (1) of

Table 4. Change identification strategy.

	(1)	(2)	(3)	(4)	(5)
	Staggered DID	Placebo Test	Reverse Causality Test		Parallel Trend Test
Dependent Variables	Agrirevenue	Agrirevenue	Whether an Agricultural Cooperative Has Emerged		Agrirevenue
Treatpost	0.065 ***
	(0.018)
Treatpost_random constructed		−0.018
		(0.019)
One-period lag of village agricultural total income			0.003
			(0.016)
One-period lag of village agricultural income per capita				−0.014
				(0.015)
Treatpost_pre5					−0.067
					(0.068)
Treatpost_pre4					−0.041
					(0.092)
Treatpost_pre3					0.031
					(0.120)
Treatpost_pre2					0.191
					(0.152)
Treatpost_pre1					0.301
					(0.185)
Treatpost_current					0.405 *
					(0.218)
Treatpost_post1					0.447 *
					(0.253)
Treatpost_post2					0.573 **
					(0.287)
Treatpost_post3					0.675 **
					(0.321)
Treatpost_post4					0.686 *
					(0.357)
Treatpost_post5					0.560
					(0.395)
Control variables	Y	Y	N	N	Y
Individual fixed effects	Y	Y	N	N	Y
Village fixed effects	N	N	Y	Y	N
Year fixed effects	Y	Y	Y	Y	Y
N	27,042	27,042	544	538	27,042
adj. R-sq	0.5564	0.5561	0.6034	0.6073	0.5581

Note: (1) ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively; (2) Standard errors clustered at the village level are reported in parentheses.

, after the establishment of agricultural cooperatives in the village, local farmers’ production income increased by 6.7% compared to the control group (since the coefficient of column (1) in Table 4 is 0.065, then the impact of agricultural cooperatives’ emergence on farmers’ agricultural revenue can be calculated as (exp(0.065) − 1) = 6.7%.), validating Hypothesis 1.

However, there are still several concerns. Firstly, to test whether the aforementioned results are driven by some unobservable variables, we conduct a placebo test by randomly assigning the year of establishment of the first agricultural cooperative in the sample villages into Equation (2). The results are presented in column (2) of Table 4, where the estimated coefficient is not significant, indicating that the relative change in farmers’ agricultural revenue is not driven by some unobservable variables but rather is a result of the establishment of agricultural cooperatives in the local area. Secondly, considering the potential reverse causality where agricultural revenue may influence the formation of agricultural cooperatives, we use lagged village agricultural total income and income per capita as explanatory variables, respectively, with the establishment of agricultural cooperatives as the dependent variable, to examine the existence of reverse causality. The estimation results in columns (3) and (4) indicate that the village’s agricultural income does not significantly influence the establishment of agricultural cooperatives, thus somewhat ruling out interference from reverse causality in the baseline estimation results. Thirdly, an important premise for the application of the difference-in-differences model is that the experimental and control groups have parallel trends before the shock occurs. We use the sixth period before the shock (Treatpost_pre6) as the baseline for the parallel trend test. The estimation results in column (5) show that the experimental group and control group exhibited similar trends before the establishment of agricultural cooperatives. After the establishment shock, farmers’ agricultural revenue in the experimental group significantly increased compared to the control group, indicating that the assumption of pre-treatment parallel trends is satisfied.

4.2.5. Use Instrumental Variable Method

In addition to the staggered DID approach, we also apply the instrumental variable method to address the endogeneity problem. As is well-known, valid instrumental variables must satisfy the requirements of relevance, exogeneity, and exclusivity. Regarding relevance, since the formation of agricultural cooperatives is a kind of collective action, representing a common expression of cooperative decision-making among farmers, such cooperative decisions may be influenced by local levels of social trust, farmers’ risk attitudes, and their value orientations. On one hand, the existing research has shown that the severity of damage during the Great Famine in China had long-term effects on people’s behaviors, including a decline in social trust [58], a tendency toward more conservative values [59], and an increased preference for household-based production among farmers [60]. On the other hand, many studies have highlighted the role of informal institutions in shaping economic decisions [61,62]. Confucian culture, in particular, is one of the most influential cultural symbols in Chinese society, deeply affecting people’s risk attitudes and value orientations [63]. Therefore, we use the population reduction rate to measure the severity of damage in various counties, and the number of ancient academies to indicate the strength of Confucian culture across counties. We then construct instrumental variables by interacting these two cross-sectional indicators with the annual total number of agricultural cooperatives nationwide [64]. Regarding exogeneity, both cross-sectional variables originate from historical periods, making them relatively exogenous to the current economic system.

The first-stage regression results, as shown in columns (1) to (3) of

Table 5. Use instrumental variable method.

First Stage	(1)	(2)	(3)
	Cooperative	Cooperative	Cooperative
The severity of damage during the Great Famine × Total number of agricultural cooperatives nationwide	−0.310 ***		−0.244 ***
	(0.023)		(0.023)
Confucian culture × Total number of agricultural cooperatives nationwide		−0.038 ***	−0.027 ***
		(0.004)	(0.004)
KP F statistic	182.01	92.43	100.53
Second stage	(4)	(5)	(6)
	Agrirevenue	Agrirevenue	Agrirevenue
Cooperative	1.080 ***	0.863 ***	0.989 ***
	(0.156)	(0.162)	(0.131)
Hansen J p-value	—	—	0.2394
Control variables	Y	Y	Y
Individual fixed effects	Y	Y	Y
Year fixed effects	Y	Y	Y
N	48,501	48,501	48,501

Note: (1) *** indicate statistical significance at the 1% levels; (2) Standard errors clustered at the village level are reported in parentheses.

, confirm a strong correlation between these instrumental variables and the explanatory variable, eliminating concerns about weak instruments (F-statistics > 10). The second-stage regression results, presented in columns (4) to (6), demonstrate that the development of agricultural cooperatives has a significantly positive impact on agricultural production revenue, with statistical significance at the 1% level. Furthermore, the over-identification test finds no evidence of endogeneity in the instrumental variables (the p-value of the Hansen J statistic exceeds 0.1, indicating that the null hypothesis of exogeneity cannot be rejected), supporting the validity of the selected instrumental variables.

Moreover, regarding exclusivity, it is possible that the selected instrumental variables may not only influence the development of agricultural cooperatives but also affect farmers’ labor allocation, farmland transfer, cropping structure, and technology adoption. These indirect effects could bias the estimation results by impacting the dependent variable through alternative channels, violating the exclusivity requirement. To address this concern, we conducted a series of tests. Specifically, we measured local labor allocation using the proportion of agricultural laborers and the proportion of migrant laborers in the village. The level of farmland transfer development was assessed using the proportion of households transferring out farmland and the proportion of farmland area transferred out. The cropping structure was evaluated based on the proportion of grain crop planting area and cash crop planting area. The adoption of new technologies was captured using the proportion of households with internet access, and local agricultural infrastructure was measured by the distance from the village to the nearest highway. Each of these indicators was used as a dependent variable in separate regressions, with the two instrumental variables serving as the key explanatory variables. The results, presented in

Table 6. Exclusivity test of the instrumental variables.

	(1)	(2)	(3)	(4)
Dependent Variables	The Proportion of Agricultural Laborers in the Village	The Proportion of Migrant Laborers in the Village	The Proportion of Households Transferring Out Farmland	The Proportion of Farmland Area Transferred Out
The severity of damage during the Great Famine × Total number of agricultural cooperatives nationwide	−0.020	0.002	0.004	−0.004
	(0.036)	(0.040)	(0.059)	(0.057)
Confucian culture × Total number of agricultural cooperatives nationwide	0.001	0.003	−0.003	−0.002
	(0.005)	(0.005)	(0.006)	(0.005)
Control variables	Y	Y	Y	Y
Village fixed effects	Y	Y	Y	Y
Year fixed effects	Y	Y	Y	Y
N	1165	1165	1165	1152
adj. R-sq	0.7251	0.6671	0.2638	0.2655

Note: Standard errors clustered at the village level are reported in parentheses.

and

Table 7. Exclusivity test of the instrumental variables.

	(1)	(2)	(3)	(4)
Dependent Variables	The Proportion of Grain Crop Planting Area	The Proportion of Cash Crop Planting Area	The Proportion of Households with Internet Access	The Distance from the Village to the Nearest Highway
The severity of damage during the Great Famine × Total number of agricultural cooperatives nationwide	0.001	−0.007	−0.005	0.049
	(0.054)	(0.040)	(0.048)	(0.111)
Confucian culture × Total number of agricultural cooperatives nationwide	−0.013	0.000	0.006	−0.002
	(0.008)	(0.006)	(0.006)	(0.014)
Control variables	Y	Y	Y	Y
Village fixed effects	Y	Y	Y	Y
Year fixed effects	Y	Y	Y	Y
N	1157	1157	1165	1010
adj. R-sq	0.6249	0.5762	0.3753	0.6864

Note: Standard errors clustered at the village level are reported in parentheses.

, show that the estimated coefficients of the two instrumental variables through alternative channels are generally small and statistically insignificant. This suggests that the instrumental variables are unlikely to influence agricultural profitability through other pathways, providing additional evidence to support their validity. Overall, the two-stage instrumental variable regression results provide robust support to Hypothesis 1.

4.3. Heterogeneity Analysis

This paper evaluates the average effect of the development of agricultural cooperatives on farmers’ agricultural revenue in the baseline regression, and the empirical results remain robust after a series of checks, alternative identification strategies, and two-stage regressions using instrumental variables. Subsequently, we will focus on exploring the heterogeneity in the role of agricultural cooperatives. Specifically, we will examine whether the impact of agricultural cooperatives differs across various rural regions and among different types of farming households.

4.3.1. Village Heterogeneity

The primary function of agricultural cooperatives is to facilitate the collective procurement of agricultural inputs and the unified sale of agricultural products, helping farmers increase their market power and reduce transaction costs. When farmers individually enter the market, they often struggle due to their small-scale operations. This limits their ability to cover the costs of establishing procurement and sales channels, forcing them to rely on local markets and wait for bulk buyers to collect their produce. As a result, individual farmers are weak market participants, typically acting as price-takers with minimal bargaining power when dealing with upstream suppliers and downstream buyers. Moreover, in the face of unexpected events such as sudden climate changes or transportation blockages, farmers may encounter difficulties in selling their products or face higher costs for inputs, exacerbating market risks. These challenges not only reduce agricultural revenues but also create significant instability in household income, which discourages agricultural production. However, when local agricultural cooperatives are established, farmers can collectively enter the market. By expanding their scale of purchases and sales and sharing the costs of developing sales channels, cooperatives improve market accessibility, allowing farmers to negotiate better prices and participate in a wider market.

Following this logic, in villages where agricultural operations are larger, the potential benefits of collective procurement and sales are greater, leading to more significant economic effects from the development of agricultural cooperatives. Therefore, we divide the sample villages into two groups based on the median percentage of agricultural income in total village income in 2009. A dummy variable is created to interact with the explanatory variable in the regression model. The results in column (1) of

Table 8. Village heterogeneity analysis.

	(1)	(2)	(3)	(4)	(5)
Dependent Variable	Agrirevenue	Agrirevenue	Agrirevenue	Agrirevenue	Agrirevenue
Cooperative	0.050 ***	0.073 ***	0.095 ***	0.091 ***	0.407 ***
	(0.018)	(0.023)	(0.014)	(0.014)	(0.037)
Cooperative× Whether the village has a relatively high proportion of agricultural income	0.122 ***
	(0.021)
Cooperative× Whether the village is located in the Central–Western Region		0.065 ***
		(0.025)
Cooperative× Whether the village is an impoverished village			0.179 ***
			(0.030)
Cooperative× Whether the village is located in a mountainous area				0.128 ***
				(0.024)
Cooperative× Whether the village has a relatively high internet usage					−0.326 ***
					(0.038)
Control variables	Y	Y	Y	Y	Y
Individual fixed effects	Y	Y	Y	Y	Y
Year fixed effects	Y	Y	Y	Y	Y
N	63,107	66,537	62,957	63,107	63,107
adj. R-sq	0.5756	0.5734	0.5756	0.5756	0.5765

Note: (1) *** indicate statistical significance at the 1% levels; (2) Standard errors clustered at the village level are reported in parentheses.

show that in villages where agriculture is the primary activity, the development of agricultural cooperatives significantly boosts farmers’ production revenue, which aligns with expectations.

Furthermore, in regions with less favorable geographic conditions and underdeveloped economies, market accessibility is constrained by factors such as poor transportation infrastructure and immature markets, resulting in higher transaction costs in the agricultural market. In these areas, agricultural cooperatives, as a means for farmers to access the market, should play a larger role. To analyze this, we classified villages based on their provincial location into Eastern and Central–Western regions and created a binary variable indicating whether the village is located in the Central–Western region. This variable was then interacted with the core explanatory variable and included in the regression model. The results in column (2) of Table 8 show that, compared to the Eastern regions, the development of agricultural cooperatives has a more significant impact in the Central–Western regions, where it is more effective in helping farmers increase agricultural production revenue. Moreover, we introduce village characteristics, such as whether the village is officially recognized as impoverished or located in a mountainous area, into the baseline regression by creating interaction terms with the explanatory variable, respectively. The results in columns (3) and (4) indicate that the development of agricultural cooperatives has a positive effect on agricultural revenue, and this effect is more pronounced in impoverished and mountainous villages. This suggests that agricultural cooperatives not only increase farmers’ production revenue on average, but also enable villages with disadvantaged geographic and economic conditions to access markets more effectively, thereby promoting balanced development across different rural regions.

Additionally, from a market accessibility perspective, in villages with better-developed internet infrastructure, farmers can use online channels to broaden their market reach and reduce the costs of developing procurement and sales channels. In such cases, the role of agricultural cooperatives may be diminished due to the proliferation of internet infrastructure. Therefore, we divide villages into two groups based on the median proportion of internet users in each village in 2009, creating a dummy variable that interacts with the explanatory variable in the regression model. The results in column (5) of Table 8 reveal that in villages with lower internet usage rates, the positive effect of agricultural cooperative development on agricultural revenue is stronger. This further supports the idea that agricultural cooperatives help farmers organize collectively, overcome market accessibility constraints, and participate in the market more effectively.

4.3.2. Household Heterogeneity

In the debate over the role of agricultural cooperatives in China, some studies argue that these cooperatives primarily serve as a tool for rural elites to engage in rent-seeking and maximize their own interests. Others, however, contend that despite their distinct features, agricultural cooperatives can still function as platforms for cooperation and shared benefits among smallholder farmers. To explore this further, we conduct household heterogeneity analysis as follows. Firstly, we measure the household’s human capital, technological levels, and economic conditions using the household head’s education level, the household’s total factor productivity, and per capita net income in 2009, respectively. The median value of these indicators within each village is used to divide the sample households into two groups: rural elites and smallholder farmers. Columns (1) to (3) of

Table 9. Household heterogeneity analysis.

	(1)	(2)	(3)	(4)
Dependent Variable	Agrirevenue	Agrirevenue	Agrirevenue	Agrirevenue
Cooperative	0.082 ***	0.081 ***	0.079 ***	0.126 ***
	(0.022)	(0.016)	(0.017)	(0.013)
Cooperative× Whether the household head’s education level is relatively low	0.060 **
	(0.024)
Cooperative× Whether the household’s TFP is relatively low		0.091 ***
		(0.021)
Cooperative× Whether the household’s net income per capita is relatively low			0.090 ***
			(0.021)
Cooperative× Whether the household head is a cadre				−0.030
				(0.026)
Control variables	Y	Y	Y	Y
Individual fixed effects	Y	Y	Y	Y
Year fixed effects	Y	Y	Y	Y
N	63,107	63,107	63,107	63,107
adj. R-sq	0.5753	0.5754	0.5754	0.5752

Note: (1) *** and ** indicate statistical significance at the 1% and 5% levels, respectively; (2) Standard errors clustered at the village level are reported in parentheses.

reveal significant heterogeneity in the impact of agricultural cooperative development on agricultural production revenue across different households. Specifically, the development of agricultural cooperatives has a stronger positive effect on households with relatively lower human capital, technological levels, and economic conditions, suggesting that, despite their unique characteristics compared to those in Western countries, agricultural cooperatives in China can help smallholder farmers achieve higher production returns.

Secondly, we introduce a dummy variable indicating whether a household is led by a national or village cadre and incorporate its interaction with the explanatory variable into the regression. The results in column (4) of Table 9 show that cadre households benefit similarly to other households, with no evidence of them extracting more production gains from the development of local cooperatives. This indicates that agricultural cooperatives in China are not solely driven by rural elites’ profit-seeking motives but rather involve a profit-sharing mechanism between rural elites and smallholder farmers. Therefore, considering China’s unique resource endowments, developmental stage, and institutional, economic, and cultural contexts, the development of agricultural cooperatives may exhibit some “localized” characteristics.

4.4. Mechanism Analysis

4.4.1. Mechanism Test

Based on the theoretical analysis, the impact of agricultural cooperatives on farmers’ agricultural revenue may result from the price-enhancing effect, productivity-enhancing effect, or cost-saving effect. To determine which mechanism contributes the most to improving agricultural revenue, we will test each individually. First, we decompose the dependent variable into three components: the unit sales price of agricultural products, the output per acre, and the production costs per acre. These are measured using weighted sales prices, land productivity, and production costs of various crops cultivated by each household, with the weights determined by the proportion of each crop’s sown area. The results in

Table 10. Mechanism test.

	(1)	(2)	(3)
Dependent Variable	Saleprice	Productivity	Productcost
Cooperative	0.002	0.061 ***	0.019 ***
	(0.009)	(0.007)	(0.006)
Control variables	Y	Y	Y
Individual fixed effects	Y	Y	Y
Year fixed effects	Y	Y	Y
N	60,534	60,738	60,738
adj. R-sq	0.1821	0.4354	0.6152

Note: (1) *** indicate statistical significance at the 1% levels; (2) Standard errors clustered at the village level are reported in parentheses.

show that the development of agricultural cooperatives leads to an increase in output per acre, indicating that land productivity improvement is the primary mechanism through which cooperatives boost farmers’ production revenue. In other words, agricultural cooperatives facilitate the optimal allocation of production factors, resulting in higher yields per unit of land through the synergy of these factors. This supports Hypothesis 2B (productivity-enhancing effect), while there is no significant evidence to support Hypotheses 2A (price-enhancing effect) or 2C (cost-saving effect).

4.4.2. Further Discussion

Through empirical analysis, we determined that agricultural cooperatives in China primarily enhance agricultural production revenue by promoting land productivity. Existing research indicates that there is a significant degree of factor misallocation in the agricultural production sector in China, which constrains the increase in agricultural productivity [13]. To further explore how the development of agricultural cooperatives drives farmers to optimize the allocation of production factors and subsequently boost productivity, we will discuss five key factors: land, labor, capital, intermediate inputs, and technology.

The first possibility is that the development of cooperatives promotes land transfer and consolidations, which leads to scale benefits in production and then a noticeable increase in output per unit of land. Thus, we use the area of farmland operated by sample households as the dependent variable in the regression model. The results in column (1) of

Table 11. Further discussion.

	(1)	(2)	(3)	(4)	(5)
	Land	Labor
Dependent Variable	Farmland	Migrant	Agriculture	Agri-Relatedind	Otherind
Cooperative	−0.010	−0.606 **	0.322	0.406 ***	−0.121
	(0.006)	(0.266)	(0.337)	(0.141)	(0.311)
Control variables	Y	Y	Y	Y	Y
Individual fixed effects	Y	Y	Y	Y	Y
Year fixed effects	Y	Y	Y	Y	Y
N	66,221	75,767	75,767	75,767	75,767
adj. R-sq	0.9082	0.6242	0.6646	0.6047	0.5482

Note: (1) *** and ** indicate statistical significance at the 1% and 5% levels, respectively; (2) Standard errors clustered at the village level are reported in parentheses.

indicate that the development of agricultural cooperatives does not significantly affect the scale of farmland operated by households.

The second possibility is that farmers increase output through greater labor investment on the same scale of cultivated land. To verify this, we first categorize the types of labor used by households. Specifically, laborers who have worked outside for more than three months in a year are defined as migrant laborers, while those who have not migrated are classified based on their primary industry into categories such as farming at home, working in agricultural-related industries, and engaging in other industries. On this basis, we calculate the proportion of each of these four types of labor for the surveyed household, which are then used as dependent variables in the regression, respectively. The results in columns (2) to (5) of Table 11 demonstrate that the development of agricultural cooperatives can attract more rural labor to stay and work locally, while these laborers do not engage in agricultural production but rather in agricultural-related industries. These findings suggest that the enhancement of land productivity achieved by agricultural cooperatives does not stem from an increase in agricultural labor input.

The third possibility is an increase in capital investment, where farmers enhance their capital input on contracted farmland by utilizing more agricultural machinery and advanced equipment. This leads to improvements in the mechanization and digitalization of production processes such as plowing, sowing, and harvesting, ultimately increasing agricultural production efficiency. To verify this potential mechanism, we measure capital investment by the number of agricultural machines owned by households, using this as the dependent variable in the regression. The results in column (1) of

Table 12. Further discussion.

	(1)	(2)	(3)	(4)	(5)
	Capital	Intermediate Inputs			Technology
Dependent Variable	Machine	Inputamount	Fertilizerprice	Pesticideprice	TFP
Cooperative	0.041 ***	0.020 ***	−0.028 ***	−0.007	0.069 ***
	(0.011)	(0.006)	(0.003)	(0.007)	(0.009)
Control variables	Y	Y	Y	Y	Y
Individual fixed effects	Y	Y	Y	Y	Y
Year fixed effects	Y	Y	Y	Y	Y
N	76,252	60,738	65,077	54,189	60,738
adj. R-sq	0.8124	0.5869	0.6237	0.5769	0.3878

Note: (1) *** indicate statistical significance at the 1% levels; (2) Standard errors clustered at the village level are reported in parentheses.

show that the development of agricultural cooperatives significantly promotes farmers’ investment in agricultural machinery, which may subsequently contribute to increased output per acre.

The fourth possibility involves an increase in intermediate inputs, such as fertilizers and pesticides. We calculate the weighted average of the expenditures on intermediate inputs per acre and use this as the dependent variable. The results in column (2) of Table 12 indicate that the development of agricultural cooperatives has led to an increase in the amount spent on intermediate inputs per unit of land. Given that the rise in expenditure could stem from either an increase in consumption or higher purchase prices, we further calculate the purchase prices of fertilizers and pesticides for each household, using these as dependent variables in the regression. The results in columns (3) and (4) reflect that the development of agricultural cooperatives enables farmers to purchase intermediate inputs at lower prices. This not only indicates an improvement in farmers’ market position but also suggests that the increased use of intermediate inputs is one of the channels through which agricultural cooperatives enhance productivity.

Finally, we examine the role of technological factors by using the agricultural total factor productivity of each household as the dependent variable in the regression analysis. The results in column (5) of Table 12 indicate that the development of agricultural cooperatives fosters the application of technology in the agricultural production process. Overall, agricultural cooperatives enhance farmers’ capital investment, intermediate inputs, and technological factors, thereby optimizing the efficiency of factor allocation and increasing land productivity, which ultimately leads to higher agricultural revenue.

5. Conclusions

To examine the impact of the development of agricultural cooperatives on farmers’ agricultural revenue in China, this paper constructed an agricultural cooperatives database on the basis of the national commercial registration enterprise dataset, and then matched it with the National Fixed Point Rural Survey (NFP). After controlling for various household and village characteristics, the empirical analysis reveals that the development of agricultural cooperatives significantly helps farmers enhance their agricultural production revenue and further increases household income. These results remain robust after considering potential measurement errors, excluding other policy impacts, switching to an alternative identification strategy, and using instrumental variable methods to address endogeneity issues.

Additionally, this paper finds strong heterogeneity in the positive effects of agricultural cooperative development on agricultural revenue at both the village and household levels. First, in villages where agriculture is the main economic activity, with poor geographical conditions, relatively backward economies, or inadequate network infrastructure, the development of agricultural cooperatives helps farmers access markets, overcoming constraints due to limited market accessibility, and leads to a more significant increase in agricultural production revenue. Second, the development of agricultural cooperatives has a more substantial positive effect on households with relatively lower human capital, technological levels, and economic conditions in the village. Moreover, there is no significant difference in the benefits received by cadre households and those of regular farmers, indicating that while agricultural cooperatives in China exhibit some unique structural characteristics, they do not merely serve the profit motives of rural elites but also embody a profit-sharing mechanism between rural elites and smallholder farmers. Furthermore, in the mechanism analysis, this paper decomposes agricultural revenue into three components and examines each one individually. The results show that agricultural cooperatives facilitate increased investment in capital, intermediate inputs, and technology, optimizing the allocation of production factors in the agricultural process, thereby improving land productivity and ultimately increasing farmers’ production revenue.

Our findings indicate that promoting the development of agricultural cooperatives offers a promising and effective strategy for boosting agricultural revenue and improving household incomes in rural China. First, while digital network technologies are advancing rapidly, many rural areas in China, particularly relatively impoverished villages, still lag in market development. In these regions, fostering collective organizations such as agricultural cooperatives can strengthen farmers’ market positions, reduce transaction costs between farmers and enterprises, and serve as a crucial mechanism to enhance production revenue, stimulate agricultural growth, and support rural revitalization. Moreover, disparities among farmers in some regions are pronounced due to differences in age, gender, education, social resources, and production technology. Under these circumstances, targeted policy guidance to encourage large-scale farmers to establish benefit-sharing mechanisms with smallholder farmers can help narrow income disparities within rural communities. Second, compared to developed countries, agricultural cooperatives in China are relatively new and still in the early stages of development. Given the less mature trade networks and financial support systems in rural China, local governments can play a vital role in helping improve agricultural cooperatives’ operational efficiency and market competitiveness. This can be achieved through initiatives such as improving transportation networks, building storage and preservation facilities, and establishing agricultural wholesale markets. As the impact of agricultural cooperatives strengthens, farmers’ incentive for agricultural production increases, thereby contributing to sustainable agricultural development.