Research on the Impact of Agricultural Production Outsourcing on Farmers’ Fertilizer Application Intensity: An Inverse U-Shaped Relationship

Abstract

Agricultural production outsourcing services encourage a shift in the way crops are grown in developing countries and make it easier for small farmers to join the social division of labor in agriculture. This makes production more efficient and has a big effect on the inputs used in agriculture, especially fertilizer. This paper empirically tests the impact of production outsourcing on farmers’ fertilizer intensity using the instrumental variables method with non-planar panel data from the 2020–2022 China Land Economy Survey (CLES) of farm plots. The results showed that there was a significant inverted U-shaped relationship between the degree of agricultural production outsourcing and the intensity of fertilization on farmers’ plots. Mechanistic analysis shows that agricultural production outsourcing affects the fertilizer intensity by changing the labor allocation of farmers. Especially as the degree of agricultural production outsourcing increases, the intensity of farm labor inputs by farmers gradually decreases, and the impact of fertilizer intensity on the plots showed a tendency to be promoted first and then suppressed. The moderating effect showed that plot size was a major moderating factor. This means that the bigger the plot, the flatter the inverted U-shaped curve became, and the same level of outsourcing could lead to less fertilizer application. This happened by moving the inflection point of the inverted U-shaped curve to the left, which stopped the fertilizer application at a lower level of outsourcing. Heterogeneity analysis showed that participation in technology-intensive production outsourcing was beneficial in terms of reducing fertilizer intensity, and that an increased degree of agriculture production outsourcing was beneficial for farmers with large plot sizes and younger heads of household.

1. Introduction

Fertilizers are a critical factor in food production and play an important role in ensuring food security, particularly in China and some developing countries. In the face of China’s huge population base, relatively limited arable land surface per capita, and strict constraints on food security, chemical fertilizers have played a pivotal role in guaranteeing China’s increased grain production, with a contribution rate of more than 40 percent (http://www.kjs.moa.gov.cn/gzdt/202101/t20210119_6360102.htm, accessed on 25 September 2024), and have provided important support to keeping grain output stable at more than 1.3 trillion pounds. However, excessive use of chemical fertilizers has caused soil nodulation, acidification, salinization, and shallow plough layers, which not only decline the quality of arable land and result in serious soil pollution [1], but may also pose a threat to food security and the green development of agriculture. It is worth noting that numerous rural laborers who go out to work are driven by economic interests in the context of China’s rapid economic development [2,3,4,5]. Between 2013 and 2023, the overall number of migrant workers rose from 268.94 million to 297.53 million (https://www.stats.gov.cn/xxgk/sjfb/zxfb2020/202404/t20240430_1948783.html, accessed on 25 September 2024). Consequently, China has a persistent manpower shortage in agriculture [6], so women and the elderly are now the primary agricultural producers [7,8]. As a result, farmers have increased their agricultural output and reduced the possibility of yield declines by making excessive investments in production inputs like fertilizers [5,8]. Therefore, knowing how to further promote fertilizer reduction and efficiency is key to the green development of high-quality agriculture [9]. China’s Ministry of Agriculture and Rural Development has released a number of policies aimed at promoting fertilizer efficiency and reduction, such as the Action Programme for Fertilizer Reduction by 2025 and the Action Programme for Zero Growth in Fertilizer Use by 2020. China’s efficiency in applying fertilizer has greatly increased recently, and the country has successfully managed its fertilizer intensity. The rate of fertilizer utilization for the three main grain crops—rice, wheat, and maize—was 41.3 percent in 2022, up 1.1 percentage points from 2020 (https://www.moa.gov.cn/ztzl/zyncgzh2022/pd2022/202212/t20221226_6417613.htm?eqid=ec6e4e880001b53e000000066458a0e5, accessed on 25 September 2024). However, this rate is less than the 50–65 percent nitrogen fertilizer utilization rate for grain crops in developed nations in Europe and the US (https://www.gov.cn/xinwen/2015-12/02/content_5019114.htm, accessed on 25 September 2024). According to FAO statistics, China’s discounted agricultural fertilizer use per unit of sown area in 2021 was 319.11 kg/hm², a value that is 2.58 times the world average and that far exceeds the internationally recognized safe upper limit. This indicates that although some progress has been made in fertilizer reduction and efficiency in China, the problems of high intensity and low efficiency of fertilizer inputs are still prominent.

Agricultural production outsourcing entails the delegation of certain or all stages of agricultural production to external service providers, or the employment of individuals to carry out specific production tasks [10]. Agricultural production outsourcing, as a new means of agricultural production, can optimize resource allocation, reduce agricultural production costs, and increase agricultural productivity [6]. In China, the agricultural production outsourcing business is currently growing quickly. Family farms are managing their farms less independently and agricultural socialization services, which mostly include large agricultural equipment, are becoming more common [2]. For example, farmers in China can outsource seedling production, ploughing, sowing, fertilizing, dosing, irrigation, and straw return to the fields to agricultural production organizations. Consequently, agricultural production outsourcing services promote a change in production methods in developing countries’ agriculture and facilitate the integration of smallholders into the social division of labor in agriculture, thereby increasing production efficiency with a significant impact on agricultural factor inputs, especially fertilizer inputs. Land fragmentation is a basic characteristic of agricultural production in China [2]. Although land transfers can promote large-scale operations and help save fertilizer and other factor inputs [11], the management pattern of China’s dispersed farmland has not entirely improved despite the transfer of farmland management rights ongoing for many years [12]. Therefore, outsourcing services in the agricultural production chain provide a new path for fertilizer reduction and efficiency. Therefore, here we discuss how outsourcing of the agricultural production chain affects farmers’ fertilizer application behavior to explain the excessive fertilizer application intensity of Chinese farmers.

An analysis of the actual production practices of farmers found that outsourcing of agricultural production processes may have contributed to the intensity of fertilizer application by farmers, and helped to reduce the intensity. On the one hand, with their cutting-edge agricultural machinery and sophisticated agricultural production techniques, agricultural production outsourcing service providers can offer their services [12,13] and alleviate the technical constraints of farmers [14,15]. Consequently, the adoption of outsourcing for agricultural production processes by producers can resolve the issue of excessive and inefficient use of agrochemical inputs by utilizing a professional division of labor and scientific production services [3,16]. On the other hand, farmers’ shift from traditional farming to modern market-oriented agricultural production methods has eased the labor constraints of farm households creating a “crowding out effect” on labor [17]. This has made non-farm employment and part-time work commo and has reduced the intensity of farm labor inputs by farm households, which may lead to the risk of reduced yields and incomes. Due to a lack of labor input, farmers may seek to apply fertilizer more intensely in order to mitigate this risk [13,18,19,20]. Based on this analysis, it is suggested that there might not be a direct correlation between outsourcing of the agricultural production chain and the amount of labor dedicated by farmers who apply fertilizer. This is due to the interplay between labor and fertilizer, as well as the outsourcing of the production process and labor. The “crowding-out effect” of outsourcing on farm labor in the production process sharpens when farmers outsource their agricultural output. The degree of production specialization rises concurrently as well. This can lead to an inverted U-shaped relationship, whereby outsourcing first encourages farmers to use fertilizer, then causes its use to be hampered. Plot size is also an important factor influencing fertilizer use by farmers. The larger the plot size, the more conducive it is to promoting the adoption of agricultural production outsourcing services by farmers, increasing the degree of vertical division of labor in agriculture and giving comparative advantages, which is conducive to reducing fertilizer intensity. Therefore, the effect of agricultural production outsourcing on the fertilizer application intensity of farmers may show an inverted U-shape relationship with an increasing degree of outsourcing, which may be moderated by plot size. Thus, there may be an inverted U-shaped relationship between the introduction of outsourcing services in the agricultural production chain, which helps to reduce the fertilizer application intensity by farmers, and the possibility of over-fertilization by farmers, in which plot size may play a moderating role.

The existing literature on the effect of outsourcing the production chain on the intensity of fertilizer application by farmers is still controversial. A part of the study suggests that agricultural production outsourcing has a dampening effect on the intensity of fertilizer application [12]. Research indicates that farmers’ involvement in outsourcing can decrease the intensity of fertilizer application by enhancing fertilizer application efficiency [21], promoting the adoption of environmentally friendly technologies, adjusting factor allocation [12,19], implementing large-scale fertilizer application, and increasing crop yields [21]. Nevertheless, certain studies have reached the opposite conclusion, positing that the outsourcing of agricultural production increases fertilizer intensity. Such studies suggest that outsourced service organizations seek to maximise profits [22] or collude with agro-dealers [14,23,24], leading farmers to buy more fertilizer. In addition, it has been argued that participation in agricultural production outsourcing has not reduced fertilizer application intensity. This study concluded that agricultural production outsourcing is not a necessary condition for achieving fertilizer reduction and does not necessarily bring about a fertilizer reduction effect [22], and also found that although the purchase of fertilizer outsourcing by part-time farmers was effective in mitigating the decline in the amount of fertilizer made by farmers, it did not bring about a quantitative effect of fertilizer reduction [13]. Studies have found that whether or not farmers are involved in outsourcing, the level of horizontal participation in outsourcing versus vertical participation in outsourcing promotes an increase in the amount of annual fertilizer inputs [25], which in turn increases the fertilizer application intensity. An additional investigation confirmed that outsourcing can enhance the effectiveness of pesticide utilization by augmenting the frequency of agrochemical inputs. However, it is ineffective in addressing the overabundance of agrochemical inputs and may even escalate their intensity [14]. Further theoretical research is needed to examine the relationship between agricultural production outsourcing and the intensity of fertilizer application by farmers, as indicated by the aforementioned study.

The reason for the conflicts in the existing studies may lie in the failure to analyze the intensity of fertilizer application from a unified framework of outsourcing the agricultural production chain and the allocation of farm labor. Both non-farm employment and part-time employment involve the allocation of labor to farm households for farming. However, rational farmers generally choose labor-saving production chain outsourcing [17] to save household labor and reallocate labor resources to maximize household returns. Due to the substitution relationship between production chain outsourcing and farm household labor, it is important to analyze production chain outsourcing and farm household labor allocation in the same analytical framework. Fertilizers must be applied several times during the crop growth cycle, at a different rate each time [12]. Therefore, in order to avoid the risk of reduced yields and harvests, farmers generally choose to reduce the number of fertilizer applications and increase the amount of fertilizer applied per application, which leads to an increase in the intensity of fertilizer application [26,27]. In addition, as the degree of outsourcing of agricultural production processes deepens, farmers may choose to purchase outsourced services for fertilizer application in addition to outsourcing labor-saving production processes. This point involves the transformation of the main body of fertilizer application. In general, when the size of the land is small, it is more cost-effective for farmers to be able to choose family labor for fertilizer application, whereas when the size of the plot gauge is large, farmers are constrained by their labor force and may outsource fertilizer application to social service organizations [12]. In reality, for Chinese farmers, when the degree of outsourcing of the production process is high, farmers will choose services such as fertilizer outsourcing, and because agricultural fertilizer outsourcing organizations have advanced machinery and professional personnel, they can uniformly apply fertilizers through mechanization to improve uniformity and reduce the intensity of chemical fertilizer application. This suggests that as the degree of outsourcing of the production chain deepens, its impact on the intensity of fertilizer application by farmers is first promoted and then suppressed with the increase in the degree of outsourcing, which may show up as an inverted U-shaped relationship. The above analysis shows that outsourcing of the agricultural production chain does not have a simple linear relationship with the intensity of fertilizer application by farmers. This effect also involves a shift in the identiy of the subject of fertilizer application and is also influenced by the moderating effect of plot size. Furthermore, there are differences in the results of studies based on the farmer level and the plot level [28], as plot-level data allow for a more precise study by avoiding the bias caused by the aggregation of data at the farmer level. The study mentioned above leads to the development of an analytical framework called “agricultural production outsourcing—labor allocation and plot size—fertilizer intensity”. This framework is built on the combination of the division of labor theory and rational smallholder theory. The framework strives to offer insights into how farmers distribute labor resources based on the concepts of division of labor and rational decision-making in the agricultural production process.

There has been a relatively rich body of research on the impact of outsourcing the production chain on the intensity of fertilizer application by farmers, but improvements can be made in the following areas. Firstly, studies have emphasised the single promotional or inhibitory effect of production chain outsourcing on the fertilizer intensity of farmers, but have not explored the non-linear relationship that may exist between the two. Secondly, studies have discussed the impact of off-farm employment or part-time work on fertilizer application intensity, but have rarely included off-farm employment or part-time work in the analysis of labor allocation. Third, studies have paid more attention to the impacts of farm size, land size, and contiguous size on fertilizer application, and less attention to the possible moderating effect of outsourcing the maximum plot size of a farm on fertilizer application intensity. Fourth, more studies have focused on whether farmers participate in the outsourcing of production processes, but not enough attention has been paid to the effect of the extent of farmers’ participation in outsourcing of production processes on the intensity of fertilizer application by farmers. Fifth, more studies have used micro cross-section data, while few studies have used micro unbalanced panel data to discuss the impact of production chain outsourcing on the intensity of fertilizer application by farm households.

Using micro unbalanced panel data from the 2020–2022 China Land Economic Survey (CLES), we analyzed the mechanism of the degree of outsourcing on fertilizer application intensity on farmers’ plots, and empirically tested it using the instrumental variables method in order to provide theoretical support to guide farmers to reduce the intensity of chemical fertilizer application and promote the green development of agriculture.

The possible marginal contributions of this paper compared to existing studies are as follows. Firstly, this paper proposes the hypothesis of an inverted U-shaped relationship between the degree of outsourcing of the production chain of farmers and the fertilizer application intensity on the farmer’s plot, which compensates for the shortcomings of the existing studies. Secondly, by unifying non-farm employment and part-time work into the analysis framework, the mechanism of the inverted U-shaped relationship is discussed from the perspective of labor allocation, and the steepness of the inverted U-shaped relationship and the movement of the inflection point are judged according to the intensity of labor. Thirdly, this paper examines the moderating role of plot size in agricultural production outsourcing on the fertilizer application intensity on farmers’ plots. While existing studies have focused on the influence of factors such as the scale of operation, land size, and contiguous size in fertilizer application, the role that plot size plays in the effect of outsourcing the production chain on the intensity of fertilizer application has been neglected. Fourth, the research perspective is expanded, not stopping at the decision to outsource the agricultural production chain, by examining the heterogeneity of the impact on the fertilizer intensity of farmers from the degree of agricultural production outsourcing. Finally, in terms of the study object, the largest plot of farmers was used as the study object and micro unbalanced panel data were used for the analysis. Compared with the farmer-based survey, the plot-level data can more accurately reflect the fertilizer application strategies adopted due to different plot characteristics, effectively avoiding the average bias that may arise during the data aggregation process and making the study more precise and reliable.

2. Materials and Methods

2.1. Theoretical Analyses and Research Hypotheses

Becker [29] argued that time allocation for farmers follows the principle of profit maximization. Sumner [30] revealed that farmers view the household as a unit to optimize the allocation of their time between farm and non-farm employment and leisure. Stark, et al. [31] and Körner, et al. [32] proposed new labor migration economics in which individual farm household decisions are incorporated into unified collective household decisions. Based on Schulz’s analytical model, Popkin [33] extended the category of “rationality” to the “economic behavior of farmers”. In his masterpiece, The Rational Small Farmer, he puts forward the central hypothesis that farmers are rational individuals or maximizers of family welfare. Therefore, combining the rational smallholder theory and the model of farmers’ behavior, farmers will rationally allocate resources such as land and labor according to the limited land and labor they have, which will grow the household income. Based on the above theories and analyses, we construct a theoretical and analytical framework for agricultural production chain outsourcing–labor allocation and plot size–agrochemical fertilizer application intensity, which is used to explain how farmers maximize the allocation of household benefits between production chain outsourcing and household labor, to make optimal decisions on how to apply fertilizers on their plots. The research framework is shown in Figure 1.

In China, due to the relatively small amount of arable land resources, farmers are not able to produce on a large scale through the acquisition of advanced and large-scale equipment, as is the case for farmers in developed countries. In this case, it is not cost-effective for small farmers to purchase agricultural machinery [34]. Therefore, with limited areas of land, it is difficult to realize great long-term wealth, grow the household economy, and increase the level of household income by outsourcing agricultural production processes and allocating household labor to non-sectoral work. In the context of this realization, outsourcing of agricultural production chain services is relatively common in China, especially in labor-intensive production chain processes [8,17].

2.1.1. Direct Impact of Production Outsourcing on the Fertilizer Application Intensity by Farmers

First, when farmers outsource their agricultural production activites, the outsourcing service can introduce advanced knowledge, technology, factors, and equipment from service organizations into farmers’ agricultural production [13], which can give play to the comparative advantages and the coordination advantage of multi-segment operations, which improve the efficiency of factor utilization and reduce the amount of fertilizer application [13,19,35]. The use of technologies such as mechanized tillage and straw return improves the soil quality and helps to reduce the intensity of fertilizer application. However, at lower levels of production outsourcing, this advantage is not obvious and the effect on fertilizer reduction by farmers may be weaker. Secondly, fertilizer outsourcing allows for the uniform application of fertilizer by mechanized means and reduces waste. Finally, agricultural production service organizations seek to maximize profits, so outsourcing of fertilizer application, for example, may lead to opportunism and moral hazard [7], thus resulting in farmers applying more fertilizer. This can occur if farmers choose to outsource tasks like applying fertilizer and the degree of outsourcing is high. Therefore, as the degree of outsourcing increases, the increase in fertilizer use due to moral hazard may outweigh the reduction brought about by the technological superiority effect of outsourcing. This can lead to an overall increase in fertilizer intensity in farmers’ plots. When outsourcing reaches a certain threshold, the technological superiority effect brought about by outsourcing becomes more significant. This effect can lead to a reduction in fertilizer use, as the outsourced activities are optimized through technological advancements.

Hypothesis 1.

The level of agricultural production outsourcing has an inverted U-shaped relationship with the intensity of fertilizer application on farmers’ plots.

2.1.2. Indirect Effects of Labor Allocation

Participation in production outsourcing has a “crowding-out effect” on the labor of farmers’ households, which changes the traditional intensively cultivated and involutional production model, as rational smallholders tend to avoid the output risk by over-applying fertilizers [13]. Diverse fertilizer application strategies result from the fact that producers select varying levels of outsourcing based on their production characteristics and household endowments.

The primary objective of farmers who engage in low levels of outsourcing is to alleviate labor shortages by outsourcing labor-intensive production tasks [17,28,34]. This typically entails activities such as land preparation and harvesting [8], which frequently involve the use of mechanized operations such as tillage, sowing, and harvesting [19]. The outsourced tasks are typically executed with minimal moral hazard from the supplying entities and low supervision costs for producers [35]. Fertilizer application outsourcing is restricted during this phase. Fertilizers are primarily applied by farmers, who are responsible for the specific application procedures and the determination of the appropriate quantities.

Intensive agricultural practices enable farmers to reduce the labor-intensive effort necessary for the application of reduced quantities of fertilizer as the degree of outsourcing increases. This transition to outsourcing results in an additional decrease in agricultural labor inputs and enables farming households to allocate a greater amount of manpower to non-farming activities. As a result, there has been a rise in non-farming and part-time employment, with the elderly, weak, ill, disabled, and women now assuming significant roles in agricultural production. Farming households frequently choose simplified production methods to reduce labor inputs as a result of the substitution relationship between chemical fertilizers and labor [13]. This trend underscores the increasing significance of non-farming activities within farming communities and the changing dynamics of agricultural labor allocation.

On the one hand, farmers often compensate for the scarcity of labor inputs by applying more affordable fertilizers, which exacerbates the over-application of fertilizers [19,20]. On the other hand, the degree of outsourcing may encourage the opportunistic behavior of the supplying entities, which in turn increases the cost of supervision and control of the farmers. This, in turn, may also result in the over-application of fertilizers [35]. The intensity of fertilizer administration increases in tandem with the depth of outsourcing.

When the level of outsourcing exceeds a certain point, the impact on the amount of labor provided by farm households becomes more noticeable. In these cases, farmers may shift their attention towards non-agricultural jobs and part-time activities, leading to a decrease in the intensity of agricultural labor input. Besides obtaining basic outsourcing services, farmers may also use outsourcing services to acquire the necessary tools and equipment for production [35]. This may include outsourcing fertilizer application, shifting the responsibility of fertilizer management from the farmers to professional service providers. The main part of fertilizer application then changes from farmers to professional service organizations [35]. In this case, fertilizer application is carried out by the outsourced service organization, which determines the amount of fertilizer to be applied. Although there may be high moral risks and supervision costs, the technology introduction effect and external learning effect are obvious [35]. Through the professional fertilizer application knowledge and equipment of the outsourced service organizations, fertilizer can be applied scientifically based on soil conditions, crop demand, and other factors [19], and mechanized uniform fertilizer application can be adopted to avoid the uneven and irregular problems of manual fertilizer application [21] in order to reduce chemical fertilizer application intensity. When the degree of outsourcing is high, the production process improves the soil’s organic matter content by taking advantage of multi-link operations, such as straw return to the field, deep ploughing, and deep polishing, thus reducing the agrochemical inputs [35]. In addition, through subsidies and other means, the Government promotes the adoption of green production techniques by organizations that outsource their agricultural production processes, which has helped to reduce the intensity of fertilizer application [36]. At this stage, as the degree of outsourcing deepens, the fertilizer application intensity decreases. Accordingly, the following research hypothesis is proposed:

Hypothesis 2.

The degree of agricultural production outsourcing affects the fertilizer intensity on farmers’ plots through labor allocation.

2.1.3. Moderating Effects of Plot Size

Plot size plays an important moderating role in influencing the degree of outsourcing on fertilizer application intensity to plots. On the one hand, large-scale plots provide favourable conditions for mechanization and scale effects, thus promoting vertical division of labor in agricultural production [8]. As plot size increases, the higher the degree of vertical division of labor, the easier it is for farmers to adopt modern agricultural production techniques and equipment, improve production efficiency, promote more accurate and efficient use of vegetation, and reduce the intensity of chemical fertilizer application, thus achieving the suppression of chemical fertilizer intensity by increasing the degree of outsourcing. On the other hand, at a given degree of outsourcing, the size of the plot directly affects the productivity and cost-effectiveness of the outsourced service organization. Larger plot sizes enable outsourcing service organizations to more fully exploit the scale effect, optimize resource allocation, reduce production costs, and reduce fertilizer inputs. Thus, an increase in plot size results in a flattening of the inverted U-shaped curve, i.e., it is easier to achieve lower fertilizer application intensity on larger plots. On the contrary, when plot sizes are small, land fragmentation is high, which not only increases the cost of outsourcing services and reduces farmers’ propensity to outsource, but also limits the degree of the vertical division of labor. At this point, it is more difficult to achieve operations at scale and reduce the fertilizer application intensity, so the inhibitory effect on fertilizer is relatively insignificant. It is proposed accordingly:

Hypothesis 3.

Plot size negatively moderates the effect of the degree of agricultural production outsourcing on fertilizer application intensity on farmers’ plots.

Hypothesis 3-1.

Plot size moderates the inverted U-shaped curve of the degree of outsourcing in the production chain on fertilizer application intensity in farmers’ plots.

Hypothesis 3-2.

Plot size shifts the inflection point of the inverted U-shaped curve of the degree of agricultural production outsourcing on fertilizer application intensity to the left.

2.2. Data and Variables

2.2.1. Study Area

The Jiangsu Province, a leading agricultural province in China, demonstrates a substantial degree of economic development in the agricultural sector. Hence, it was suitable to choose rice and maize farmers in the Jiangsu Province as the participants of this research to examine the influence of outsourcing production processes on the level of fertilizer usage. The Jiangsu Province demonstrates a significant level of agricultural mechanization. The predicted mechanized agricultural cultivation area in 2022 was 6955.49 thousand hectares. The mechanical sowing area was expected to be 5797.70 hectares, the mechanical plant protection area to be 6582.65 hectares, and the mechanical harvesting area to be 5965.52 thousand hectares. In the year 2022, the province of Jiangsu successfully planted 2221.42 thousand hectares of rice paddy, resulting in a total production of 199,161 tons (https://tj.jiangsu.gov.cn/2023/nj10/nj1002.htm, accessed on 25 September 2024). Figure 2 displays the precise geographical position of the Jiangsu Province.

2.2.2. Data Sources

All the data utilized in this study originate only from the China Land Economy Survey (CLES), which was carried out by Nanjing Agricultural University. A baseline study was conducted in 2020, followed by two follow-up studies in 2021 and 2022. The survey employed the probability proportional sampling (PPS) method to select 26 sample counties and districts from 13 prefecture-level cities in the Jiangsu Province in 2020. A total of 52 administrative villages were included in the survey, representing a total of 2628 farm households. This baseline survey was conducted to gather data. In 2021, a sample of 2420 households in 48 villages in 12 prefectural cities was completed; in 2022, a sample of 1203 farm households in 24 villages in 12 counties in 6 cities was completed. The survey recorded detailed information on farmers and agricultural production, providing important support for this study to explore the impact of agricultural production outsourcing on fertilizer application intensity. After using logical reasoning to approximate missing values, certain variables with a significant number of missing values were filled in. For example, missing values for the refinement of operating arable land were replaced with the number of corresponding contracted land parcels. Similarly, missing values for the size of the operation were filled in using the maximum parcel area. Any remaining missing values and outliers that could not be addressed were removed, resulting in a total of 1836 valid samples.

2.2.3. Variables

• Dependent variables. Referring to studies such as Huang, et al. [37] and Sun, et al. [38], the actual fertilizer input per unit area was used as the dependent variable, i.e., the intensity of fertilizer Jin per mu (1 Jin = 0.5 kg, 1 mu = 1/6.07 acre). In order to eliminate heteroskedasticity, guarantee data smoothness, and eliminate spurious regressions, the dependent variable was treated as logarithmic. Meanwhile, the cost of fertilizer input per mu was used for the robustness test [37]. The annual fertilizer application intensity of farmers is gradually declining, as shown in Figure 3;

2.

Agricultural production outsourcing is the explanatory variable that describes the practice of farming families assigning duties related to agricultural production to persons or organizations [3,39,40]. A binary variable does not adequately reflect the degree of farmers’ involvement in outsourcing, even if some studies utilize the adoption of outsourcing procedures as an indicator of farmers’ participation in agricultural outsourcing [10,12,17];

In order to overcome this constraint, other studies have investigated alternative metrics, such as the number of farmers involved in outsourced production processes [15] or the cost of outsourcing per unit area [14], in order to more accurately assess the level of farmers’ participation in outsourcing. This study focuses on quantifying farmers’ level of involvement in outsourcing production linkages, referred to as the “degree of outsourcing”. The extent of outsourcing was assessed by analyzing the several production tasks that farmers delegated to external parties, such as plowing, seedling growing, planting, pesticide spraying, fertilizer application, harvesting, and straw return to the field. It is worth mentioning that farmers were identified as outsourcing fertilizer application by obtaining technical services for soil testing and fertilizer application and using formulated fertilizer. This is based on a 2020 finding from the Jiangsu Province, where thousands of villages and ten thousand households are involved in enterprises’ actions to reduce fertilizer usage and improve efficiency. The agriculture-related counties, cities, and districts in the province are conducting soil testing and formulating fertilizer in order to establish more than 5 demonstration areas for “soil testing and formula, double reduction of chemical fertilizers”, covering 1000 administrative villages in the province. This initiative provides free soil testing and fertilizer formulation services for 10,000 farmers, farms, and societies, with the participation of 100 fertilizer enterprises and farmers working together to enhance fertilizer supply services and increase the number of farmers receiving technical assistance. In the Jiangsu Province, the strategy of reducing fertilizer usage and improving efficiency through the socialized service organization of ‘unified measurement, unified allocation, unified supply, and unified application’ has effectively addressed the challenges of ‘measurement-allocation-production’ and bridged the gap in the five main stages of ‘measurement-allocation-production-supply-application’. In addition, the extent of outsourcing was evaluated by computing the expense of outsourced production connections per unit of land (mu), which was then subjected to a test of reliability. Figure 4 illustrates the examination of farmers’ yearly delegation of production operations. It indicates that harvesting has the greatest level of outsourcing, while fertilizer application has the lowest. In general, there is a clear and consistent increase in the extent of outsourcing over the period of time.

3.

Control variables. Considering the results of existing studies [9,11,14], the main controlled variables were the head of household, family, plot, crop, village, and regional and temporal characteristics. The variable settings are shown in

Table 1. Variable settings and descriptive statistics.

Variable	Description/Measurements	Mean	Std. Dev	Min	Max
Fertilizer application intensity	Fertilizer input intensity (Jin per mu) of the plot, taken in logarithms	4.807	0.514	2.398	5.707
Fertilizer costs	Amount of fertilizer applied per mu on the plot (¥1000/mu), taken as a logarithmic figure	5.250	0.426	3.912	6.397
Degree of outsourcing (number)	Number of segments involved in the production outsourcing chain (number)	2.957	1.743	0.000	7.000
Degree of outsourcing squared (number)	Square of the number of links outsourced in the agricultural production chain	11.782	11.027	0.000	49.000
Degree of outsourcing (cost)	Average cost per mu of participation in the agricultural production chain (¥1000/mu)	0.246	0.239	0.000	1.660
Degree of outsourcing squared (cost)	Square of the per-mu cost of outsourcing the agricultural production chain	0.118	0.325	0.000	2.756
Labor allocation	Plot agricultural labor input hours (days/mu), plus 1 to take logarithms	2.386	1.055	0.000	5.591
Plot size	Maximum parcel size (mu) in logarithms	1.006	0.949	−0.693	4.456
Gender	Gender of the head of household, male = 1, female = 0	0.721	0.449	0.000	1.000
Age	Age of head of household (years)	61.448	10.174	31.000	81.000
Self-reported health	Self-assessment of the health status of the head of household: 1 = laborless, 2 = poor, 3 = medium, 4 = good, 5 = excellent	3.946	1.044	1.000	5.000
Education	Educational attainment of head of household: 1 = illiterate, 2 = primary school, 3 = lower secondary school, 4 = upper secondary school and secondary school, 5 = university and above	1.606	0.982	0.000	5.000
Risk	1 = risk averse, 0 = other	0.739	0.439	0.000	1.000
	1 = risk neutral, 0 = other	0.181	0.385	0.000	1.000
Social network	Number of mobile phone contacts for the head of household (persons)/max	0.087	0.150	0.000	1.000
Party	Communist Party member in the household: 1 = yes, 0 = no	0.255	0.436	0.000	1.000
Village cadres	Village cadre in the family, 1 = yes, 0 = no	0.156	0.363	0.000	1.000
Policy household	Whether the household is established as a poor household, five-guarantee household, low-income household, or disabled household: 1 = yes, 0 = no	0.096	0.295	0.000	1.000
Agricultural training	Whether household members receive training in agricultural technology: 1 = yes, 0 = no	0.399	0.490	0.000	1.000
Non-farm training	Whether household members receive training in off-farm technologies: 1 = yes, 0 = no	0.287	0.453	0.000	1.000
Agricultural labor	Number of agricultural laborers in farming households as a percentage	0.544	0.291	0.000	1.000
Aged dependent population	Percentage of older persons aged 60 and over in households	0.349	0.336	0.000	1.000
Child dependency population	Percentage of children under 16 in households	0.101	0.136	0.000	0.500
Value of productive fixed assets	Value of productive fixed assets of agriculture in the household (in thousands of yuan), plus 1 to take logarithms	0.902	1.571	0.000	6.066
Own mechanical farming	Whether or not they use their machinery: 1 = Yes, 0 = No	0.157	0.260	0.000	1.000
The size of arable land operated	Scale of rice and maize cultivation (mu) in logarithms	1.717	1.450	−0.693	6.273
Degree of agricultural insurance	Amount of agricultural insurance purchased (¥), plus 1 to take logarithms	2.859	2.797	0.000	9.550
Irrigation	Whether irrigation is possible: 1 = yes, 0 = no	0.895	0.306	0.000	1.000
Soil	1 = loam, 0 = other	0.181	0.385	0.000	1.000
Fertility	1 = poor, 2 = moderate, 3 = good	2.351	0.635	1.000	3.000
Road	Distance to the nearest hardened concrete road (kilometres), plus 1 to take logarithms	0.212	0.279	0.000	1.609
Disaster	1 = affected, 0 = no	0.469	0.499	0.000	1.000
Land transfer	1 = transferred land, 0 = contracted land	0.249	0.433	0.000	1.000
Fractionalisation	Total number of plots of operational arable land	5.787	9.670	0.000	70.000
Crop	Whether rice is grown: 1 = yes, 0 = no	0.812	0.391	0.000	1.000
	Whether or not maize is grown: 1 = yes, 0 = no.	0.198	0.398	0.000	1.000
Village economics	Average village income level (¥1000/person in logarithms)	2.823	1.039	−6.152	3.689

Note: Data are 1 percent and 99 percent rounded off.

;

4.

Mechanism variables. Drawing on existing research [41,42,43], labor allocation was characterised using the input intensity of farm labor on the plot. Plot sizes are inscribed with the plot area;

5.

Instrumental variable. Referring to the study of Chang et al. [14] and Zhang et al. [19], the average level of other farmers’ participation in outsourcing within the same village was used as an instrumental variable. In terms of relevance, rural China is a typical humane society; there are relatively frequent social interactions and imitative learning between neighbors within the village, and the degree of outsourcing by other farmers has a demonstrative effect and a peer effect on other farmers [44]. In terms of exclusivity of instrumental variables, the decision to purchase outsourcing services was viewed as the outcomse of collective rational decision-making within the farmer’s family. The average level of outsourcing purchases by other farmers in the same village would not directly impact the farmer’s fertilizer intensity, thus meeting the exclusivity constraint. Similarly, the mean value of the agricultural labor input intensity of other farmers’ plots in the same village was used as an instrumental variable for labor allocation.

For the sample’s descriptive statistics and variable meanings, see Table 1. The mean value of 122.364 Jin/mu (the logarithmic value in the table is 4.807, the same as below), the maximum value of 300.967 Jin/mu, and the minimum value of 11.001 Jin/mu showed large variations in fertilizer application intensity. In terms of the level of outsourcing, farmers took part in 2.957 production stages on average, which is almost half of the total production stages. This suggests that farmers have a strong incentive to take part in the outsourcing of production stages.

The degree of outsourcing was further curve-fitted to the fertilizer application intensity in the plots, as shown in Figure 5. It can be seen that the degree of outsourcing shows an inverted U-shaped relationship with the fertilizer application intensity in the plots, which tentatively validates research Hypothesis 1. However, more rigorous measurement methods are needed to confirm the existence of the inverted U-shaped relationship.

2.3. Model Speccification

The extent to which farmers, as rational individuals with limited resources, engage in the outsourcing of production processes is determined by their desire to maximize profits. This decision is influenced by a variety of factors that can be both observed and unobserved. Not considering unobservable characteristics, such as social interaction, skill, interests, etc., might lead to a biased evaluation. In order to tackle the issue of endogeneity arising from self-selection and the potential exclusion of variables, this study utilizes unbalanced panel data and incorporates area and time fixed effects as supplementary controls to alleviate the impact of the unobservable factors. In addition, the study employs instrumental variable (IV) regression to address the issue of endogeneity and improve the precision of the estimation.

2.3.1. Basic Regression Model

In this study, the dependent variable is the intensity of fertilizer application by farmers, which is a continuous variable. The degree of outsourcing of production processes by farmers is expressed using the number of outsourced processes, which is a discrete variable ranging from 0 to 7. Therefore, the instrumental variable method was used for the estimation in this paper. To investigate the effect of the degree of outsourcing on fertilizer application intensity in farmers’ plots, the following model was set up:

$$\ln Y_{it}=Cons+{\beta }_{1}O{s}_{it}+{\displaystyle \sum {\lambda }_j{X}_{it}}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(1)

where $O{s}_{it}$ is the degree of outsourcing of the ith farmer in year t, $\ln Y_{it}$ is the logarithm of the fertilizer application intensity of the ith farmer in year t; $X_{it}$ are the control variables, such as head of household, family, plot, village level and other characteristics, and ${\beta }_1$ and ${\lambda }_j$ are the parameters to be estimated. If ${\beta }_1$ is significantly positive, it means that the degree of outsourcing promotes the fertilizer application intensity in the farmers’ plots; vice versa, it inhibits the fertilizer application intensity in farmers’ plots. If it is not significant, it may be in a non-linear relationship; ${\mu }_i$ is a region fixed effect, ${\delta }_t$ is the time fixed effect, and ${\epsilon }_{it}$ is the error term. To explore the possible inverted U-shaped relationship, a squared term for the degree of outsourcing is introduced into Equation (1):

$$\ln Y_{it}=Cons+{\beta }_{1}O{s}_{it}+{\beta }_{2}O{s}_{it}^2+{\displaystyle \sum {\lambda }_j{X}_{it}}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(2)

In Equation (2), if it is significant, this means that there is a non-linear relationship between the degree of outsourcing and the fertilizer application intensity of farmers. An inverted U-shaped (or U-shaped) relationship is implied if it is significantly negative (or positive), and its axis of symmetry lies within the study data interval.

2.3.2. Mechanism Analysis Model

In order to investigate the mechanism of the effect of the degree of outsourcing on the fertilizer application intensity in the plots, regressions of outsourcing on the mechanism variables and regressions of the mechanism variables on fertilizer application emphasis were conducted. The following model was set up:

$$\ln M_{it}=Cons+{\gamma }_{1}O{s}_{it}+{\displaystyle \sum {\lambda }_j{X}_{it}}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(3)

$$\ln Y_{it}=Cons+{\alpha }_1\ln M_{it}+{\alpha }_2{(\ln M_{it})}^2+{\displaystyle \sum {\lambda }_j{X}_{it}}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(4)

Equation (3) is a regression of outsourcing on the mechanism variable, and Equation (4) is a regression of the mechanism variable on fertilizer. Where $M_{it}$ is the mechanism variable, i.e., agricultural labor input intensity, ${\gamma }_1$ is the parameter to be estimated; when this is significantly positive, it indicates that agricultural production outsourcing significantly promotes the mechanism variable and vice versa. The remaining parameters are the same as in Equation (1). When ${\alpha }_2$ is significant, there is a mediating effect. The instrumental variables approach was used to address the endogeneity of outsourcing and labor allocation. Based on Equation (2), the moderating effect of plot size is further explored by setting up the following model:

$$\begin{array}{c}\ln Y_{it}=Cons+{\gamma }_{1}O{s}_{it}+{\gamma }_2\ln M_{it}+{\gamma }_{3}O{s}_{it}\times \ln M_{it}+{\gamma }_{4}O{s}_{it}^2+{\gamma }_{5}O{s}_{it}^2\times \ln M_{it}+\\ {\displaystyle \sum {\lambda }_j{X}_{it}}+{\mu }_i+{\delta }_t+{\epsilon }_{it}\end{array}$$

(5)

where ${\gamma }_5$ is significant, $M$ has a moderating effect. For the inverse $U$ relationship, ${\gamma }_5$ is significantly positive, and the moderating variable weakens the inverse relationship, i.e., the curve flattens as the moderating variable increases, implying that the same level of outsourcing makes farmers with larger plot sizes lower their fertilizer application intensity; conversely, it increases the intensity of fertilizer.

Referring to Haans et al. [45], the inflection point is obtained by taking the inverse of (5) concerning the moderating variable $M$ and making it zero:

$$O{s}^{\ast }={\displaystyle \frac{-{\gamma }_1-{\gamma }_3\ln M}{2{\gamma }_4+2{\gamma }_5\ln M}}$$

(6)

Solving Equation (6) for the derivative for the moderator variable $\ln M$ yields the change in the inflection point:

$${\displaystyle \frac{\partial O{s}^{\ast }}{\partial \ln M}}={\displaystyle \frac{{\gamma }_1{\gamma }_5-{\gamma }_3{\gamma }_4}{{(2{\gamma }_4+2{\gamma }_5\ln M)}^2}}$$

(7)

The effect of the change in the moderating variable on the inflection point of the curve can be determined with (7). Specifically, when ${\gamma }_1{\gamma }_5-{\gamma }_3{\gamma }_4>0$ holds, the inflection point shifts to the right as the moderating variable increases, implying that the suppression effect is realised at a higher degree of outsourcing; when ${\gamma }_1{\gamma }_5-{\gamma }_3{\gamma }_4<0$ holds, the opposite is true. Similarly, the instrumental variables approach was used to address the endogeneity.

3. Results

3.1. Benchmark Regression

Table 2. Benchmark regression results.

Variable	(1)	(2)	(3)	(4)	(5)
Degree of outsourcing	0.006	0.012	0.374 ***	0.340 **	0.028 ***
	0.024	0.029	0.119	0.135	0.008
Degree of outsourcing squared			−0.062 ***	−0.056 ***
			0.019	0.021
Instrumental variable					−0.010
					0.019
Constant	4.836 ***	4.397 ***	4.441 ***	4.161 ***	4.376 ***
	0.079	0.242	0.160	0.294	0.232
Control variable	no	yes	no	yes	yes
Regional fixed effects	yes	yes	yes	yes	yes
Time fixed effect	yes	yes	yes	yes	yes
N	1836	1836	1836	1836	1836
Phase I F-value	1117.11	148.93	906.81/367.64	143.84/58.56
Instrumental variable correlation	0.701 ***	0.635 ***	0.376/1.039 ***	0.258/1.025 ***

Note: Second line is robust standard error; significance ** p < 0.05, *** p < 0.01.

reports the regression results of the instrumental variable method, where (1) and (2) are listed as the results of the primary term regression and (3) and (4) are listed as the results of the secondary term regression. The results show that the F-value of the first-stage regression is much larger than 10, indicating that there is no weak instrumental variable problem; meanwhile, the instrumental variable and its squared term are significantly correlated with the degree of outsourcing and the squared term of the degree of outsourcing, respectively, which satisfies the correlation requirement. Furthermore, column (5) shows that the instrumental variable is not correlated with the dependent variable, which satisfies the exclusivity requirement. Therefore, the instrumental variables are valid and the regression results are reliable.

From the results of the primary regression in columns (1) and (2), it can be seen that the effect of the degree of outsourcing on fertilizer application intensity on farmers’ plots is not significant. This implies that the correlation between the extent of outsourcing and the intensity of fertilizer application in farmers’ plots is not a straightforward linear relationship. The regression results indicate that after including the squared term of the degree of outsourcing in columns (3) and (4), the coefficients are both statistically significant. The symmetry axes for these coefficients are 2.997 and 3.037, respectively. The differences between these axes are relatively small, ranging from 0 to 7. This suggests that there is a significant inverted U-shaped relationship between the degree of outsourcing and the intensity of fertilizer application on farmers’ plots. The inverted U-shaped relationship indicates that the fertilizer application intensity on the plots tends to increase and then decrease as farmers’ participation in the vertical division of agricultural labor deepens. This may be because when the degree of outsourcing is low, farmers are the main applicators of fertilizers and decide the amount of fertilizer to be applied, and the substitution effect of outsourcing on farmers’ labor and production technology is weak. Farmers still need to put in a lot of labor, thus tend to cultivate their plots carefully and reduce the intensity of chemical fertilizer application; with the deepening of outsourcing, the intensity of their agricultural labor inputs decreases. In order to avoid the risk of possible yield reduction and crop loss, farmers tend to increase the use of “cheap” fertilizers instead of “expensive” fertilizers, which increases the fertilizer application intensity in the plots. However, when the degree of outsourcing exceeds a certain threshold, the degree of substitution of labor and production technology by outsourcing service organizations significantly increases, and the intensity of farm labor inputs significantly decreases, resulting in the main fertilizer application being from outsourcing service organizations. This, with the dual support of professional equipment and professional technicians, can reduce the fertilizer application intensity through scientific fertilizer application. Therefore, the effect of the degree of outsourcing on the intensity of fertilizer application in farmers’ plots is first promoted and then inhibited, showing an inverted U-shaped relationship. Hypothesis 1 was thus proven.

3.2. Robustness Test

The results of the robustness tests are shown in

Table 3. Robustness test results.

Variable	(1)	(2)	(3)	(4)	(5)
	Replacement of the Dependent Variable	Replacement of Explanatory Variables	Adding Cubic Terms	Excluding Organic Fertilizer Data	Raw Data
Degree of outsourcing	0.099	−3.101 *	−1.361	0.314 ***	0.396 ***
	0.120	1.644	1.064	0.116	0.141
Degree of outsourcing squared	0.099	2.455 *	−1.361	0.314 ***	0.396 ***
	0.120	1.272	1.064	0.116	0.141
Degree of outsourcing cubic term			−0.084
			0.052
Constant	5.481 ***	4.735 ***	4.041 ***	4.249 ***	3.190 ***
	0.242	0.421	0.371	0.287	0.400
Control variable	yes	yes	yes	yes	yes
Regional fixed effects	yes	yes	yes	yes	yes
Time fixed effect	yes	yes	yes	yes	yes
N	1836	1836	1836	1780	1836

Significance * p < 0.10, *** p < 0.01.

.

First, the dependent variables were replaced. The cost of fertilizer per mu (taken as a logarithm) was used as a proxy for fertilizer application intensity and regressed using the instrumental variable method. The squared term of the degree of outsourcing is significantly negative and the symmetry axis 1.568 is between 0 and 7, so the inverted U-shaped relationship holds.

Second, the core explanatory variables were replaced. The cost of outsourcing the purchased production chain per mu was used as a proxy variable for the degree of outsourcing and regressed using the instrumental variable method, and the results are shown in Table 3 (2). It can be seen that the squared term of the degree of outsourcing is significantly negative and the symmetry axis is 0.632 thousand yuan/mu, and the range of outsourcing costs is between 0.000 thousand yuan and 1.66 thousand yuan, so the inverted U-shaped relationship is significant.

Again, the cubic term was introduced. After the introduction of the cubic term for the degree of outsourcing in the baseline regression, the coefficient was found to be insignificant when estimated by the instrumental variable method, ruling out a possible regression-type parabolic relationship between the degree of outsourcing and the fertilizer application intensity on farmers’ plots.

Fourth, the estimation model was replaced. For the possible self-selection problem, the total number of outsourced links was used to convert the degree of outsourcing into a ratio between 0 and 1; the 0–1 interval was then divided by quartiles, and the generalised propensity score matching method was used for the estimation, which yields an inverted U-shaped dose-response function (shown in Figure 6), indicating that there is a significant inverted U-shaped relationship.

Fifth, the data were transformed. After excluding the organic fertilizer data, the quadratic term for the degree of outsourcing remained significantly negative and the axis of symmetry, 3.077, remains between 0 and 7, confirming the existence of an inverted U-shaped relationship. Further regressions using unshrunken data showed that the quadratic term for the degree of outsourcing remained significantly negative and the symmetry axis of 3.143 remains between 0 and 7.

Finally, the Utest was performed. Referring to the study of Lind et al. [46], the inverted U-shaped relationship was tested using the Utest. The test showed that the extreme point of the inverted U-shaped curve was 3.036, between 0 and 7; the slopes of the left and right sides of the axis of symmetry were significant at 0.340 and −0.444 at a 1% confidence level. The T-value of the overall inverted U-shape existential was 2.51, which was significant at a 1% confidence level, suggesting that the inverted U-shape relationship was established. All the above methods confirmed the existence of a significant inverted U-shape between the degree of outsourcing and the intensity of fertilizer application in the plots, indicating that the benchmark results are robust.

3.3. Mechanism Analysis

The theoretical analysis showed that the effect of the degree of outsourcing on the intensity of fertilizer application in the plots was achieved through labor allocation, and was also moderated by the moderating effect of the plot size. Regressions were conducted using the instrumental variable method and the results are presented in

Table 4. Regression results for mediating and moderating effects.

Variable	(1)	(2)	(3)
	Labor Allocation	Fertilizer Application Intensity	Fertilizer Application Intensity (Plot Size)
Degree of outsourcing	−0.240 ***		0.981 **
	0.060		0.392
Degree of outsourcing squared.			−0.155 **
			0.064
Labor allocation		1.306 **
		0.556
Labor allocation squared		−0.257 **
		0.103
Degree of outsourcing×plot size			−0.450 *
			0.242
Degree of outsourcing squared×plot size.			0.059 *
			0.032
Plot size	0.072	0.057	0.698 *
	0.071	0.035	0.373
Constant	2.296 ***	2.955 ***	3.704 ***
	0.440	0.705	0.434
Control variable	yes	yes	yes
Regional fixed effects	yes	yes	yes
Time fixed effect	yes	yes	yes
N	1836	1836	1836

Significance * p < 0.10, ** p < 0.05, *** p < 0.01.

.

3.3.1. Identification of the Role Path of Labor Allocation

From the results in Table 4 (1), the increase in the degree of outsourcing significantly reduced the labor input intensity of farming. The probable reason for this is that the choice of outsourcing services by the farmers enabled the household labor force to be freed from its original agricultural production for a non-farm allocation, thus reducing the labor input intensity.

The data presented in Table 4 (2) indicate that the squared term of labor input intensity has a statistically significant negative effect. The symmetry axis is located at 2.538, which lies between the highest value of 5.591 and the minimum value of 0. This indicates that there is a non-linear relationship between the amount of work used and the amount of fertilizer applied. Initially, as the labor input increases, the fertilizer application also increases. However, at a certain point, as the intensity of the farm inputs decreases, the fertilizer application decreases. At low levels of outsourcing, farmers usually take care of fertilizer application themselves, deciding on the right quantity to be utilized. During this stage, there is a high level of labor required on the farm, which leads to a focus on intensive cultivation methods. As a result, the application intensity of fertilizer may be reduced. As the level of outsourcing rises, the amount of farm labor required decreases. When farmers are concerned about their crop yields, they may respond by increasing the amount of fertilizer they use in order to make up for the decrease in labor inputs, resulting in an overall increase in fertilizer usage. When the amount of outsourcing exceeds a certain critical point, farmers demonstrate a greater degree of specialization in their agricultural output. This shift results in the primary responsibility for fertilizer application transitioning from farmers to outsourced service organizations. The implementation of scientific fertilizer application practices by these service providers can help reduce the intensity of chemical fertilizer application. Hypothesis 2 is thus proved.

3.3.2. Identification of Moderating Effects of Plot Size

The correlation coefficient between the degree of outsourcing and plot size was considerably positive, suggesting that plot size had a considerable moderating effect on the link between the degree of outsourcing and fertilizer application intensity. The inverted U-shaped relationship becomes flatter as the area of the plot increases, implying that the intensity of fertilizer application is relatively low on larger plots, even if the degree of outsourcing is the same, reflecting the scale effect. The value of −0.012 was found to be less than 0 by calculating Equation (7), indicating that the larger the plot size, the point of inflection of the inverted U-shaped curve shifted to the left, i.e., on larger plots, farmers were able to achieve suppression of the intensity of fertilizer application at a lower level of outsourcing. The expansion of plot size is conducive to the vertical division of labor in agricultural production among farmers, increasing the degree of outsourcing and specialisation, enhancing the substitution effect, and reducing the intensity of fertilizer application; at the same time, under the established degree of outsourcing, the larger the plot size, the more conducive it is to optimize the allocation of resources, improving the efficiency of production, giving full play to the scale effect, and lowering the cost of production, which in turn reduces the inputs of chemical fertilizers. Hypotheses 3, 3-1 and 3-2 are proved.

3.4. The Effect of Different Types of Production Outsourcing on Fertilizer Application Intensity

There are differences in the labor intensity and degree of mechanization required for different production stages in agriculture [44], which have different “crowding out” effects on farm labor and different fertilizer application strategies. Outsourcing was classified into labor-intensive (ploughing, planting, harvesting, straw return) and technology-intensive (seedling, plant protection and fertilizer outsourcing) for the regression, and the results are shown in

Table 5. Effect of different types of agricultural production outsourcing on the intensity of fertilization in plots.

Variable	(1)	(2)	(3)	(4)
	Fertilizer Application Intensity (Labor-Intensive)		Fertilizer Application Intensity (Technology-Intensive)
Degree of outsourcing	1.031		0.041
	0.795		0.178
Degree of outsourcing squared	−0.240		−0.018
	0.192		0.031
Labor-intensive production outsourcing		0.589 ***
		0.226
Technology-intensive production outsourcing				−0.825 **
				0.332
Constant	3.983 ***	3.763 ***	4.858 ***	5.184 ***
	0.433	0.397	0.283	0.348
Control variable	yes	yes	yes	yes
Regional fixed effects	yes	yes	yes	yes
Time fixed effect	yes	yes	yes	yes
N	1137	1137	892	892

Significance ** p < 0.05, *** p < 0.01.

.

It can be seen that the inverted U-shaped relationship was not significant in both labor and technology-intensive production outsourcing samples, further regression using whether labor and technology-intensive production chain outsourcing was used found that participation in labor-intensive production chain outsourcing significantly promoted the intensity of fertilizer application to the farmers’ plots, while on the contrary, participation in technology-intensive production chain outsourcing significantly suppressed the intensity of fertilizer application to the plots. The possible reasons are that farmers involved in purely labor-intensive production chain outsourcing have a lower degree of specialisation in agricultural production, and their fertilizer application is carried out by the farmers themselves who decide on the amount of fertilizer to be applied, and avoid a reduction in the input of farm labor, they tend to apply more fertilizer, thus increasing the intensity of fertilizer application; whereas farmers involved in technologically-intensive production chain have a higher degree of specialisation, and their fertilizer application is organized by outsourced service providers who decide on the amount of fertilizer to be applied, which is conducive to the use of technical comparative advantages and reduces the amount of fertilizer applied. This further confirms the theoretical analyses in this paper.

3.5. Heterogeneity Analysis

3.5.1. Heterogeneous Effects of Education

The level of education is an important factor influencing the allocation of labor and the intensity of fertilizer application by farm households. Typically, farmers with higher levels of education are more likely to find employment opportunities in the off-farm market and thus tend to adopt outsourcing of production processes as an alternative to traditional agricultural labor. As a result, at lower levels of outsourcing, farmers may still apply fertilizer on their own and may increase fertilizer application to compensate for labor shortages, thus driving up the intensity of fertilizer application; whereas at higher levels of outsourcing, farmers are more inclined to purchase outsourcing services, which may reduce the need for farmers to apply fertilizer on their own, thus suppressing the intensity of fertilizer application, resulting in an inverted U-shaped relationship. In contrast, farmers with lower levels of education are more likely to rely on traditional agricultural production methods and less likely to use outsourced alternative labor due to relatively fewer opportunities to find work in the off-farm market and are therefore more likely to rely on traditional experience in fertilizer application, which may lead to over-fertilization and thus promote an increase in fertilizer application intensity, with an inverted U-shaped relationship that is not significant. To analyze in depth the effect of education level on the inverted U-shaped relationship, the average education level of farm households was divided into two groups of high and low according to the mean value and regressed and the results are shown in

Table 6. Heterogeneity regression results.

Variable	(1)	(2)	(3)	(4)
	Highly Educated	Poorly Educated	Older	Younger
Degree of outsourcing	0.415 ***	0.430	0.106	0.618 ***
	0.143	0.370	0.163	0.211
Degree of outsourcing squared	−0.068 ***	−0.069	−0.013	−0.104 ***
	0.024	0.057	0.027	0.033
Constant	4.073 ***	4.132 ***	4.422 ***	3.313 ***
	0.350	0.580	0.548	0.465
Control variable	yes	yes	yes	yes
Regional fixed effects	yes	yes	yes	yes
Time fixed effect	yes	yes	yes	yes
N	931	905	889	940

Significance *** p < 0.01.

. The inverted U-shaped relationship was significantly present in the subgroups with a high level of education, whereas in the subgroups with a low level of education, the relationship was not significant. This indicates that the relationship between the level of outsourcing and fertilizer application intensity is influenced by heterogeneity in education level.

3.5.2. Heterogeneous Effects of Age

Age is one of the important factors affecting labor allocation and fertilizer application intensity in farm households. Generally, as farmers age, their physical strength diminishes and job market opportunities diminish, so they are more likely to be involved in the labor-intensive aspects of agricultural production. At this time, the “crowding out effect” of outsourcing on farmers’ labor is weak and farmers are still the main applicators and take the lead in applying fertilizers, often based on their experience, which may increase the intensity of chemical fertilizer application and lead to over-fertilizing. On the contrary, younger farmers are more likely to find employment in non-farm markets, which makes them invest less time in agricultural production. Therefore, when the degree of outsourcing is low, farmers may compensate for their lack of labor inputs by increasing the amount of fertilizer applied, leading to more intense fertilizer application. As the degree of outsourcing increases, the main body of fertilizer application gradually changes from farmers to outsourced service organizations, which have the advantage of professional staff and advanced equipment to apply fertilizer scientifically, helping to curb the intensity of chemical fertilizer application. The sample was divided into older and younger age groups for the regression according to the median age of the head of the household, and the results are presented in Table 6. The results showed that the inverted U-shaped relationship was significantly present in the younger age subgroups, while it was not significant in the older age subgroups. This finding suggests that the younger the age of the household head, the more beneficial the increase in outsourcing within the production chain is for reducing fertilizer application.

4. Discussion

The degree of farmers’ involvement in outsourcing the production chain and the amount of fertilizer used intensively in the plot was shown to have a strong inverted U-shaped relationship in this study, which is consistent with previous research [35]. As the practice of outsourcing becomes more prevalent, farmers may be inclined to increase their use of fertilizers as a result of risk aversion, resulting in a higher intensity of fertilizer application. As outsourcing increases, farmers rely more on outsourced service organizations for fertilizer application. The use of specialized services and advanced technologies can assist farmers in achieving more accurate fertilizer application, resulting in a decrease in the amount of chemical fertilizers used. This finding is a useful addition to the existing controversy in the literature. It indicates that improving the efficiency of the market for outsourced agricultural production chain services and raising the level of outsourcing by farmers is crucial to decreasing the amount of fertilizer used and attaining sustainable agricultural development. Furthermore, the distribution of labor significantly influences the extent of outsourcing and the level of fertilizer usage. This study explains the mechanism of the impact of production chain outsourcing on the fertilizer application intensity of farm households from the perspective of labor allocation, i.e., the intensity of farm labor input, and unifies the factors of non-farm employment and part-time work into the analytical framework to more comprehensively reflect the impact of labor force changes on fertilizer input factors. Labor and fertilizer have a significant substitution connection as crucial factors of production. As the level of outsourcing rises, the impact of the “crowding out” effect on agricultural labor also grows. In order to mitigate the potential negative impact of lower crop yields and harvests, this shift in labor distribution can result in a rise in the level of fertilizer usage. However, as the degree of outsourcing increases further and the main fertilizer applicator shifts from the farmer to an outsourced service organization, the introduction of specialized services may help farmers to achieve more efficient agricultural production, dampening the increase in fertilizer application intensity to some extent. Finally, plot size is an important factor influencing both the adoption of outsourcing production chains and fertilizer application by farmers. The expansion of plot size is conducive to the vertical division of labor in agricultural production among farmers, increasing the degree of outsourcing and specialisation, enhancing the substitution effect, and reducing the intensity of fertilizer application. At the same time, under the established degree of outsourcing, the larger the plot size, the more conducive it is to optimizing the allocation of resources, improving the efficiency of production, giving the full scale effect, and lowering the cost of production, which in turn reduces the inputs of chemical fertilizers. This finding is consistent with the studies of Zhang et al. [47] and Chen et al. [12], which suggests that the inhibitory effect of outsourcing the production chain on the intensity of fertilizer application is more pronounced on larger plots. The significance of this finding coincides with China’s current policy of “merging large fields with small ones”, and foreshadows the need for active land consolidation, which will help to reduce the degree of land fragmentation.

This study not only enriches the research on the relationship between the market for outsourced agricultural production services and the intensity of fertilizer application, but also provides useful policy recommendations for sustainable agricultural development. By optimizing the market for agricultural production outsourcing services, increasing the degree of outsourcing production processes, rationally allocating agricultural labor, and encouraging farmers to purchase agricultural insurance, the intensity of chemical fertilizer application can be effectively reduced, and green, efficient, and sustainable development in agricultural production can be achieved.

This paper also has some limitations. For example, it mainly focuses on rice and maize crops and does not consider other factor inputs such as pesticides. The geographical area of the study is mainly the Jiangsu Province, which can be further expanded to the whole country in the future. In the future, whether there is an inverted U-shaped relationship between the degree of outsourcing in the production chain and other factors of production such as pesticides could be explored.

5. Conclusions

This paper empirically tested the impact of production outsourcing on farmers’ fertilizer intensity on farm plots using the instrumental variables method with non-planar panel data from the 2020–2022 China Land Economy Survey (CLES). The conclusions are as follows. (1) There is a significant inverted U-shaped relationship between the degree of agricultural production outsourcing and the intensity of fertilizer application in the plots, and the results hold after a series of robustness tests. (2) Mechanistic analyses showed that the degree of outsourcing affected the intensity of fertilizer application in plots, mainly through labor allocation. Specifically, the degree of outsourcing had a facilitating and then inhibiting effect on fertilizer application intensity as the labor input intensity decreased. Furthermore, plot size plays a moderating role; i.e., larger plot size makes the inverted U-shaped curve flatter and achieves lower fertilizer application intensity with the same degree of outsourcing. At the same time, this facilitates an increase in the degree of the vertical division of labor, which shifts the inflection point of the inverted U-shaped curve to the left and achieves an inhibitory effect on the intensity of fertilizer application at a lower degree of outsourcing. (3) Participation in the outsourcing of technology-intensive production processes is conducive to reducing the intensity of fertilizer application on plots, and increasing the degree of outsourcing of production processes on large plot sizes and with young heads of households is conducive to reducing the intensity of fertilizer application.