Why China’s AMS Market Is Difficult to Develop Sustainably: Evidence from the North China Plain

Abstract

The agricultural machinery service (AMS) market enables China to rapidly realize agricultural mechanization on the basis of smallholder farmers. Academics generally believe that China has created a model of agricultural mechanization that matches small-scale farmers with large-scale machinery. However, in recent years, the transformation of the AMS market from prosperity to decline has been under-analyzed. This paper used farm and AMS supplier survey data from the North China Plain to estimate the links between acquaintance transactions, segmentation of the AMS market and losses in terms of transaction efficiency. We found that although the AMS from acquaintances is more expensive and less efficient than from non-acquaintances, farmers still buy from acquaintances because this reduces the transaction costs of the AMS. However, acquaintance transactions have led to the segmentation of the AMS market, reducing the transaction efficiency of the AMS. The AMS market has gradually changed from open to closed in China. The Chinese government should determine how to adjust policies to promote the sustainable development of the AMS market.

1. Introduction

The traditional view holds that it is difficult to realize agricultural mechanization with smallholder farmers, and large-scale land is the basic premise for realizing agricultural mechanization [1]. Countries such as the United States and Canada, which are dominated by large-scale farmers, quickly realized agricultural mechanization after World War II. However, the progress of agricultural mechanization in Asian countries dominated by smallholder farmers was slow. Scholars gradually shifted their research perspective to the relationship between land scale and agricultural mechanization. Different types of machinery need to match different processes in agricultural production, but each process has a short duration [2]. Agricultural machinery has the characteristics of high investment cost, low frequency of use and long return period. Therefore, smallholder farmers lack the willingness and ability to invest in agricultural machinery [3]. This is the reason for the slow progress of agricultural mechanization in Asian countries dominated by smallholder farmers. Consequently, if Asian countries wish to realize agricultural mechanization, they need to promote land transfer and large-scale operation [4].

However, China has quickly realized the popularization of large machinery with smallholder farmers. Practice in China breaks the traditional judgment that it is difficult to realize agricultural mechanization with smallholder farmers [5,6]. This is not only different from the matching mode between large-scale farmers and large machinery in the United States and Canada, but also different from the matching mode between smallholder farmers and small machinery in Japan and Southeast Asia. China has embarked on a road of agricultural mechanization with Chinese characteristics [7,8]. The key to the rapid realization of agricultural mechanization in China is the market of the agricultural machinery service (AMS) represented by cross-regional service [9]. The data released by the Ministry of Agriculture and Rural Affairs of China showed that the comprehensive mechanization rate of wheat and corn are 97.19% and 89.76% in 2020, respectively. Chinese practice shows that agricultural mechanization and smallholder farmers are not mutually exclusive.

The agricultural mechanization model in China has attracted the attention of scholars who hope to study the factors that promote these positive results. First, by purchasing the AMS, smallholder farmers are involved in the division of labor in agricultural machinery operations [10]. Using agricultural machinery without bearing the investment cost is a rational decision to minimize costs for smallholder farmers. Therefore, the development of the AMS market is the result of the division of labor. Second, the development of the AMS market is the result of government support. In the early stage of AMS development, the government introduced a cross-regional service fee reduction, agricultural machinery purchase subsidies, agricultural machinery operation subsidies, etc., which greatly promoted the popularization and application of the AMS [11]. Agricultural socialized services are classified into different categories, such as agricultural infrastructure, production sources, mechanization, and informatization [12]. The AMS is the core component of agricultural socialized services. Therefore, the Chinese government’s policies aimed at promoting agricultural socialized services also promoted AMS development. Third, the AMS is the result of the imbalance in the quantity of regional agricultural machinery. Areas with excess quantity will take the initiative to export the AMS to areas with insufficient quantity [5]. The positive impacts of the AMS on Chinese agricultural growth has been appreciated by scholars, which include the improvement of production efficiency, the growth of household income, and land transfer [10,13,14].

However, with the increase in the amount of agricultural machinery, the AMS market competition intensified, and the profit margin of the AMS shrunk. Consequently, the AMS market capacity began to decline. First, according to the China Agricultural Machinery Industry Yearbook (CAMIY), the cross-regional harvested area of wheat, rice and corn peaked in 2013, but showed a downward trend after 2013. The cross-regional harvested area of wheat decreased from 14.426 million hectares in 2013 to 6.035 million hectares in 2019, a decrease of 58.17%. The cross-regional harvested area of rice decreased from 7.696 million hectares in 2013 to 4.473 million hectares in 2019, a decrease of 41.88%. The cross-regional harvested area of corn decreased from 3.251 million hectares in 2013 to 2.57 million hectares in 2019, a decrease of 20.95%. Second, the AMS revenue continued to decline after reaching its peak in 2015, and the profit margin of agricultural mechanization in 2019 was nearly five percentage points lower than the peak in 2015. Third, the number of AMS practitioners decreased from a peak of 55.715 million in 2015 to 53.412 million in 2019. The number of AMS professionals decreased year by year from 5.2508 million in 2014 to 4.243 million in 2019. In response to the above, scholars generally believe that China has found an ideal model to realize agricultural mechanization, that is, the AMS market represented by cross-regional services. However, the reality is that the AMS, especially cross-regional services, is rapidly declining. Why is there a conflict between reality and theory? How should the rapid decline in cross-regional services be explained? The existing literature cannot answer the above questions.

Our survey of the AMS in the North China Plain found that farmers are more inclined to buy the AMS from acquaintances, despite it being more expensive and inefficient than that from non-acquaintances. Among the 795 surveyed farmers, 535 chose to purchase the AMS from acquaintances, accounting for 67.30% of the total sample. The price of the AMS from acquaintances in six agricultural machinery operation links were all higher than those from non-acquaintances. For example, the average prices of the AMS from acquaintances of machine farming, machine sowing, and machine harvesting for wheat were 56.62 yuan/mu, 18.15 yuan/mu and 57.90 yuan/mu, respectively. However, the average prices of the AMS for the same service from non-acquaintances were 49.17 yuan/mu, 14.39 yuan/mu and 51.81 yuan/mu, respectively. Meanwhile, the performance of agricultural machinery from non-acquaintances was more efficient than those from acquaintances. For example, the average horsepower of tractor, wheat combine, and corn combine from acquaintances was 107.25 horsepower, 122.37 horsepower, and 145.16 horsepower, respectively. However, the average horsepower of the same agricultural machinery from non-acquaintances was 139.08 horsepower, 158.60 horsepower, and 170.28 horsepower, respectively. Although the price of AMS from acquaintances was more expensive and inefficient, farmers still chose to buy it, which is not consistent with common sense. Farmers choose the AMS from acquaintances, and as a result, the AMS market is divided into small-scale closed markets with acquaintances as the margin. This makes it possible to explain the rapid decrease in cross-regional services. Thus, if we can explain why farmers abandon the lower-priced AMS from non-acquaintances and buy the higher-priced AMS from acquaintances, we can determine why the cross-regional service has decreased rapidly.

In rural China, a village is a typical acquaintance society. The farmers in villages live close to each other and are familiar with each other, and most of them have blood or clan relations. Acquaintances in the village help each other in the process of agricultural production and consumption, forming acquaintance networks, and this plays an important role in maintaining the order of production and consumption in rural China [15]. However, transactions between acquaintances may not be efficient and can adversely affect agricultural production [16]. Clearly, transacting with acquaintances has an important impact on the AMS market, and may lead to fragmentation of the AMS market. However, to our best knowledge, few studies have analyzed the impact of transactions between acquaintances on the AMS market, and discussed AMS market segmentation and the resultant trading efficiency losses. Moreover, segmentation of the AMS market has important implications for future policy formulation and implementation.

The evaluation in this paper was three-fold. First, we investigated the relationship between the acquaintance-transaction AMS and opportunistic behaviors of the AMS suppliers. The purpose was to discuss why farmers buy more expensive and inefficient AMS from acquaintances instead of cheaper and more efficient AMS from non-acquaintances. Second, we assessed the links between transactions of the AMS inside acquaintance networks and the service radius of AMS suppliers. Third, we further discussed the impact of a transaction between acquaintances on AMS trading efficiency. This not only helps to explain why farmers buy a more expensive AMS from acquaintances, but also helps to explore the causes of AMS market segmentation and its efficiency loss.

This paper contributes to the extant literature in two ways: First, to the best of our knowledge, prior studies have paid little attention to the acquaintance transactions of the AMS market and it’s segmentation. This paper addresses this gap and documents systematic evidence on whether and how acquaintance transactions can influence the segmentation of the AMS market. Second, extant studies have primarily focused on the importance of the AMS market, but provided little evidence on the sustainable development of China’s AMS market. This paper is the first to shed light on the development trend in the AMS market and the role of acquaintance transactions in the development of the AMS market.

2. Background and Theoretical Framework

2.1. Background in China

Prior to 1978, the main body of agricultural mechanization in China was divided into two parts: one included the state-owned tractor station, and the other included the collective-owned agricultural machinery teams. By 1979, China had 879 state-owned tractor stations, 31,300 commune tractor teams and 1,060,800 agricultural machinery teams [17]. These agricultural machinery organizations provided a free AMS for farmers. In 1978, the reform of the rural land system rapidly developed in China, which destroyed the state-owned and collective-owned agricultural machinery system, and led to a short-term decline in the rate of agricultural mechanization from 1980 to 1982. In 1982, the Chinese government established the agricultural mechanization service station (AMSS) to coordinate the supply of the AMS, which curbed the decline in the agricultural mechanization rate. Document No.1 in 1983 implemented restrictions on the private purchase of agricultural machinery for the first time. Meanwhile, the government contracted state-owned and collectively owned agricultural machinery to farmers, who were responsible for their own profits and losses. Obviously, the Chinese government has played an important role at initial stages of the AMS market, which is similar to the situations in Africa and Southeast Asia today [18,19].

After 1985, the rural labor force of China began to transfer to cities on a large scale, which led to a rapid increase in labor price, and the AMS suppliers began to appear in large quantities. Except for a small decrease after 2015, the number of AMS employees has increased rapidly, which is similar to the changing trend in the number of farmers providing the AMS. In 2004, the Chinese government introduced the subsidy policy for purchasing agricultural machinery. According to the data released by the Ministry of Agriculture and Rural Affairs of China, the amount of subsidies for purchasing agricultural machinery increased from 75 million yuan in 2004 to 23.755 billion yuan in 2015. There is no doubt that the subsidy policy has promoted the rapid development of the AMS market in China. In addition, cross-regional services have increased rapidly and become the most important component of the AMS.

However, in recent years, the development speed of the AMS market in China has obviously slowed down and even started to shrink. First, cross-regional services began to decrease rapidly after reaching their peak in 2013. In the introduction, we introduced the changes in cross regional services data. Second, the number of AMS suppliers decreased year by year after reaching its peak in 2015. According to the CAMIY, the number of AMS employees increased from 38.48 million in 2004 to 55.72 million in 2015, and then decreased year by year to 53.41 million in 2019. Third, AMS income decreased from 446.76 billion yuan in 2013 to 353.47 billion yuan in 2019. The AMS has been gradually decreasing, and cross-regional services have experienced a significant reduction. Where the AMS market will go is still worthy of in-depth study.

2.2. Theoretical Framework

A theoretical framework for the analysis of the influence of transactions between acquaintances in terms of the transaction cost of the AMS, service radius of AMS suppliers and service benefits of AMS suppliers is presented in this section. Although many studies have suggested that transactions between acquaintances are an important approach for transactions of production and consumer goods in rural China, the literature has paid little attention to the impact of transactions between acquaintances on the AMS market. Agricultural production processes have a long time span and labor supervision is difficult; therefore, agricultural production is unsuitable for division of labor and outsourcing [20]. Farmers adopt the AMS and participate in the division of the labor economy, which is a form of outsourcing and hiring labor to operate agricultural machinery. However, any form of outsourcing will be subject to the difficulty of agricultural labor supervision, especially transaction costs [21]. Therefore, tracing the effect of transaction costs on the transaction between acquaintances will help to confirm the causes of the AMS market segmentation.

Farmers who adopt the AMS are involved in the division of the labor economy by purchasing the services from third-party suppliers. Division of labor not only brings about increasing returns, but also brings about an increase in transaction costs. The transaction cost of the AMS is determined by the basic characteristic of agricultural production, that is, the labor time is inconsistent with the crop growth time, thus the input and output cannot form a one-to-one corresponding relationship. This characteristic determines that (i) farmers cannot assess the operation quality of the AMS according to the yield, and (ii) it is difficult for farmers to assess the quality of AMS operations on site, because observing the depth of cultivation, sowing density and harvest loss is almost impossible. Therefore, information asymmetry of the quality of agricultural machinery operation between farmers and AMS suppliers is very serious. In order to save fuel consumption, and reduce the operation time and depreciation of machinery, AMS suppliers provide low-quality services, for example, shallow tillage, uneven sowing, and excessive harvest losses. Therefore, adopting the AMS involves a high labor supervision cost, which is the endogenous transaction cost. Moreover, the AMS needs to conform to the farming season. Searching for AMS suppliers may run the risk of delaying the farming season, which will reduce production [10]. Therefore, farmers who adopt the AMS face the cost of searching for information, which is the exogenous transaction cost. Both exogenous and endogenous transaction costs will have an impact on the behaviors of both parties.

First, the transaction between acquaintances is likely to help to reduce the transaction cost of the AMS. Farmers only reach oral agreements with AMS suppliers, and it is costly and unrealistic to sign perfect contracts. If choosing self-service, farms need to buy agricultural machinery, which is not economical. Farmers tend to use the reputation of the acquaintance society to reduce the transaction cost of the AMS. Meanwhile, the AMS from acquaintances can inhibit AMS suppliers from exhibiting opportunistic behavior. Although the probability of being discovered exhibiting opportunistic behavior is very low, once discovered by farmers, it will affect their reputation, which may exclude them from acquaintance network relationships. This will cause the AMS suppliers to lose all potential customers. On the contrary, if farmers buy the AMS from non-acquaintances, both parties have only one chance, and the transaction cost borne by the farmer will be higher.

Second, a transaction between acquaintances is likely to lead to fragmentation of the AMS market. In order to reduce the transaction cost of the AMS, farmers are more inclined to buy the AMS from acquaintances. If farmers generally choose to buy the AMS from acquaintances in their villages and towns, the AMS from non-acquaintances across counties and provinces (cross-regional services) will be excluded from the market. Therefore, the service radius of AMS suppliers consequently shrinks from cross-province and cross-county services (cross-regional services) to the villages or towns where they are located, which means that AMS suppliers can only provide services to acquaintances. In the case of insufficient agricultural machinery, farmers will not care about the transaction cost of a non-acquaintance AMS, but in the case of sufficient agricultural machinery, the role of acquaintances in saving transaction costs will be emphasized.

Third, a transaction between acquaintances is likely to reduce the service effectiveness of AMS providers. The AMS from non-acquaintances has the characteristics of a large service radius, wide service range, large service area, and many farmers being covered. However, the service objects of the AMS from acquaintances are fixed, and the service scope of AMS providers is limited to the acquaintance relationship network. Therefore, a transaction between acquaintances reduces the service area of the AMS suppliers while reducing the service radius. Although the price of the AMS from acquaintances is higher than from non-acquaintance, the reduction in the service area will reduce the service income and profit of the AMS suppliers. This means that low-price, high-efficiency and cross-regional services are replaced by a high-price and low-efficiency AMS from acquaintances, which causes AMS market segmentation, and leads to AMS resource misallocation and loss of efficiency.

3. Materials and Methods

3.1. Data Sources

In this paper, the data used for statistical analysis came from a farm survey conducted in the North China Plain from August 2019 to January 2020. The surveyed provinces included Henan, Hebei, and Shandong province, which are the core components of the North China Plain. Household-level data on agricultural production, and supplier-level data on the AMS were collected. This survey focused on studying farmers’ adoption of the AMS, and the service behaviors of AMS suppliers.

In terms of the survey methodology, we used stratified random sampling to select household farms and AMS suppliers. First, eight counties were randomly selected from Henan, Hebei and Shandong, based on their geographic location, grain sowing acreage, economic level, etc. The counties included Sui, Xingyang, Wuqiang, Jing, Gaocheng, Guan, Wenshang, and Yucheng. Second, we divided all towns in each county into three groups based on their economic level, and one town was then randomly selected from each group. Third, we randomly selected 15 suppliers from the AMS supplier list of names provided by the town government. Fourth, the administrative villages in each town were divided into two groups based on the economic level, and one village was randomly selected from each group. Fifth, according to the number of farmers, 15–20 farm households were randomly selected from each village. We convened the heads of the farm household and the AMS suppliers with the village committee and town government, respectively, and collected household-level and supplier-level data through face to face interviews. The interviewer was a student at our school. It should be noted that the AMS suppliers were entities engaged in the AMS supply services, e.g., agricultural machinery operators and the agricultural machinery service cooperative. Considering that the agricultural machinery service cooperatives were composed of agricultural machinery operators, we chose agricultural machinery operators as the survey objects.

The survey had a total of 840 household-level observations and 360 supplier-level observations. Since 19 farm household responses had missing values, 26 farm households did not plant grain in 2019, 12 AMS suppliers responses had missing values, and 10 AMS suppliers quit the AMS market in 2019, the data for these farms and AMS suppliers were removed. Therefore, a sample covering 795 grain farmers and 338 AMS suppliers was used in this paper.

3.2. Variables

The dependent variables in this study were transaction costs, service radius, and service benefit. The measurement of transaction costs has been controversial. Every transaction incurs a transaction cost, but transaction costs are not standardized and the law of one price does not apply. The ordinal comparison method was used to measure the relative size of transaction costs between different individuals [22]. Therefore, farmers’ perception of the AMS quality was used as the endogenous transaction cost of the AMS, and farmers’ perception of the timeliness for the AMS was used as the exogenous transaction cost of the AMS. We used the farthest service area of the AMS suppliers in 2019 to represent the service radius of the AMS suppliers. The working capacity and efficiency of different power machinery varies greatly; therefore, we used the service area per horsepower and service profit in 2019 to express the service efficiency of the AMS suppliers.

A transaction between acquaintances was the main independent variable. First, in the analysis of the impact of the transaction between acquaintances on transaction costs, we focused on evaluating whether farmers’ choice of the AMS from acquaintances can help save AMS transaction costs, so we used transaction partner type to represent acquaintance relationships. Farmers were asked if the AMS came from an acquaintance. The acquaintances here included AMS suppliers who were familiar with the farmer, lived near to the farmer, or had blood ties with the farmer. Second, in the analysis of the impact of the transaction between acquaintances on service radius and service benefit, we focused on evaluating whether transaction between acquaintances could lead to the shrinkage of the service radius and a decrease in efficiency of AMS suppliers. Thus, we used the proportion of the acquaintance service area of the AMS suppliers to the total service area to represent the acquaintance relationship.

In the analysis of the impact of the transaction between acquaintances on transaction costs, other variables were controlled. It is known that signing the transaction contract can improve the stability of transaction and reduce opportunistic behavior on both sides of the transaction. We therefore used indicators of a transaction contract, such as a dummy transaction contract and dummy transaction agreement to represent the contractual relationship. Meanwhile, we controlled for the variables of land characteristics including land size and land fragmentation. In addition, some studies have proposed that the characteristics of both parties in a transaction are likely to affect the transaction costs [23,24]; thus, we controlled for the variables of the identity of the AMS supplier. Meanwhile, a person’s family may affect both the farmers’ production behavior and perception of the AMS transaction costs. Finally, considering the difference in the quantity of agricultural machinery in different regions, we controlled for the province using a dummy.

Table 1. Variable definitions and summary statistics of farmer sample.

Variable Name	Definition	Mean	SD
Endogenous transaction costs	Farmers’ perception of AMS quality, 1 = very good, 2 = good, 3 = central, 4 = bad, 5 = very bad	1.873	1.314
Exogenous transaction costs	Farmers’ perception of timeliness, 1 = very good, 2 = good, 3 = central, 4 = bad, 5 = very bad	1.786	1.290
Transaction partner type	1 if transaction partner type is acquaintance, and 0 otherwise	0.630	0.483
Transaction contract	1 if farmer signs contracts with AMS supplier, and 0 otherwise	0.023	0.149
Transaction agreement	1 if the farmer appoints the operation task with AMS supplier in advance, and 0 otherwise	0.565	0.496
Land size	Size of land operated by the farm household (mu a)	53.064	101.638
Land fragmentation	The proportion of land parcels less than 1 mu	0.116	0.357
Identity of AMS supplier	1 if AMS suppliers providing services to farmers belong to service organizations, and 0 otherwise	0.313	0.464
Age	Age of the household head (years)	56.435	10.481
Gender	1 if the household head is male, and 0 otherwise	0.860	0.347
Education	Education of the household head, 1 = uneducated, 2 = elementary school, 3 = junior high school, 4 = high school or secondary school, 5 = junior college, 6 = college and above	2.688	0.818
Health	Self-identified health of the household head,1 = very good, 2 = good, 3 = central, 4 = bad, 5 = very bad	4.152	0.908
Agricultural training	1 if the household head has received agricultural training, and 0 otherwise	0.343	0.475
Agricultural labor	Proportion of the labor force mainly engaged in farming in the farm household	0.330	0.288
Aging labor	Proportion of aging labor in farm household	0.577	0.416
Non-farm income	Proportion of non-farm income to total farm household income	0.436	0.365
Cooperative membership	1 if the farm household has cooperative membership, and 0 otherwise	0.133	0.340
Province dummy	Province dummy variable	—	—

Note: ^a. 1 mu = 1/15 ha.

describes the model variables and summary statistics.

In the analysis of the impact of a transaction between acquaintances on service radius and service benefit, other variables were controlled. The performance of agricultural machinery is on the basic premise that determines the operational ability [25,26]; therefore, we controlled for the variables of machine value, machine consumption, and machine subsidy. We controlled for the variables of the AMS working years and the AMS price. Meanwhile, the AMS suppliers join service organizations (e.g., agricultural machinery cooperatives), which may affect the AMS service radius and benefits by changing their business acquisitions. Personal and family characteristics may affect the AMS service radius and benefits [26]; therefore, we controlled for the variables of a person’s individual and family characteristics(e.g., age, gender, education, self-identified health, labor structure, family income). We also controlled for the province by using a dummy.

Table 2. Variable definitions and summary statistics of AMS supplier sample.

Variable Name	Definition	Mean	SD
Service radius	The farthest service area of the AMS supplier, 1 = village, 2 = town; 3 = county, 4 = province, 5 = other provinces	2.190	1.253
Service area	Service area of the AMS supplier pe horsepower in 2019 (mu)	5.748	4.232
Service profit a	Service profit of the AMS supplier in 2019 (ten thousand yuan)	0.989	0.742
Rate of acquaintance transactions	Proportion of the acquaintance service area to the total service area	0.617	0.290
Machine value	Original value of agricultural machinery(ten thousand yuan)	23.773	32.039
Machine consumption	Time that agricultural machinery has been used (year)	5.371	3.876
Machine subsidy	The proportion of agricultural machinery purchase subsidies in the price	0.261	0.092
AMS experience	Time as an AMS supplier (year)	7.344	6.338
ASM price b	ASM price per mu (yuan)	219.881	21.405
Organizational dummy	1 if the AMS supplier join an agricultural machinery organization, and 0 otherwise	0.389	0.488
Age	Age of the AMS supplier	49.935	7.681
Education	Education of the AMS supplier, 1 = uneducated, 2 = elementary School, 3 = junior high school, 4 = high school or secondary school, 5 = junior college, 6 = college and above	2.789	0.748
Health	Self-identified health of the AMS supplier, 1 = very good, 2 = good, 3 = central, 4 = bad, 5 = very bad	4.377	0.858
Agricultural machinery training	1 if the AMS supplier has received operation training of Agricultural machinery, and 0 otherwise	0.433	0.496
Province dummy	Province dummy variable	—	—

Notes: (^a) Service profit is equal to total service income minus the cost of machinery depreciation, the cost of machinery maintenance, and fuel expense. (^b) The price of AMS is the sum of the prices of six links for wheat and corn. Wheat includes machine farming, machine sowing, and machine harvesting. Corn includes machine sowing, machine harvesting, and straw returning. Wheat and corn are planted on alternate years in North China Plain, and corn is directly sown without needing arable land.

describes the model variables and summary statistics.

3.3. Estimation Strategy

First, we needed to assess whether a transaction between acquaintances could help reduce the transaction costs of adopting the AMS for farmers. We adopted the following multiple linear regression model:

$${Transactionc}_i={\alpha }_0+{\alpha }_1{Servicerad}_i+{\displaystyle \sum _{i=1}^n}{\beta }_j{X}_{ij}+{\epsilon }_i$$

(1)

where ${Transactionc}_i$ represents transaction costs, which consists of the endogenous and exogenous transaction costs. ${Servicerad}_i$ represents a transaction between acquaintances, which takes a value of 1 if the transaction partner type is acquaintance, and 0 otherwise. $X_{ij}$ are the control variables including transaction contract, transaction agreement, land size, etc. ${\alpha }_0$ is a constant term. ${\alpha }_1$ and ${\beta }_j$ are the regression coefficients to be estimated. ${\epsilon }_i$ is the random error, which is assumed to be independent and normally distributed.

Second, the relationship between transactions of the AMS inside acquaintance networks and the service radius of AMS suppliers was assessed. Meanwhile, we further assessed the relationship between transactions of the AMS inside acquaintance networks and the service efficiency of AMS suppliers. The models are specified, as follows:

$${Radius}_i={\gamma }_0+{\gamma }_1{Servicerat}_i+{\displaystyle \sum _{i=1}^n{\eta }_j{X}_{ij}}+{\epsilon }_i$$

(2)

$${Efficiency}_i={\kappa }_0+{\kappa }_1{Servicerat}_i+{\displaystyle \sum _{i=1}^n{\rho }_j{X}_{ij}}+{\epsilon }_i$$

(3)

where ${Radius}_i$ denotes the service radius of AMS suppliers. ${Efficiency}_i$ represents the service efficiency of AMS suppliers. ${Servicerat}_i$ represents the rate of acquaintance transactions, which is measured by the proportion of the acquaintance service area to the total service area. $X_{ij}$ are the control variables including machine value, machine consumption, machine subsidy, etc. ${\gamma }_0$ and ${\kappa }_0$ are constant terms. ${\gamma }_1$, ${\eta }_j$, ${\kappa }_1$ and ${\rho }_j$ are the regression coefficients to be estimated. ${\epsilon }_i$ is the random error, which is assumed to be independent and normally distributed.

Decision-making on acquaintance transactions is not random but may be intentional, which can lead to selection bias and inconsistent estimates. For example, if farmer households are affected by others in the same village and government action, then these households will have a higher probability of adopting the AMS from acquaintances. Both observed and unobserved factors may result in a self-selection bias. To overcome this problem, propensity score matching (PSM) was used to re-estimate Models (1), (2) and (3). The PSM is based on a counterfactual framework and can be used to construct a control group and a treatment group.

Moreover, omitting variables and reverse causality may be confounded by parameter estimations and, hence, the statistical inference. To some extent, the reverse causality of an acquaintance transaction can impact the transaction costs of adopting the AMS, the service radius and the efficiency of AMS suppliers. Specifically, farmers who perceive higher transaction costs are more inclined to adopt the AMS from acquaintances to reduce transaction costs. In order to cope with the competition in the AMS market, AMS suppliers with a reduced service radius and efficiency also tend to give priority to providing the AMS to acquaintances. Although we have tried our best to control variables that affect transaction costs and service efficiency, we may still have omitted other variables.

We used the instrumental variable (IV) method to solve the above endogenous problem. The development of the AMS market at the village level is utilized to serve as the IV for the transaction partner type in Model (1). Theoretically, if there is AMS supplier in the village, the farm household is more likely to adopt the AMS from acquaintances. A dummy variable for whether the village has an AMS supplier was used to measure the development of AMS markets at the village level. The development of AMS markets at the village level was assessed by the village-level variable, and the transaction costs of adopting the AMS as assessed by an individual-level variable, which means that IV was exogenous to the dependent variable.

Meanwhile, the situation of AMS suppliers’ acquaintance transactions at the town level was utilized to serve as the IV for the rate of acquaintance transactions in Model (2) and (3). If the average acquaintance-transaction rate of AMS suppliers at the town level is higher, the AMS suppliers may provide more AMS to acquaintances, and the rate of acquaintance transactions of the AMS supplier will be higher. The situation of AMS suppliers’ acquaintance transactions is a town-level variable, but the service radius and efficiency of AMS suppliers is an individual-level variable; therefore, the IV was exogenous to the dependent variable. The town-level variable could not directly affect the service radius and efficiency of AMS suppliers, except through the rate of acquaintance transactions.

The endogenous variable was the binary indicator in Model (1), the two-stage least squares method (2SLS) could not be used to estimate the model parameters. We used a conditional mixed process (CMP) instead. A CMP allows continuous, binary, and ordered endogenous variables.

4. Empirical Results and Discussion

4.1. Impacts of an Acquaintance Transaction on Transaction Costs of Adopting AMS

Table 3. Impacts of acquaintance transaction on transaction costs of adopting AMS.

Variable Name	Exogenous Transaction Costs			Endogenous Transaction Costs
	(1) Oprobit	(2) CMP Fist-Stage	(3) CMP Second-Stage	(4) Oprobit	(5) CMP Second-Stage
Transaction partner type	−0.479 *** (0.112)		−0.632 *** (0.226)	−0.796 *** (0.113)	−0.778 *** (0.220)
IV for transaction partner type		1.577 *** (0.126)
Transaction contract	−0.102 (0.318)	−0.076 (0.436)	−0.104 (0.318)	−0.212 (0.311)	−0.213 (0.311)
Transaction agreement	0.179 (0.110)	0.009 (0.128)	−0.194 * (0.112)	0.122 (0.111)	0.120 (0.113)
Land size	0.010 *** (0.001)	−0.006 *** (0.001)	0.010 *** (0.001)	0.013 *** (0.002)	0.013 *** (0.002)
Land fragmentation	0.762 *** (0.284)	0.261 (0.240)	0.740 *** (0.284)	0.035 (0.166)	0.034 (0.167)
Identity of AMS supplier	−0.077 (0.114)	−0.528 *** (0.124)	−0.087 (0.115)	−0.258 ** (0.118)	−0.257 ** (0.119)
Age	0.000 (0.006)	0.001 (0.007)	0.000 (0.006)	−0.006 (0.006)	−0.006 (0.006)
Gender	0.198 (0.160)	−0.033 (0.173)	0.190 (0.160)	0.015 (0.151)	0.016 (0.151)
Education	0.056 (0.072)	0.101 (0.084)	0.059 (0.072)	0.088 (0.073)	0.088 (0.073)
Health	0.230 *** (0.066)	−0.139 * (0.075)	0.221 *** (0.067)	−0.046 (0.063)	−0.045 (0.064)
Agricultural training	−0.138 (0.118)	0.270 * (0.142)	−0.135 (0.118)	0.034 (0.118)	0.033 (0.118)
Agricultural labor	0.750 *** (0.233)	0.208 (0.276)	0.760 *** (0.233)	0.150 (0.240)	0.147 (0.241)
Aging labor	−0.369 ** (0.151)	0.144 (0.173)	−0.354 ** (0.152)	−0.439 *** (0.150)	−0.441 *** (0.151)
Non−farm income	−0.904 *** (0.167)	−0.177 (0.189)	−0.902 *** (0.167)	−0.291 * (0.166)	−0.291 * (0.166)
Cooperative membership	−0.396 *** (0.130)	−0.155 (0.159)	−0.393 *** (0.130)	−0.294 ** (0.144)	−0.294 ** (0.144)
Province dummy	yes	yes	yes	yes	yes
Constant		0.002 (0.689)
Observations	795	795	795	795	795
Atanhrho_12			0.112 *** (0.037)		−0.013 ** (0.006)
Pseudo R2	0.328			0.312

Note: Robust standard errors are presented in parentheses; *, **, and *** indicate significance levels at 10%, 5%, and 1%, respectively.

reports the parameter estimates for Model (1). The results in Columns (1) and (4) are estimated using the ordered probit model (Oprobit). Column (2) presents results of the first-stage regression for the CMP. Columns (3) and (5) present results of the second-stage regression for the CMP. The results of the first-stage in the equation of endogenous transaction costs were similar to the equation of exogenous transaction costs; thus, we did not show the first-stage results in the equation of exogenous transaction costs. Atanhrho_12 test in Columns (3) and (5) indicate the presence of an endogenous problem. In the first-stage results, the coefficients on the IV passed the significance test at the 1% level, which indicates that IV is available.

Columns (1) and (3) both show that a transaction between acquaintances had a negative impact on the endogenous transaction costs of adopting AMS, and the coefficient on the type of transaction partner in Column (3) was smaller than that in Column (1), indicating that the negative impact of a transaction between acquaintances on the endogenous transaction costs was greater when the endogenous problem was considered. Columns (4) and (5) both show that transaction between acquaintances has a negative impact on the exogenous transaction costs of adopting the AMS, indicating that purchasing AMS from acquaintances can help farmer households reduce the exogenous transaction costs of adopting the AMS.

Farmers adopt the AMS from acquaintances with higher prices and lower efficiency, which is not economical. The results in Table 3 help us to understand this uneconomic behavior of farmer households. Clearly, acquaintances can help reduce transaction costs. Through the repeated game and the reputation within the acquaintance network, the opportunistic behaviors of both sides of the transaction is suppressed. By purchasing AMS from acquaintances, farmers can avoid delays in the farming season and improve the quality of agricultural machinery operations, which helps to improve production efficiency and planting profits. Although the AMS from acquaintances is more expensive, this is still not enough to offset the savings in transaction costs.

Among the other control variables, the larger the farmer’s land size, the higher the transaction cost of adopting the AMS. Elderly farmers lack the energy to monitor the behavior of AMS suppliers and are not sensitive to transaction costs. The higher the proportion of non-agricultural income, the lower the transaction cost of adopting AMS. Cooperatives help farmers get a more convenient AMS; therefore, joining cooperatives reduced the transaction cost of adopting the AMS. Land fragmentation increased the endogenous transaction costs of adopting AMS. Meanwhile, we found that AMS from service organizations had lower endogenous transaction costs than individual suppliers. The better the self-perceived physical health, the lower the endogenous transaction costs.

4.2. Do Acquaintance Transactions Cause Fragmentation of the AMS Market?

Table 4. Impacts of an acquaintance transaction on the service radius of AMS suppliers.

Variable Name	(1) Oprobit	(2) CMP First-Stage	(3) CMP Second-Stage
Rate of acquaintance transactions	−4.785 *** (0.316)		−3.583 *** (0.322)
IV for rate of acquaintance transactions		0.677 *** (0.016)
Machine value	0.006 ** (0.002)	0.000 * (0.000)	0.006 ** (0.002)
Machine consumption	−0.029 * (0.017)	−0.000 (0.001)	−0.029 (0.018)
Machine subsidy	−0.915 (0.723)	0.007 (0.048)	−0.887 (0.719)
AMS experience	0.003 (0.013)	−0.000 (0.001)	0.004 (0.013)
ASM price	0.005 (0.003)	−0.000 (0.000)	0.005 (0.003)
Organizational dummy	0.164 *** (0.036)	0.020 ** (0.009)	0.096 *** (0.035)
Age	−0.014 * (0.009)	−0.001 (0.001)	−0.014 * (0.009)
Education	−0.041 (0.095)	−0.016 ** (0.006)	−0.042 (0.096)
Health	0.037 * (0.018)	0.002 (0.005)	0.015 * (0.008)
Agricultural machinery training	0.158 (0.145)	0.001 (0.010)	0.160 (0.145)
Province dummy	yes	yes	yes
Constant		0.098 (0.065)
Observations	338	338	338
Atanhrho_12			−0.010** (0.005)
Pseudo R2	0.321

Note: Robust standard errors are presented in parentheses; *, **, and *** indicate significance levels at 10%, 5%, and 1%, respectively.

reports the estimates for Models (2). Columns (2) and (3) present results of the first-stage and second-stage regression for the CMP, respectively. Atanhrho_12 in Column (3) passed the significance test at the 5% level, which indicates the existence of endogenous transaction costs in Column (1). IV has a positive impact on the rate of acquaintance transactions at the 1% significance level in Column (2), which means that IV is available.

The results in Columns (1) and (3) indicate that the influence coefficient of acquaintance transactions on the service radius of AMS suppliers was negative at the 1% significance level. The coefficient on the rate of acquaintance transactions in column (3) was larger than that in Column (1), indicating that the negative impact had been reduced when the IV had a positive impact on the rate of acquaintance transactions is considered. The result of farmers purchasing the AMS from acquaintances is that the service scope of AMS suppliers is limited to the network of acquaintances. The results in Table 4 support this theoretical logic, and acquaintance transactions trigger the contraction of the service radius of AMS providers. This means that acquaintance transactions lead to segmentation of the AMS market. The national cross-regional service market is divided into a local, closed, small AMS market with the network of acquaintances as the boundary. With the increase in the number of agricultural machinery, cross-regional services have been gradually replaced by the AMS from small-radius acquaintances. This result helps to explain why the cross-regional services have declined rapidly in China from the perspective of AMS suppliers.

Among the other control variables, the larger the machine value, the larger the service radius of AMS suppliers. The service radius of small agricultural machinery is smaller than that of large agricultural machinery. The AMS suppliers who join an agricultural machinery organization have a larger service radius than individual AMS suppliers. The older the age, the smaller the service radius, and the poorer the health, the smaller the service radius, which is consistent with common sense. The other control variables have insignificant impacts on the service radius of AMS suppliers.

4.3. Impacts of Acquaintance Transaction on Transaction Efficiency

Table 5. Impacts of an acquaintance transaction on transaction efficiency.

Variable Name	Service Area		Service Profit
	(1) OLS	(2) CMP Second-Stage	(3) OLS	(4) CMP Second-Stage
Rate of acquaintance transactions	−11.380 *** (0.557)	−10.069 *** (0.527)	−2.019 *** (0.105)	−2.026 *** (0.091)
Machine value	0.001 * (0.000)	0.001 * (0.000)	0.002 *** (0.001)	0.002 ** (0.001)
Machine consumption	−0.017 (0.051)	−0.016 (0.039)	−0.004 (0.006)	−0.004 (0.007)
Machine subsidy	0.064 (1.657)	0.015 (1.570)	−0.088 (0.281)	−0.087 (0.270)
AMS experience	0.018 (0.031)	0.016 (0.028)	0.010 ** (0.005)	0.010 ** (0.005)
ASM price	0.019 *** (0.007)	0.020 *** (0.007)	0.001 (0.001)	0.001 (0.001)
Organizational dummy	0.347 *** (0.109)	0.341 *** (0.108)	0.048 (0.051)	0.048 (0.051)
Age	−0.040 * (0.023)	−0.041 ** (0.019)	−0.006 (0.004)	−0.006 * (0.003)
Education	0.208 (0.209)	0.197 (0.208)	0.006 (0.036)	0.006 (0.036)
Health	0.026 * (0.014)	0.016 * (0.008)	−0.001 (0.032)	−0.001 (0.030)
Agricultural machinery training	0.455 (0.350)	0.474 (0.319)	0.028 *** (0.009)	0.028 *** (0.008)
Province dummy	yes	yes	yes	yes
Constant	6.580 *** (2.394)	6.294 *** (2.149)	2.344 *** (0.381)	2.351 *** (0.370)
Observations	338	338	338	338
Atanhrho_12		−0.121 ** (0.058)		0.016 (0.058)
R2	0.630		0.649

Note: Robust standard errors are presented in parentheses; *, **, and *** indicate significance levels at 10%, 5%, and 1%, respectively.

reports the estimates for Model (3). The results in Columns (1) and (3) are estimated using the ordinary least squares method (OLS). Columns (2) and (4) present the results of the second-stage regression for the CMP. Results of the first-stage regression of Model (3) are similar to Column (2) in Table 4; therefore, we did not show the results of the first stage for the CMP in Table 5. Atanhrho_12 in Column (2) passed the significance test at the 5% level, which indicates the existence of endogenous transaction costs in Column (1). A tanhrho_12 in Column (4) is not significant; thus, there are no endogenous issues in Column (2).

The estimation results in Columns (1), (2) and (3) indicate that the influence coefficient of acquaintance transactions on the service area and profit of AMS suppliers were negative at the 1% significance level. The acquaintance transactions reduced the service area per horsepower of AMS suppliers, which leads to a decline in the utilization of agricultural machinery and an increase in the idle rate of agricultural machinery. It is known that the sharing of agricultural machinery among farmer households is considered to be efficient, and AMS suppliers are known for their efficient use of agricultural machinery [7,8]. However, the efficient use of such agricultural machinery does not seem to be sustainable. Meanwhile, acquaintance transactions also reduce service profit of AMS providers, which means that the profitability of AMS suppliers is constantly being challenged. The decrease in service area and profit indicates the declining efficiency of the AMS transactions.

Among the other control variables, the larger the machine value, the larger the service area and profit of AMS suppliers, which indicates the transaction efficiency of large agricultural machinery is higher than that of small agricultural machinery. The price of the AMS helps to increase the enthusiasm of AMS suppliers to provide services, thereby increasing the service area per horsepower. However, the AMS prices are often linked to fuel prices and have no significant impact on the profit of AMS suppliers. The older the age, the lower the transaction efficiency, and the worse the health, the lower the transaction efficiency. The other control variables have insignificant impacts on the service area and profit of AMS suppliers.

4.4. Robustness Check: Using the Propensity Score Method

For the robustness check of the results in Table 3, Table 4 and Table 5, farm households adopting the AMS from acquaintances were defined as the treatment group, those adopting AMS from non-acquaintances were defined as the control group. As the variable of acquaintance transaction rate in Models (2) and (3) was not a binary indicator, we constructed a treatment group of AMS suppliers whose acquaintance transactions rates were more than the average acquaintance transaction rate, and a control group of AMS suppliers whose acquaintance transactions rates were less than the average acquaintance transaction rate. Five matching strategies such as nearest neighbor (1 to 1), nearest neighbor (1 to 4), radius (caliper), kernel, and local linear regression were used to estimate the average treatment effect for the treatment groups (ATT).

Figure 1 shows the probability distributions of the propensity score between treatment and control groups before and after matching for Model (1). It can be judged from Figure 2 that the probability distribution of the propensity scores in the treatment group and the control group before matching was quite different. However, the probability distributions of propensity scores converged after matching, which indicates that the differences between the treatment and control groups narrowed significantly. Figure 2 shows the probability distributions of the propensity score between treatment and control groups before and after matching for Models (2) and (3). The difference between the treatment group and the control group was large before matching, but the difference was reduced after matching. Therefore, we can judge that the matching balance was ideal.

Table 6. Results of PSM estimation.

Matching Method	Farm Productivity	Treatment Group	Control Group	ATT
Nearest neighbor (1 to 1)	Exogenous transaction costs	1.391	1.578	−0.186 *** (0.062)
	Endogenous transaction costs	1.259	1.706	−0.447 *** (0.156)
	Service radius	1.442	3.459	−2.017 *** (0.107)
	Service area	3.302	9.936	−6.633 *** (0.634)
	Service profit	0.537	1.660	−1.123 *** (0.110)
Nearest neighbor (1 to 4)	Exogenous transaction costs	1.391	1.568	−0.177 *** (0.052)
	Endogenous transaction costs	1.259	1.679	−0.420 *** (0.148)
	Service radius	1.442	3.446	−2.004 *** (0.155)
	Service area	3.302	9.091	−5.789 *** (0.514)
	Service profit	0.537	1.540	−1.003 *** (0.085)
Radius (caliper)	Exogenous transaction costs	1.399	1.608	−0.209 *** (0.044)
	Endogenous transaction costs	1.261	1.660	−0.400 *** (0.140)
	Service radius	1.463	3.407	−1.945 *** (0.146)
	Service area	3.368	9.492	−6.124 *** (0.497)
	Service profit	0.551	1.601	−1.050 *** (0.080)
Kernel	Exogenous transaction costs	1.391	1.610	−0.219 *** (0.044)
	Endogenous transaction costs	1.259	1.693	−0.434 *** (0.140)
	Service radius	1.442	3.415	−1.973 *** (0.148)
	Service area	3.302	9.304	−6.002 *** (0.488)
	Service profit	0.537	1.576	−1.039 *** (0.082)
Local linear regression	Exogenous transaction costs	1.391	1.572	−0.181 *** (0.062)
	Endogenous transaction costs	1.259	1.697	−0.438 *** (0.156)
	Service radius	1.442	3.412	−1.970 *** (0.207)
	Service area	3.302	9.347	−6.045 *** (0.634)
	Service profit	0.537	1.552	−1.015 *** (0.110)

Note: Robust standard errors are presented in parentheses; *** indicate significance levels at 1%.

reports the results of PSM. Farm households purchasing the AMS from acquaintances are likely to have lower exogenous and endogenous transaction costs, as the ATTs are significantly negative in rows 1, 2, 6, 7, 11, 12, 16, 17, 21 and 22. The ATT in rows 3, 8, 13, 18 and 23 were significantly negative, indicating that the service radius of AMS suppliers whose rate of acquaintance transactions was higher than the average was narrower. AMS suppliers whose rate of acquaintance transactions was higher than the average had a lower service area per horsepower and service profit, as the ATT was significantly negative in rows 4, 5, 9, 10, 14, 15, 19, 20, 24 and 25. In general, we can conclude that self-selection bias or omitted variables do not confound our estimates.

5. Summary and Conclusions

Relying on the AMS market represented by cross-regional services, China has rapidly realized agriculture mechanization on the basis of smallholder farmers, which broke the traditional judgment of scholars that it is difficult for smallholder farmers to achieve mechanization. China has created a special mode of realizing agricultural mechanization. However, since 2013, the cross-regional service area has rapidly decreased, and the total revenue of the AMS market has continued to decline, which indicates that the AMS market of China is facing unprecedented challenges. Why do widely recognized cross-regional services decrease rapidly? Why is the AMS market changing from prosperity to decline? Existing research cannot provide an answer to these questions.

It is known that rural China is a society of acquaintances, and the acquaintance network is an important medium for the transaction of means of production and consumer goods. Our survey on the North China Plain found that although acquaintances’ AMS is more expensive and less efficient than that of non-acquaintances, farmers still choose to buy the AMS from acquaintances, which is not consistent with common sense. We realize that acquaintance transactions are likely to bring additional benefits to farmers and have an impact on the AMS market. Unfortunately, there are very few studies on this in the existing literature; in particular, there is a lack of analysis of AMS market segmentation, resource mismatch and efficiency loss. AMS market segmentation and efficiency loss determine whether the government needs to adjust its support policies for the AMS market.

In this study, we theoretically analyzed the impact of acquaintance transactions on the transaction costs of adopting the AMS, the service radius, and the profit of AMS suppliers. Then, we used survey data collected from 795 farm households and 338 AMS suppliers from the Henan, Hebei, and Shandong provinces in the North China Plain to empirically examine the impact of acquaintance transactions on the transaction costs of adopting the AMS, the service radius, and the profit of AMS suppliers. The results indicate that acquaintance transactions can help to reduce the transaction costs of adopting the AMS for farmers when the endogenous problem is dealt with. This is also the reason why the AMS from acquaintances is more expensive and less efficient, but farmers still buy it. We also found that acquaintance transactions lead to segmentation of the AMS market. The national cross-regional service market is divided into a local, closed, small AMS market with the network of acquaintances as the boundary, which helps to explain why the cross-regional service area is rapidly decreasing. Furthermore, acquaintance transactions reduce the use efficiency of agricultural machinery and transaction efficiency of the AMS, which causes the decline in service area and profit of AMS suppliers. Robustness checks using the PSM supports our findings.

Obviously, AMS transaction between acquaintance is a double-edged sword, which reduces the transaction cost of farmers adopting the AMS, but fragments the AMS market, reduces the transaction efficiency of the AMS, and hurts the profitability of AMS suppliers. The AMS market has gradually shifted from an open to closed market in China, and many market ingredients have been segmented by networks of acquaintances, which means that the scope of the division of labor is shrinking, and the phenomenon of a “reverse division of labor” appears. The consequent resource mismatch of the AMS and loss of transaction efficiency will also challenge the sustainable development of the AMS market.

Based on the research conclusion of this paper, we have tried to include information specially for policy makers. The Chinese government’s policies need to be adjusted in a timely manner. Instead of blindly pursuing the development of the AMS market, the government should address the segmentation of the AMS market and correct resource misallocation and efficiency loss. The most effective way would be to suppress the opportunistic behavior of AMS suppliers and reduce information asymmetry of the AMS. Moreover, China’s agricultural purchase subsidy policy also needs to be adjusted, especially to improve the flexibility of the subsidy policy. For surplus agricultural machinery, the government should encourage the exchange of old agricultural machinery for new. For insufficient agricultural machinery, the government should increase subsidies.

Our study, of course, has its limitations. We only examined the transaction costs in terms of farmers’ subjective perceptions. Due to the limitations of the data, we cannot measure the absolute value of transaction costs. Moreover, we limited our research to food crops; however, an AMS market for non-food crops in China has emerged. As a result, we cannot analyze the AMS market for non-food crops. In closing, we call for more detailed cross-crop studies given the vast differences in non-food crops and food crops. We believe that the cross-crop studies can achieve an important, useful, and better understanding of the development of the AMS market.