Does Farmland Transfer Lead to Non-Grain Production in Agriculture?—An Empirical Analysis Based on the Differentiation of Farmland Renting-In Objects

Abstract

The study aims to estimate different land leasing entities’ intentions and drivers to grow non-grain crops. In 2021, following a multistage sampling technique based on non-grain farmland, 264 farmers from the Zengcheng District of China were interviewed using a well-structured questionnaire based on the theory of planned behavior and transaction cost. The structural equation model was used to quantitatively reveal the influence mechanism of the non-grain use of the transferred farmland. The difference in the non-grain use of the transferred farmland was analyzed from the perspective of the differentiation of the renting entities. The results showed that the profit margin of non-grain and food crops, and the follow-up behavior of business entities, all promote the non-grain utilization of transferred farmland; however, the transaction costs of non-grain utilization and the endowment constraints of agricultural businesses inhibit the non-grain utilization of farmland. The non-grain crops in the suburbs are more profitable, and the transaction costs of the farmland leasing entities are low, so they tend to be grain-free; the rents of the farmland in the outer suburbs are low and can be operated on a large scale, and the leasing entities tend to be grain-oriented. Large-scale leasing entities tend to grow grain, while small-scale leasing entities tend to grow non-grain crops. In general, large-scale leasing entities in the outer suburbs have high transaction costs and low land rents and tend to be grain-oriented. The small-scale leasing entities in the suburbs are close to the market, the transaction costs are low, the rental price of farmland is high, and they are more inclined to grow non-grain crops. The non-grain utilization of the leased farmland should be treated separately, the supervision of the grain production capacity of the leased farmland should be carried out, and the rotation of grain and non-grain crops should be encouraged; the moderate scale operation in outer suburbs should be encouraged, and the construction of high-standard basic farmland for grain-growing farmland should be promoted.

1. Introduction

With the gradual withdrawal of the older generation of farmers from agricultural operations, the possibility of the second generation of farmers continuing to farm has been greatly reduced [1]; who will farm the land has become a major issue facing China’s agricultural development. To compensate for the productivity vacuum that has emerged in agricultural development, the government has put forward a policy orientation that actively promotes the transfer of farmland and promotes large-scale agricultural businesses [2,3]. Different from traditional smallholders’ inertia in growing grain, most farmland leasing businesses take profit maximization as their business goal. The non-grain use of farmland is often the result of rational decision-making by leasing businesses under established constraints, causing widespread concerns about food security in all walks of life. However, according to data from the National Bureau of Statistics, in 2020, the Chinese grain sown area is 116.8 million hectares, with a total output of 669.5 billion kilos. The grain sown area and output have increased year after year. Obviously, there is a certain contradiction between theory and statistical data. The question that arises from this is: does farmland transfer necessarily lead to non-grain production in agriculture? At present, there is a certain controversy on this issue in academic circles. There are mainly two views, as follows. First, farmland transfer promotes non-grain production in agriculture. This type of research is mainly based on cost-benefit theory. It is believed that when food prices are stable and agricultural production materials and farmland rents are steadily rising, the production benefits of most non-grain crops are higher than those of food crops, and based on the consideration of maximizing profit, agricultural businesses preferentially choose to grow non-grain crops. For example, Zhang et al. [4], based on case data, believe that the agricultural land management structure after farmland transfer shows obvious non-grain. Xu et al. [5] also believe that the development of the farmland transfer market will lead to a decline in the sown area of grain and affect food security. Second, farmland transfer does not necessarily promote non-grain production in agriculture, and whether farmland transfer promotes non-grain is strictly situationally dependent. For small-scale farmers, farmers with an abundant labor force or pure farmers, farmland transfer will aggravate non-grain. For rural households that have undergone labor migration or large-scale farmland transfer of households, under the background of rigid labor constraints and agricultural socialized services, the renting-in farmland will appear to be grain-oriented. For example, Qiu et al. [6] pointed out that land renting-in positively affects rice acreage, especially when there is less labor available for agriculture. Bi et al. [7] believe that the effect of scale expansion on planting structure is nonlinear; this process will be restricted by domestic labor endowment, and that is, there is a threshold effect. When the average scale of family labor operating in agriculture is less than the threshold scale, the expansion of the operation scale will encourage farmers to grow cash crops; otherwise, it will promote grain-oriented planting structures. Luo et al. [8] believe that small-scale farmland transfer households tend to have non-grain production. Whether renting-in farmland intensifies non-grain is affected by the abundance of labor and the level of social services. With the continuous transfer of rural labor and the development and improvement of the agricultural production service market, the planting structure will shift to grain-oriented. Zhang et al. [9] pointed out that due to the huge difference in labor productivity between non-grain crops and food crops, the economic scale of non-grain crops and food crops is greatly different. The larger the scale of land operation, the more business entities tend to have grain-oriented operation. Han et al. [10] believe that whether farmland transfer promotes non-grain has value differentiation characteristics. The increase in land rent price will increase the possibility of non-grain planting of farmland [11,12].

The non-grain utilization of farmland is essentially the decision-making result of the business entity to achieve the set goals under certain constraints. The core of the research on farmland non-grain should focus on two points: one is the target difference of business entities, and the other is the structure and form of constraints [13]. Therefore, the research on the non-grain of renting-in farmland cannot be limited to the main body of scale operation, nor can it stop at the assumption of profit maximization. It should include more subjects and have a more comprehensive analysis perspective. Generally, different business entities have different resources and different ways of using farmland. It is necessary to classify the farmland renting-in entities and take into account factors such as the endowment constraints and transaction costs of different entities. Otherwise, because the research object is not specific and the decision-making goal is too narrow, the conclusions and suggestions drawn are not oriented, and it is difficult to specifically guide the practice. The research of Kuang et al. [14], based on the perspective of peasant household differentiation, provides good enlightenment to this paper, but the above household classification does not take into account geographical differentiation. According to agricultural location theory put forward by Thunnen, there is a large difference between the non-grain behaviors of the farmland renting-in entities in the suburbs and outer suburbs.

This paper takes the Zengcheng District of Guangzhou, with a high land rental ratio, as an example and constructs a theoretical model of the non-grain utilization of farmland based on the theories of comparative benefit, endowment constraint, transaction cost, and planning behavior. Based on the spatial distribution of farmland, questionnaires were randomly collected in the field, and a structural equation model was used to quantitatively reveal the driving mechanism of the non-grain of renting-in farmland. The differences in influencing factors between farmland renting-in subjects from suburban and exurb areas and small-scale and large-scale farmland transfers were analyzed from two dimensions of region and scale, and specific suggestions were proposed to better manage the non-grain of renting-in farmland.

2. Theoretical Framework and Hypotheses

2.1. Theoretical Framework

The theory of planned behavior (TPB) was revised on the basis of the theory of reasoned action (TRA). The TPB has used in analytical framework that explicitly recognizes the importance of the influence of attitude and action, behavior of others, and the perception of a decision-maker [15]. A use of TPB has been found in agricultural sciences, for example, to identify the intention of farmers to adopt a practice of improved grassland [16], farmers’ conservation behavior [17]. The TPB was used in the study to predict the formation mechanism of non-grain renting-in farmland. However, the TPB was also criticized for only considering the influence of individual cognition on behavioral intention but not for increasing other external environmental variables, which limits the prediction ability of behavioral intention on individual behavior. In this article, behavioral attitudes are formed by behavioral beliefs and show the positive or negative evaluations of farmland renting-in entities toward the non-grain, and subjects evaluate the results mainly through the comparative profit of grain growing and non-grain growing. While the grain price has been stable for a long time, grain production costs have been rising steadily, and the low comparative profit of growing grain is undoubtedly the key factor that determines the attitude of farmland renting-in subjects [18]. Following this, in the study, the behavioral attitude was modified into comparative profit. Perceived behavior control is the understanding of the difficulty of non-grain planting of farmland renting-in entities. Endowment characteristics, such as the labor force and the technology of farmland renting-in subjects, can comprehensively reflect their perceived behavior control. The more abundant the labor force and the stronger the planting technology is, the stronger the tendency toward non-grainization is. Therefore, in this paper, perceived behavior control is modified into an endowment constraint. In addition, the non-grain farmland is also affected by transaction costs [19]. The factors that influence the non-grain of farmland include the contract cost of leasing farmland, the search cost of sales of non-grain products and the transportation cost [20]. Therefore, this paper introduces the transaction cost variable and adds the path of “transaction cost → behavioral intention” in the TPB. Finally, there are some differences in factors such as endowment constraints, transaction costs, and subjective norms among different entities, which are due to the different ways of using agricultural land and resources mastered by different farmland renting-in subjects. The type of farmland renting-in entities is introduced as a regulating variable, and the regulating effects of the renting-in subjects in suburban and exurban, small-scale, and large-scale farmland renting-in entities on each variable are analyzed from two dimensions of region and scale. The theoretical analysis framework of non-grain renting-in farmland, based on the improved TPB, is shown in Figure 1.

2.2. Hypotheses

(1) Effects of comparative profit on intention of non-grain. Comparative profit refers to the mutual comparison of the profits of grain growing and non-grain growing, which is a kind of relatively different condition in obtaining profit under the same resource input. In the decision-making of non-grain growing, the relatively low profit or even loss of grain production is the main internal cause of non-grain [21]. As a rational economic man, profit maximization is the main driver of farmland renting-in subjects growing non-grain crops. Generally, the higher the profit difference between non-grain crops and food crops per unit area, the greater the proportion of non-grain crops growing by farmland renting-in entities [9]. Hence, Hypothesis H1 is put forward:

H1: 

The comparative profit of farmland renting-in subjects about non-grain growing positively affects their intention of non-grain.

(2) Effects of subjective norms on intention of non-grain. Subjective norms refer to the social pressures perceived by an individual deciding whether to conduct a specific behavior and reflect the effects of significant others or teams on behavioral decision-making by the individual. Subjective norms of non-grain are the external pressures perceived by farmland renting-in entities during non-grain growing behavioral decision-making. Subjective norms are divided into injunctive norms and descriptive norms. Specifically, injunctive norms indicate that governmental staff guides farmland renting-in subjects in the behaviors of non-grain. The food-supporting policies of governments, including grain-growing subsidies and minimum purchase prices, all positively affect the non-grain intention of farmland renting-in entities. Moreover, descriptive norms mainly refer to the effects of family members, relatives, friends, and neighbors on decision-making. Family members and relatives are the most important social relations for farmland renting-in subjects, and their attitudes and opinions are the reference and basis for decision-making by farmland renting-in entities. Moreover, the non-grain growing behaviors of neighbors and friends may induce a herd effect among renting-in subjects, who will decide to use farmland not for growing grains. Hence, Hypothesis H2 is put forward:

H2: 

The subjective norms of farmland renting-in entities about non-grain growing positively affect their intention of non-grain.

(3) Effects of endowment constraints on intention of non-grain. Non-grain growing, as a direction of planting structure adjustment, actually results from the profit maximization pursued by operation subjects under specific constraints, and these constraints are mainly decided by the resources and capabilities of farmland renting-in entities conducting non-grain growing behaviors and are called the endowment constraints. When the farmland renting-in subjects feel more endowment constraints during non-grain growing behaviors, and when the resources and conditions are more unfavorable, the perceived difficulty is higher, and the intention of conducting non-grain growing behaviors is lower. Generally, when farmland renting-in entities are old-aged, incapable of farming, and more restricted by labor force shortages and labor force hiring, they are more willing to use farmland for purposes other than growing grains [22]. Technically, farmland renting-in subjects receive less technical training on non-grain growing and have low levels of mechanized non-grain crop planting, which restricts their intention and behaviors [8]. Thus, labor and skills are chosen as two endowment constraint variables in this paper, and Hypothesis H3 is proposed:

H3: 

The endowment constraints of farmland renting-in entities about non-grain negatively affect the intention of non-grain.

(4) Effects of transaction cost on intention of non-grain. Transaction cost refers to the expenditure paid by an affiliation or department to facilitate transaction affairs in the market. Farmland renting-in subjects involved in market transactions have to pay different transaction costs owing to varying degrees of external interventions. In particular, during land transfer transactions, contract costs are generated after negotiation, contract signing, implementation and maintenance [23]. However, due to restrictions by region, zones, market development level, and endowments of renting-in subjects, the renting-in entities to sell agricultural products have to pay certain searching and transportation costs [24]. When a farmland renting-in subject thinks the expected transaction costs for non-grain growing are excessively high and significantly weaken the net incomes of non-grain growing, the intention of non-grain is lowered and further resists non-grain growing behaviors. Hence, Hypothesis H4 is put forward:

H4: 

The transaction cost of non-grain growing paid by a farmland renting-in subject negatively affects the intention of non-grain.

(5) Effects of non-grain intention on non-grain growing behavior. Under the premise that the conditions are fully satisfied, behavior is a concrete action expression of intention, and behavioral intention directly determines behavior. Therefore, the non-grain intention of farmland renting-in subjects directly determines the non-grain growing behavior. The stronger the non-grain intention, the higher the possibility of farmland renting-in entities conducting non-grain growing behaviors. Hence, Hypothesis H5 is put forward:

H5: 

The non-grain intention of farmland renting-in subjects positively affects their non-grain growing behaviors.

(6) Regulating effect of type of renting-in entities on non-grain growing decision-making. First, the non-grain growing decision-making differs among subjects of different renting-in scales. Large-scale renting-in entities are rigidly restricted by the labor force and tend to select grain crops with high labor productivity. In comparison, small-scale renting-in subjects are less restricted by the labor force and can hardly expand scales to acquire more profits and thus tend to select non-grain crops with a high multi-cropping index or high planting benefits. Second, the non-grain growing decision-making largely differs among renting-in entities from different regions. Suburban renting-in subjects prefer non-grain growing because of high rent prices and short distance to markets. In contrast, exurban renting-in entities prefer grain-growing because of the large transaction costs of non-grain growing. Hence, Hypothesis H6 is put forward:

H6: 

The type of renting-in subject can regulate the decision-making of non-grain growing behaviors.

3. Study Area and Data Collection

3.1. Study Area

Zengcheng District is located in the Middle East of Guangzhou and the northeast corner of the Pearl River Delta. It is located at 113°32′–114°00′ E and 23°05′–23°37′ N, with an area of 1616.47 km² (Figure 2). Zengcheng is high in the north and low in the south, with complex and diverse terrain. The north and middle are dominated by hills and mountains, and the south is a plain. It has a subtropical marine monsoon climate and is warm and rainy with sufficient light and heat. It is rich in hydrological and biological resources and has superior natural endowment. By 2020, Zengcheng has a permanent resident population of 1.46 million, with an urbanization rate of 73.16%. The region achieved a GDP of CNY 14.84 billion, of which the total agricultural output value was CNY 1.52 billion, with a growth rate of 10.3%, ranking first in Guangzhou, China. The development trend of agricultural large-scale operation is obvious. A total of 162.93 km² of farmland was transferred in the region, accounting for 65.38% of the total farmland area. In 2020, the sown area of crops in Zengcheng was 565.40 km², of which the sown area of grain crops was 96.47 km² and the sown rate of non-grain crops reached 82.94%. The phenomenon of non-grain is relatively serious. Therefore, it is typical to select Zengcheng District as the study area.

3.2. Data Collection

The data in this article mainly include geospatial data, field survey data, and socioeconomic data. The geospatial data include 0.5 m resolution remote sensing images, 2018 land use status survey change data, and village-level administrative division maps of Zengcheng District, which come from the Natural Resources Bureau of Zengcheng District. The survey data come from first-hand data obtained by the research team entering the village in August 2021. The survey was carried out according to the distribution of farmland. First, the research base map was formed by using current farmland use data, administrative division data, and high-definition remote sensing image data; second, according to the principle of uniform distribution, the distribution of farmland was more concentrated in the 11 towns and streets of Zengcheng District. A total of 33 administrative villages were selected as the survey area (Figure 3), and the farmland concentrated distribution area of each administrative village was demarcated. A multistage sampling technique was used [25], to collect 264 questionnaires, and 10 invalid questionnaires were excluded, with an effective recovery of 96.2%. The socioeconomic data come from the 2020 Zengcheng District National Economic and Social Development Statistical Bulletin and Statistical Yearbook, the seventh national census bulletin of Zengcheng District and the information published on the website of government departments.

4. Methods

4.1. Model Building

According to the theoretical model, this paper includes many explanatory variables, such as non-grain willingness and non-grain behavior, which are difficult to analyze by the traditional multiple regression model. The structural equation model, which integrates the factor and path analysis methods, can effectively deal with the structural relations among the variables and overcome the collinearity among the independent variables and is often used in structured questionnaire regression analysis. The formula of the structural equation model is as follows:

X = Λ_x ξ + δ

(1)

Y = Λ_y η + ε

(2)

η = Bη + Γξ + γ

(3)

In the formula, X represents the observed variable of the exogenous latent variable ξ. Y represents the observed variable of the endogenous latent variable η; Λx and Λy represent the factor loading matrix of the exogenous variable and the endogenous variable, respectively; δ and ε are the residuals; B represents the path coefficient, reflecting the relationship between endogenous latent variables; Γ is the path coefficient, reflecting the influence of exogenous latent variables on the endogenous latent variables; and γ is the error term of the structural equation.

4.2. Variable Setting

Based on the theoretical model, this paper sets up six latent variables: endowment constraints, transaction costs, subjective norms, comparative benefits, non-grain willingness and non-grain behavior. Because the latent variable cannot be observed directly, it needs to be measured through the observed variable. Based on theoretical analysis and previous studies, this paper constructs a structured questionnaire with 32 observational variables in the form of a seven-level Likert scale with 1, 2, 3, 4, 5, 6 and 7 representing “completely disagree”, “disagree”, “slightly disagree”, “have no opinion”, “somewhat agree”, “agree” and “completely agree”, respectively. The farmland renting-in subjects were invited to evaluate the questionnaire scale one by one. The corresponding relationship of specific variables and the statistical description of the variables are shown in

Table 1. Variable settings and their interpretation.

Latent Variable	Measured Variable	Serial Number	Mean	Standard Deviation
Comparative benefit CB	Non-grain crops have higher incomes than grain crops [26]	CB1	4.89	0.7
	Non-grain crops are more profitable than grain crops	CB2	4.99	0.89
Subjective norm SN	Village collectives encourage non-grain farming	SN1	3.77	0.7
	Non-grain growers think that non-grain grows are good	SN2	4.44	0.73
	Family supports non-grain cultivation	SN3	4.68	1.12
Endowment constraint EC	Insufficient labor for non-grain planting [7]	EC1	3.77	0.85
	No physical energy to grow non-grain	EC2	3.69	0.86
	Employing farm labor is difficult and costly	EC3	4.16	0.73
	Unfamiliar with non-grain crop planting technology	EC4	3.56	0.98
	It is difficult to learn non-grain planting techniques	EC5	4.85	0.83
	Non-grain crops are less mechanized [9]	EC6	4.81	0.81
Transaction costs TC	Not familiar with the market sales of non-grain crops	TC1	3.61	1.03
	There is no regular sales channel	TC2	3.46	1.07
	Non-grain crops are harder to sell [19]	TC3	3.68	0.94
	Centralized leasing is difficult	TC4	3.61	0.82
	Immature form of contract signing	TC5	3.7	0.85
	Unstable contract period	TC6	3.75	0.78
	Long distance from town and high transportation cost	TC7	3.41	0.96
	Poor road conditions and inconvenient transportation	TC8	3.32	0.79
Non-food willingness NFW	Willing to rent land to grow non-grain crops [16]	NFW1	4.56	1.07
	Will lease farmland to grow non-grain crops	NFW2	4.58	1.17
	Willing to invest in the cultivation of non-grain crops	NFW3	4.61	1.15
Non-food behavior NFB	The leased land is mainly used for growing non-grain crops [17]	NFB1	4.45	1.58
	The leased land is rarely used for growing food crops	NFB2	4.22	1.6
	Ready to plant non-grain crops	NFB3	4.5	1.46

.

4.3. Sample Characteristics

The basic characteristics of the survey sample are shown in

Table 2. Basic characteristics of the sample.

Personal and Family Endowments	Selected Indicator	Frequency	Percentage
Age	≤30	1	0.4%
	31–40	19	7.5%
	41–50	128	50.4%
	51–60	84	33.0%
	≥61	22	8.7%
Education level	Elementary school and below	50	19.7%
	Junior high school	152	59.8%
	High school/Technical secondary school	35	13.8%
	Junior college/Higher vocational education	14	5.5%
	Bachelor degree and above	3	1.2%
Agricultural business entity	Family farm/Large-scale transfer subject	61	24.0%
	Land cooperative stock cooperative operating subject	1	0.4%
	Subject of agricultural business	21	8.3%
	Companies that provide social services	1	0.4%
	Small-scale transfer subject	170	66.9%
Main source of income	Agriculture	212	83.5%
	Agriculture-based part-time business	31	12.2%
	Nonagricultural part-time business	11	4.3%
	Non agriculture	0	0.0%

. The respondents were mainly middle-aged, with a low overall education level. Approximately 50.4% of the respondents were between 41–50 years old, and 76.2% of the respondents had a high school education or less. The main source of income of the respondents was agriculture, accounting for 83.3% of the total. The survey area can be divided into suburban villages and suburban villages. Suburban villages are villages far away from the urban area. In this paper, the villages within the central towns determined in the 14th five-year plan of Zengcheng District are designated suburban villages, accounting for 50% of the survey sample, and the villages outside the central towns are designated suburban villages, accounting for 50% of the survey sample. According to the scale of operation, the respondents can be divided into small-scale leasing entities and large-scale leasing entities. Small-scale leasing entities are mainly vegetable farmers from Guangxi Province, whose operating scale is generally less than 1 hectare. Large-scale leasing entities are large local growers or agricultural enterprises, with operating scales of more than 1 hectare, and the ratio of the two is approximately 1:1.

There are large differences in the non-grain willingness and behavior of different leasing entities. Generally, the non-grain tendency of the suburban and small-scale leasing entities is relatively strong. The average values of the non-grain willingness and behavior of the rental entities in the suburbs are 5.14 and 5.12, respectively, which are greater than those of the rental entities in the suburbs. The average values of the non-grain willingness and behavior of small-scale leasing entities are 4.89 and 5.02, respectively, which are greater than those of large-scale leasing entities. The statistical description of the non-grain willingness and behavior of different types of leasing entities is shown in

Table 3. Statistical description of non-grain willingness and non-grain behavior of different business entities.

Non-Grain Willingness and Behavior	Suburbs Renting-In Subject	Outer Suburbs Renting-In Subject	Large-Scale Renting-In Subject	Small-Scale Renting-In Subject
Non-grain willingness	5.14	4.03	3.73	4.89
Non-grain behavior	5.12	4.03	3.10	5.02

.

4.4. Reliability and Validity Test

(1) Trust level analysis. Reliability is used to test the internal consistency of the observed variables in the latent variables [27]. It is generally evaluated by Cronbach’s alpha (CA) and combined reliability. It is recommended that both are above 0.70 [28]. The reliability test results are shown in

Table 4. Confirmatory factor analysis (CFA) data.

Latent Variable	Cronbach’s Alpha	Rho	Combination Reliability
Subjective norm	0.820	0.867	0.891
Transaction costs	0.884	0.897	0.928
Comparative benefit	0.830	0.868	0.921
Endowment constraint	0.750	0.750	0.889
Non-food willingness	0.954	0.955	0.970
Non-food behavior	0.950	0.950	0.968

. Cronbach’s alpha and the combined reliability of each latent variable are between 0.75~0.95 and 0.89~0.97, respectively, which is higher than the acceptance standard of 0.70. It is proven that the above latent variable data have good reliability.

(2) Validity analysis. Validity can be divided into convergent validity and discriminative validity. Convergent validity is generally tested by the average variance extracted (AVE) and combined reliability (CR). The evaluation criteria are as follows: the single factor standard factor load should exceed 0.5 and reach a significant level; the composition reliability should exceed 0.8; and the AVE of each variable should be greater than 0.5 [29]. It can be seen from Table 4 that all the latent variables in this paper are up to the standard, indicating that the convergence validity is good. Discriminative validity refers to the degree of difference between a latent variable and other latent variables. The discriminative validity is discriminated by the standardized correlation coefficient between each factor and the square root of the AVE value. If the absolute value of the correlation coefficient is less than the root mean square of AVE and all AVEs are greater than 0.5, the discriminant validity is good [30]. As shown in

Table 5. Differential validity of the model.

Latent Variable	Subjective Norm	Transaction Costs	Comparative Benefit	Endowment Constraint	Non-Food Willingness	Non-Food Behavior
Subjective norm	0.857
Transaction cost	−0.777	0.901
Comparative benefit	0.729	−0.671	0.924
Endowment constraint	−0.723	0.738	−0.670	0.895
Non-food willingness	0.845	−0.828	0.727	−0.773	0.957
Non-food behavior	0.807	−0.802	0.678	−0.764	0.889	0.954

Note: The diagonal boldface is the root mean square of AVE, and the lower triangle is the Plzen correlation.

, the root mean square of the AVE of each latent variable is greater than the correlation coefficient between the latent variables, so it can be judged that this scale has good discriminative validity.

4.5. Fitness of Structural Model

To examine the evaluation of model, fit indices were used, particularly root mean squared error of approximation (RMSEA), the goodness of fit index (GFI), adjusted goodness of fit index (AGFI), and CMIN/DF [31]. As can be seen from

Table 6. Fitness of structural model.

Fit Indices	CMIN/DF	GFI	AGFI	CFI	RMSEA
Fit criteria	<3	>0.9	>0.8	>0.9	<0.05
Fit result	1.563	0.934	0.921	0.957	0.032
Fit or not	yes	yes	yes	yes	yes

, CMIN/DF < 3.00, GFI > 0.90, AGFI > 0.80, CFI > 0.90, RMSEA < 0.05, all fitting indices are in line with the general research standards, so this model has a good fit.

5. Empirical Results and Discussion

5.1. Analysis of the Impact Mechanism of Non-Grain Renting-In Farmland

According to the theoretical model, the results obtained after running the SmartPLS software are shown in Figure 4. The four exogenous variables of non-grain willingness passed the significance test at the 1% level, and R² is 0.752, which is greater than 0.670 [30], indicating that the four exogenous variables have a good degree of explanation for non-grain willingness. The path relationship between non-grain willingness and non-grain behavior also passed the significance test at the 1% level, with an R² of 0.761, indicating a good corresponding relationship between non-grain willingness and behavior.

(1) The impact of comparative benefits (CB) on non-grain willingness (NFW). Comparative benefits are positively correlated with non-grain willingness, with a path coefficient of 0.109, which is significant at the 1% level. This shows that under certain other conditions, the greater the comparative benefits of non-grain crops, the stronger the non-grain willingness of farmland leasers, so the verification of H1 is valid. Figure 4 shows that the factor loading (0.943) of non-grain crops is higher than that of food crops (CB2), indicating that comparative profit is the main factor driving the non-grain planting decision of farmland renting-in subjects [26]. Compared with ordinary smallholders, the main purpose of land renting-in entities is to maximize profits. The survey also found that without considering the rent of farmland, the average net profit per hectare for the two rice plants is approximately CNY 2982, while the average annual rent of farmland has reached CNY 2330 per hectare, and there is a gradual upward trend. The profit of growing rice is very low, and losses may even occur. Non-grain crops such as vegetables and fruits have a much higher profit per hectare than rice, which drives farmland renting-in subjects to plant non-grain crops [32].

(2) The impact of subjective norms (SN) on non-grain willingness (NFW). Subjective norms are positively correlated with non-grain willingness, with a path coefficient of 0.385, which is significant at the 1% level. It shows that under certain other conditions, the greater the pressure (supervision or demonstration) of the farmland renting-in subjects from outside, the higher the willingness of non-grain, so the verification of H2 is valid. The social pressure for the renting-in entities mainly comes from family relatives (0.906) and neighboring neighbors (0.906), and the village committee’s role in publicity and guidance is relatively small (0.748). The survey found that the following suit of the nongrain-based renting-in entities is obvious, and non-grain farmland in suburban villages is relatively concentrated, mostly with substitute farmers from Guangxi Province, who grow vegetables. The leading role of relatives, friends, and fellow villagers is obvious. However, the village committee did not intervene too much in the use of farmland. Therefore, compared with the village committee, the pressure from the neighborhood has a greater impact on the non-grain willingness and behavior of farmland renting-in entities.

(3) The impact of endowment constraints (EC) on non-grain willingness (NFW). Endowment constraints are negatively correlated with non-grain willingness, with a path coefficient of −0.191, which is significant at the 1% level. It shows that under certain other conditions, the greater the endowment constraint, the lower the non-grain willingness of farmland renting-in entities, so the verification of H3 is valid. Among them, the rigid constraint (0.783) of labor (EC1~EC3) is greater than the rigid constraint (0.709) of technology (EC4~EC6). In the decision-making of non-grain, the renting-in entities must not only consider maximizing profits but also consider whether they have the conditions to achieve non-grain management. Compared with the standardized production of food crops, the degree of mechanization and labor productivity of non-grain crops is lower, which means that it requires a larger labor input [6], resulting in greater constraints on the non-grain of renting-in entities that are in short supply of labor resources. In addition, non-grain planting has a certain technical threshold, which will also restrict the renting-in subjects.

(4) The impact of transaction costs (TC) on non-grain willingness (NFW). Transaction costs are negatively correlated with non-grain willingness, and the path coefficient is −0.313, which is significant at the 1% level. It shows that under certain other conditions, the greater the transaction cost, the lower the non-grain willingness of farmland renting-in entities, so the verification of H4 is valid. Transaction costs mainly include search costs, transportation costs, and contract costs [19]. Among them, search costs (TC1~TC3) have the largest average factor loading (0.871). The search cost can be characterized by the difficulty of selling agricultural products. The sales of non-grain products are mainly wholesale and retail, and the wholesale search cost is relatively small. However, due to the preservation and distribution problems in the wholesale market, the transaction window is slightly short, and the transaction price is slightly low. Retail search costs are relatively high, it takes much time to sell, and there is a risk that it will be difficult to sell all of them. The transportation cost (TC7~TC8) average factor loading (0.797) followed. The closer the distance to the market is, the better the conditions of roads are, especially the roads in the fields, the lower the transportation cost is, and the more conducive to non-grain production. The average factor loading (0.735) of contract costs (TC4~TC6) is the smallest. The stable contract period and mature contract form are conducive to large-scale investment in land by renting-in entities. However, because some small-scale renting-in entities did not sign a contract during the land rent process, they did not understand the contract form, the stability of the contract term [33], or the difficulty of signing a contract, resulting in a small factor loading.

5.2. The Moderating Effect of the Renting-In Subject Types

(1) The adjustment effect of renting-in entities based on scale differentiation. There are certain differences in the business objectives and constraints of the renting-in entities of different business scales. This article uses the farmland renting-in scale as a moderating variable and analyzes the difference in non-grain factors between large-scale and small-scale farmland renting-in entities. According to the research results of Bi et al. [7], this article combines the average business scale of ordinary farmers in descriptive statistics and uses 1 hectare as the substandard for small-scale and large-scale farmland renting-in entities [34]. Nested model comparison item test results show that the p values of the measurement coefficient model and the structural coefficient model grouped by scale are 0.365 and 0.654, respectively, which are both greater than 0.05, indicating that the measurement models of different scale groups have inter-group invariance and the structural models are equivalent. From the results of path coefficient estimation (

Table 7. Estimation results of path coefficients of large-scale and small-scale renting-in entities.

Path Relationship	Group 1 (Large-Scale Renting-In Entities)		Group 2 (Small-Scale Renting-In Entities)
	Estimate	C.R.	Estimate	C.R.
SN→NFW	0.296 ***	4.563	0.534 ***	5.642
TC→NFW	−0.332 ***	3.867	−0.207 **	7.589
BC→NFW	0.104 *	3.090	0.053 *	2.035
EC→NFW	−0.278 ***	2.344	−0.122 *	5.345
NFW→NFB	0.897 ***	7.473	0.72 ***	6.485

Note: ***, **, * represent significance levels of 1%, 5%, and 10%, respectively. C.R. is the critical ratio coefficient.

), the path coefficients of the large-scale group are all smaller than those of the small-scale group. It shows that subjective norms and comparative benefits have a greater positive impact on the small-scale group than on the large-scale group, while the negative impact of transaction costs and endowment constraints on the large-scale group is stronger than that of the small-scale group. Small-scale leasing entities are ordinary farmers who have a low level of education and easily follow the crowd. In addition, they are generally in the stage of increasing returns to scale and are more sensitive to operating profits; the difference in the comparative returns of non-grain crops and food crops has a more obvious impact on the non-grainization of small-scale leasing entities. Large-scale farmland renting-in entities have stronger risk sensitivity to farmland management and face greater management constraints. On the one hand, due to the large investment in farmland, market risks brought about by transaction costs drive farmland renting-in entities to avoid risks and choose to grow food crops. On the other hand, there is a ceiling for the optimal planting scale of non-grain crops, and large-scale farmland renting-in entities are more likely to be constrained by production input factors such as labor, technology, and capital. Compared with non-grain crops, food crops have higher labor productivity and can replace labor through machinery, thereby reducing labor costs and alleviating labor constraints. The government has adopted policies such as grain planting subsidies and agricultural material subsidies to reduce capital investment in farmland and lower the investment threshold for large-scale grain planting. Therefore, large-scale leasing entities are more inclined to grow grain, and transaction costs and endowment constraints have a stronger negative impact on non-grainization.

(2) The adjustment effect of the renting-in entities based on geographical differentiation. This article uses region as a moderating variable to analyze the difference in non-grain factors between the suburban and outer suburbs of farmland renting-in entities. Nested model comparison item test results show that the p values of the measurement coefficient model and the structural coefficient model grouped by region are 0.753 and 0.775, respectively, which are both greater than 0.05, indicating that the measurement models of different region groups have inter-group invariance and the structural models are equivalent. The grouping estimation results are shown in

Table 8. Estimation results of path coefficients in remote and suburban villages.

Path Relationship	Group 1 (Renting-In Entities in the Outer Suburbs)		Group 2 (Renting-In Entities in the Suburbs)
	Estimate	C.R.	Estimate	C.R.
SN→NFW	0.417 ***	9.374	0.452 ***	12.374
TC→NFW	−0.396 ***	10.374	−0.230 ***	8.374
BC→NFW	0.016 **	2.456	0.152	8.826
EC→NFW	−0.158 **	2.342	−0.162 **	2.038
NFW→NFB	0.894 ***	5.374	0.665 ***	7.374

Note: ***, **, represent significance levels of 1%, 5%, respectively. C.R. is the critical ratio coefficient.

. The effect of comparative benefit on non-grain willingness was not significant in the outer suburb group, and the other measurement paths passed the significance test below the level of 5%. Among them, subjective norms have a greater positive impact on the non-grain willingness of the suburban group, the non-grain willingness has a stronger impact on the non-grain behavior of the outer suburbs group, and the transaction cost and endowment constraints have a more negative impact on the non-grain willingness of the outer suburbs. The rent of farmland in the suburbs is relatively high, most of them are “substitute farmers” who come from Guangxi Province to grow vegetables, and the scale of operation is small. The leading role of fellow villagers is obvious, so the positive influence of subjective norms is greater. The rents of farmland in the outer suburbs are relatively low, far away from the market, and non-grain-based transaction costs are relatively high. They are more constrained by the endowment of labor, technology, and capital, and they are mostly large-scale grain-growing entities [35].

(3) The adjustment effect of the renting-in entities based on the interaction of scale and location. To explore in depth the impact of scale and region on the non-grain willingness of farmland renting-in entities, this study uses interactive analysis to divide the research objects into large-scale and small-scale farmland renting-in entities in outer suburbs and large-scale and small-scale renting-in entities in suburbs. Nested model comparison item test results show that the p values of the measurement coefficient model and the structural coefficient model grouped by the interaction of scale and location are 0.067 and 0.545, respectively, which are both greater than 0.05, indicating that the measurement models have inter-group invariance and the structural models are equivalent. The group estimation results are shown in

Table 9. Interaction path coefficient table.

Path Relationship	Group 1 (Large-Scale Renting-In Entities in the Outer Suburbs)		Group 2 (Small-Scale Renting-In Entities in the Outer Suburbs)		Group3 (Large-Scale Renting-In Entities in the Suburbs)		Group4 (Small-Scale Renting-In Entities in the Suburbs)
	Estimate	C.R.	Estimate	C.R.	Estimate	C.R.	Estimate	C.R.
SN→NFW	0.321 **	2.780	0.480 ***	7.375	0.387 ***	5.374	0.557 ***	6.363
TC→NFW	−0.486 ***	4.274	−0.376 **	2.012	−0.364	9.374	−0.120	8.736
BC→NFW	−0.149	6.485	0.182	10.342	0.391	5.366	0.068	9.364
EC→NFW	−0.313 **	4.532	−0.056 **	3.450	−0.189 *	3.475	−0.016	2.367
NFW→NFB	0.890 ***	3.563	0.899 ***	2.045	0.897 ***	6.364	0.902 ***	11.263

Note: ***, **, * represent significance levels of 1%, 5%, and 10%, respectively. C.R. is the critical ratio coefficient.

. The subjective normative path coefficient is significant below the 5% level in the four groups. In contrast, the subjective norms of small-scale renting-in entities from suburban areas have the greatest impact, the second is the small-scale renting-in entities from outer suburbs, and the smallest is the large-scale renting-in subjects from outer suburbs. In general, subjective norms have a greater positive impact on the non-grain willingness of small-scale and suburban renting-in entities. Transaction costs have a significant impact on the large-scale group and the small-scale group in the outer suburbs, and transaction costs have a greater impact on large-scale renting-in entities. The comparative benefits were not significant among the four groups. The effect of endowment constraints is not significant in the suburban small-scale group, while other groups are significant. Among them, the endowment constraint has the greatest impact on the large-scale renting-in entities of the outer suburbs; the second is the impact on the large-scale renting-in entities of the suburbs and the least impact on the small-scale group of the outer suburbs. The non-grain willingness is significant in the four groups, and the path coefficients of the four groups have little difference, indicating that the non-grain willingness and behavior have good consistency.

6. Conclusions and Recommendations

6.1. Conclusions

The article uses the improved theory of planned behavior (TPB) to investigate different farmland renting-in entities’ cognitive and socio-psychological behavior to grow non-grain crops in the Zengcheng District of China. The structural equation models are used to conduct an empirical analysis of field survey data of 254 farmland renting-in entities in 37 administrative villages, and analyzes the differences in the factors affecting the non-grain of different types of leasing entities from the dimensions of scale and region. The conclusions are as follows:

(1) The non-grain farmland renting-in entities are mainly affected by comparative benefits and follow-up effects. The greater the realizable profit of non-grain crops than that of food crops, the stronger the non-grain willingness of renting-in entities. Transaction costs and labor, capital, and technical constraints restrict the degrainization of farmland.

(2) There are scale and geographical differences in the non-grain willingness of farmland leasing entities. Small-scale renting-in entities in the suburbs tend to be nongrain-oriented, while large-scale renting-in entities in the outer suburbs tend to be grain-oriented. This is related to the labor productivity of different crops and the difficulty of sales. Non-grain crops in the suburbs are less difficult to sell and relatively profitable, but the labor productivity of non-grain crops is low, and the amount of labor is large, which restricts the expansion of its farming scale. It is more difficult to sell non-grain crops in the outer suburbs, which is more conducive to planting grain crops with less difficulty in selling. However, the profit per unit of food crops is low, and the government encourages large-scale land transfer through gradient subsidies. Only by expanding the area of gain planting can the planting cost be reduced, and large-scale operating profits can be obtained.

6.2. Recommendations

(1) In the outer suburban areas, the government should encourage moderate-scale operation through gradient land transfer subsidies and grain subsidies, the construction of high-standard basic farmland in the outer suburbs should be strengthened, and the convenience of growing grain should be improved through measures such as field consolidation, mechanical improvement, and the construction of ditches and roads. In addition, the current small-scale leasing entity in the outer suburbs is still an important entity for food security, but the comparative disadvantage of food production is difficult to reverse. In the short term, on the one hand, it is necessary to increase the cultivation of socialized service organizations for grain planting so that small-scale leasing entities are involved in the division of labor economy to improve the convenience of growing grain. On the other hand, small-scale leasing entities should be included in the subsidies for growing grains to increase the income per unit output of growing grains, thereby increasing their enthusiasm for growing grains. In the long run, the younger generation’s interest in agricultural operations has weakened. As farmers gradually withdraw from agricultural operations, the question of “who grows grain” should be transferred to the main body of large-scale operations.

(2) Due to high land rents in suburban areas, the transaction costs of nongrain-based operations are lower, and the tendency toward nongrain-based operations is obvious. For small-scale renting-in entities, the spatial distribution rules of crops should be respected, and at the same time, non-grain behavior should be supervised, and the binding role of the land contract and the review and supervision role of the village committee should be given full play on the nongrain-based operations. Resolutely stop non-grain behaviors that damage the cultivated layer, such as digging ponds and planting trees. For large-scale leasing entities in the suburbs, in addition to the supervision of non-grain behaviors, the rotation of non-grain crops and food crops should also be encouraged.

(3) The theory of planned behavior only reveals the factors that affect the intention of non-food behavior. However, the non-grain behavior is also affected by the personal characteristics of farmers (gender, age, etc.), family characteristics (population, annual income, etc.), and characteristics of farmland (area, quality). Subsequent studies can explore the correlation between latent variables under the premise of controlling characteristic variables, and further improve the explanatory and predictive power of the model.