2.1 Review on farmers’ decisions about land transfer-out or transfer-in and transfer scale

Several studies have been undertaken in the past to investigate the decision-makings of whether farmers transfer land in or out of and the transfer scale. These decisions first depend on their comparative advantages in agricultural management or non-farm employment, transaction costs and risks of land transfer. However, the comparative advantage of farmers depends on their labor force, land resources, environmental conditions of agricultural operation and non-agricultural employment opportunities and other factors. Most studies have demonstrated that farmers’ labor force and health status have a negative impact on land transfer-out [5, 6], and a positive effect on land transfer-in [7, 8]; farmer emigration and rural labor transfer are conducive to land transfer [9, 10]. Specifically, the age of farmers has a negative [8] but decreasing influence on land transfer-out [9]. For elderly farmers, age has a positive impact on land transfer-out due to physical decline [6]. In addition, Some studies recommend that the more educated farmers are more likely to choose non-farm employment, which has a positive effect on land transfer-out [11]. However, some other researchers find that the higher the education, the stronger the operation capability, and thus the more likely to transfer into land for large-scale operation, which has a negative influence on land transfer-out [9].

Generally speaking, non-farm employment, part-time farming, non-farm income and its proportion, and the transfer income from government reduce farmers’ dependence on land and thus have a positive impact on land transfer-out [6, 9, 11, 12]. There are also studies concluding that the probability of farmers’ land transfer-out increases in the order of farmers mainly engaged in agriculture, farmers living only by farming, farmers mainly engaged in non-farm and non-farmers; the proportion of the area of transfer-out in the area of the household’ land shows a U-shaped relationship with the degree of concurrent employment [5].

Some scholars have discussed the impacts of land resources and other social resources on land transfer. The unit transaction cost of land transfer-out decreases with the increase of land owned by farmers, which has a positive impact on transfer-out. But the scale effect gradually appears and the possibility of transferring into land for large-scale operation gradually increases when the land owned by farmers increases to a certain scale, which has a negative impact on land transfer-out [6]. Therefore, farmers’ land resources are inversely U-shaped with the land transfer-out [11]. Besides, the willingness to further transfer into land becomes lower when farmers’ land meets or exceeds the requirements of scale economy. Thus, farmers’ land resources also show an inverted U-shaped relationship with land transfer-in [7]. The more fragmented the land, the less convenient the operation is, the worse the quality of agricultural land and thus the more likely to transfer out of land [11, 13]. Furthermore, farmers’ financing opportunities have a positive impact on land transfer-in [7] and a negative impact on land transfer-out [9]. Village cadres have more social resources and are more likely to transfer into land for large-scale operation than ordinary farmers, and thus have a negative impact on land transfer-out [6] and a positive effect on land transfer-in [7]. The information, trust and reputation network of farmers reduce transaction costs of land transfer, which has a positive impact on farmers’ participation in the land market [14].

In addition, some studies have examined the impacts of environmental conditions such as the distance from cities and towns, terrain and infrastructure on land transfer. Generally speaking, the farther away from the town, the worse the topography conditions, and thus the higher the farmers’ willingness to transfer out of land [8]; but for the rural elderly, the farther away from the town, the worse the terrain, the less opportunities to earn money, and thus the more dependent on land for pension, which has a negative impact on land transfer-out [6]. The better the transportation and other infrastructures, the stronger the attraction of land to capital, the greater the demand for land, and the higher the transfer price, which has a positive impact on farmers’ land transfer-out [5, 9, 11], and a negative impact on land transfer-in [7]. In addition, the higher the level of local economic development, the more opportunities for non-agricultural employment. Therefore, it has a positive effect on farmers’ land transfer-out and a negative impact on land transfer-in [7].

The transaction costs of land transfer could be reduced by the improvement of land rights, the formality and convenience of transactions, the development of land market intermediaries, the participation of rural collective organizations, and the transaction experience and land reserves and so forth, which is conducive to developing land market [14–21]. But the adjustment of village collective to the contracted land is adverse to land transfer [22]. Besides, the development of rural social security diminishes farmers’ dependence on land and has a positive impact on land transfer-out [5, 8]. Agricultural technology training and subsidies for large-scale operation have an important impact on farmers’ land transfer [13, 23]; basic farmland protection policy is unfavorable to land transfer [24]. Moreover, the motivation of farmers’ land transfer is not entirely profit maximization; the irrational factors such as the attachment to land, agricultural fun and conservative tendency have an important impact on farmers’ behavior to transfer land. Innovative and optimistic farmers are more likely to participate in the land market [25].

2.2 Review on the mode choice of farmland transfer

With regard to the mode choice of rural land transfer, the previous literature have more investigated the choice between market-based transfer (spontaneous transfer) and government or village committee-led transfer, and between formal contract and informal contract. Some studies explore that there are more spontaneous transfer among farmers in land transfer in non-suburban rural areas; however, in suburban rural areas, the government or village committee-led circulations occur more frequently [26]. Driven by political achievements [27], government-led or village committee-led land transfer easily breeds forced transfer and rent-seeking [28], distorts land resource allocation and reduces allocation efficiency [29]. Therefore, some scholars advocate promoting market-oriented circulation [28, 29]. Nevertheless, other scholars have emphasized that government or village committee dominance decreases transaction costs and risks of land transfer and is conducive to maintaining the agricultural purposes of land [30].

Farmland property rights and its supporting system, the relationship between the parties to the transaction and the participation of government or village collective, land use and so forth have an important effect on the choice of the contract form of land transfer (formal contracts and informal contracts). The signing of formal contracts are significantly promoted by the confirmation of contracted land rights, the establishment of land transfer dispute mediation institutions [31], the constraints of the transfer system, and the participation of village collectives [32]. However, the joint relationship between the parties and the uses of land for food crop cultivation significantly promote oral contracts [32]. Nguyen, Rigg [10] demonstrates that the lease or exchange of agricultural land mainly relies on oral agreements which depend on social capital and trust networks since the relevant system has been unestablished in Vietnam during the reform period.

Besides, some studies examine the choice of price of land transfer and contract flexibility. For instance, the land rights [33] and its risks [34], transaction costs [35], economic development level, land market maturity [36], human feelings [37], and emotions about land [38] have an important impact on the price of land transfer. Land fragmentation, poor land quality, poor irrigation conditions and marginalization of mountainous areas are key factors causing unrewarded land transfer [39]. Because farmers’ decisions on land transfer are based on the principle of satisfaction, there is a deviation between the actual price of land transfer and the market price. Generally speaking, this deviation is relatively small in mature market and economically developed regions [36]. Tretiak & Moskalenko etc. [40] studies land transfer under the land share system in Ukraine and demonstrates that transfer price is artificially formed at the standard monetary valuation level of land rather than by supply and demand. Additionally, Yang [41] discusses the flexibility of land lease contracts and the choice of rent payment, and indicates that lease agents increase the flexibility of contracts and reduce rents; social proximity helps non-stranger entrepreneurs obtain low-flexibility contracts and increase rents.

Some researchers have explored how farmers choose among specific land transfer modes. They show that those choices are mainly affected by farmers’ age, education level, risk preference, land endowment, as well as non-agricultural employment, family income and its stability. For instance, Wang and Gu [42] examines households’ selection in several modes of cultivated land allocation, like self-farming, substituting farming, renting and shareholding. They find that farmers with higher education and family income are more likely to choose lease and shareholding, and risk lover are more likely to choose shareholding. Wang & Yang [43] confirms that part-time farming has an important influence on the mode choices of farmland transfer in Chongqing. The household heads’ age has a positive effect on farmers living only by farming and farmers mainly engaged in agriculture choosing substituting farming, and farmers living only by farming and farmers mainly engaged in non-agriculture choosing subcontracting and leasing, while has a negative influence on farmers mainly engaged in non-agriculture and non-farmers choosing TCMR. Farmers’ land endowment only significantly affects mode choices of farmland transfer of farmers living only by farming and farmers mainly engaged in agriculture. The distance of non-farm employment only significantly influences mode choices of farmland transfer of farmers mainly engaged in non-agriculture and non-farmer, while the stability of non-farm employment only has a positive impact on non-farmers’ choice of TCMR. Land ownership also affects mode choices of transfer [44]. Alves & Maeda etc. [45] discovers that land ownership makes it easier to sell land, but has no effect on the convenience of leasing and bequests in Brazil. Ali & Deininger [46] studies the large farms in Zambia and shows that land ownership reduces participation in the rental market. Furthermore, some researchers examine that subcontracting, lease and shareholding show certain spatial rules [47]. Lease has always been the most common for farm integration among the various specific modes of farmland transfer [48]. Therefore, the existing literature are more about research on land leasing. Karwat-Wozniak & Buks [49] discusses that land leasing in Poland occurs mainly between neighbours; neighborhood leasing will be the basic way to provide land to commercial farms with the depletion of unallocated treasury land resources and farmers’ dependence on heritage.

Moreover, some scholars analyze the choice and change of fixed rent contract and shared tenancy contract. Barzel [50] argues that the costs of supervising the use and improvement of land under fixed rent contracts are high. The costs of supervising labor and output are high under the shared tenancy contract, thus, its transaction costs seem to be higher than that of fixed rent contract. But shared tenancy contract can disperse risks. people will choose shared tenancy contract when its higher transaction costs can be compensated by the benefits of risk diversification [51]. Sun & Yang [52] explains the changes of land tenancy contract in the Republic of China from the perspective of transaction costs, and believes that the replacement of shared tenancy contract and the permanent tenancy system by the short-term fixed rent contract is due to the rise of landlords’ moving to towns and agencies.

As mentioned above, regarding the mode choice of farmland circulation, the previous literature have mainly examined the selection between market-based transfer (spontaneous transfer) and government or village committee-led transfer, as well as between formal contract and informal contract. However, the problem that how the transferor and transferee in farmland market choose specific circulation modes such as TCMR, lease, and shareholding, namely the circumstance that the circulation subjects with different characteristics may choose which specific modes under what conditions has not yet attracted extensive attention of the existing literature.

Based on contract theory and transaction cost economics, this paper adopts the public samples of the Third National Agricultural Census, and takes the case that large-scale households transfer into cultivated land to analyze how they select among three basic circulation modes of TCMR, lease and shareholding, focusing on the impacts of circulation areas, uses and family characteristics on the mode choices of circulation.

(1) Effects of areas and uses of circulation on the mode choice of farmland transfer-in

Due to the land system that rural land is collectively owned by farmers and only the collective members enable to enjoy contractual management rights of the collective land, the scope of TCMR is narrow and is limited to occur among the same collective members (See the Rural Land Contract Law of the People’s Republic of China). Meanwhile, TCMR will make the transferor lose the contractual rights and the security function of land, thus, only a few households are willing to choose TCMR to transfer out of land [59]. According to the statistics of the Ministry of Agriculture and Rural Affairs of the People’s Republic of China, the area of TCMR only accounts for 2.41% of the circulation area of the farmland contracted by rural households in China by 2020. Therefore, TCMR is difficult to meet the requirements of households’ largely transferring into farmland. However, lease and shareholding have been expanded beyond the collective, overcoming the narrow scope of TCMR. In terms of lease and shareholding, although the contractors transfer out of land management rights, they retain contractual rights. It makes a large number of contractors who are willing to transfer out of the contracted land but are unwilling to give up their contractual rights enter the market and the transferee can gather land in a large range. Consequently, ceteris paribus, with the increase of the areas of transfer-in, the probability of choosing lease and shareholding may increase relatively, whereas the probability of choosing TCMR may decrease relatively.

In addition, circulation areas may also affect the mode choice of farmland transfer-in through capital demand. The greater the areas of transfer-in (the scale of management), the greater the capital demand, and the more likely the transferee is to choose the contractual form which can reduce capital demand. Under the contracts of TCMR and the fixed rent contract, the transferee or lessee needs to pay the price of TCMR or rent, while it is not required to pay that price or rent under the shareholding contract. Therefore, ceteris paribus, in order to reduce the demand for funds, the probability of large-scale households’ choosing shareholding may increase faster than that of choosing lease with the increase of the areas of transfer-in.

The uses of farmland transfer-in include crop cultivation, garden and forestry management, breeding, etc. Different uses have different production cycles. The longer the cycle, the greater the uncertainty (For example, within the contract period, due to the laws or policies related to farmland circulation and the market price of farmland circulation may change, the transferor may request to terminate the contract under the pretext of the change of the agreed conditions of the contract.), and the higher the operational risks may be [60]. Therefore, the uses of farmland transfer-in may have an important influence in the mode choice of transfer-in owing to the differences in operational risks.

(2) Effects of the characteristics of human capital of large-scale households on the mode choice

The characteristics of human capital of large-scale households include the ability to search for trading partners and information, organizational ability, and the specificity of human capital. Firstly, when large-scale households transfer into farmland, the costs and scope of searching for trading partners are affected by search capability. The stronger the search ability, the more likely they are to lease land or absorb others to share with land management rights beyond the collective, so the probability of choosing lease and shareholding may rise relatively. Secondly, from the perspective of management decision-making, the transferee or the lessee enjoys the exclusive management decision-making rights under the contracts of TCMR and lease. While, under the contract of shareholding, the costs of decision-making are more higher as shareholders collectively execute their decision-making rights [61, 62]. In order to reduce the costs of decision-making, shareholding requires large-scale households to have higher organization and control capability as the core subjects of the contract. Therefore, ceteris paribus, as large-scale households’ organization and control capability becomes stronger, the probability of their choosing shareholding when transferring into cultivated land may relatively increase. Thirdly, the stronger the specificity of large-scale households’ human capital dedicated to agriculture (such as cultivation experiences), the more likely they are willing to choose a stable and persistent contract to play the role of that human capital when transferring into farmland. The survey finds that so far, due to the imperfection, great uncertainty, low default cost and weak contract spirit of the market of farmland management right transfer, the lease and shareholding contracts have poor stability (Specifically, the period of lease is short (mostly less than 5 years), and the lessor’s request for raising rent or cancellation of the contract is common within the contract period. For shareholding contract, not only the costs of decision-making are high, but also the phenomenon that land shareholders require withdrawing shares occurs sometimes.) [63, 64]. Compared with lease and shareholding, the transfer of contractual management rights (TCMR) is the most stable and persistent [65]. Once TCMR occurs, the transferee stably obtains the contractual management rights of land. Therefore, ceteris paribus, the probability of choosing TCMR may relatively increase when large-scale households transfer into farmland with stronger specificity of human capital. This paper chooses gender, age, and educational attainment of household heads of large-scale households and the number of labor forces to reflect their characteristics of human capital.

(3) Effects of the environment of farmland on the mode choice of land transfer-in

The environment of cultivated land includes topography, traffic, irrigation water and so forth, which may have multiple effects on the mode choice of farmland transfer-in. First, ceteris paribus, the contractors’ willingness to transfer out of the contractual management rights may be relatively lower in areas with better environmental conditions, such as plains, traffic-developed areas and water-sourced areas. That is, for the transferee, the number of potential counterparties choosing TCMR is relatively smaller, and the costs of searching and waiting for transaction are relatively higher. Thus, subject to this, the probability of large-scale households’ choosing TCMR to transfer into farmland may be reduced relatively. However, there may also be an opposite forces, that is, for land with better environmental conditions, the transferee may relatively prefer to establish a stable and persistent contractual relationship. While as mentioned above, compared with lease and shareholding, TCMR is the most stable and lasting. Therefore, the probability of choosing TCMR may relatively increase. Subsequently, environmental conditions may affect the future value of cultivated land. Nevertheless, due to the complexity and uncertainty of environmental factors that affect the future value of land and the bounded rationality of human beings, it is fairly difficult to predict the future value of land, and the costs of finding transaction prices and bargaining may be quite high in TCMR and lease. In the shareholding, equity returns automatically reflect the change of land future value, thus saving the costs of predicting the change of land future value. As a result, ceteris paribus, the more complex and uncertain the environmental factors that affect the future value of land are, the more likely large-scale households are to choose shareholding when they transfer into farmland. Therefore, the influence of environment of cultivated land on the mode choice of transfer-in is uncertain and needs to be tested.

(4) Effects of the security function of land on the mode choice of farmland transfer-in

As mentioned above, large-scale households’ mode choices of transferring into cultivated land are subject to the number of potential counterparties with different modes in the market, namely households’ mode choices of transferring out of cultivated land. The land security function (Although there are different opinions on the security function of rural land, the land does bear different levels of social security functions for most farmers since the social security system is not perfect and the security level is low in Rural China. Many elderly farmers rely on land management to make up for the lack of pension. Land management is also the employment guarantee for a large number of farmers who have transferred employment after the failure of transfer employment [68–70].) is an important factor that influences households’ decision-making. Meanwhile, under different circulation modes, land has different security function to the contractors. For instance, if contractors choose TCMR, they will completely lose the land security function. Although lease or shareholding will weaken the security function of land to contractors, it will not entirely make them lose the security function of land (First, as the lease term expires or enterprises where contractors take shares are dissolved, the land management rights will be returned to the contractors. At this time, if necessary, the contractors enable to choose farming by themselves to achieve the survival and employment security function of land. Second, the rent or dividend income also provides a certain survival guarantee for contractors.). Consequently, the land security function may have an important impact on households’ mode choices of transferring out of cultivated land and then on that of transferring into cultivated land. In addition, in terms of regions, the land security function mainly depends on the development level of non-agricultural industries and the rural social security and so forth. Ceteris paribus, the more developed the non-agricultural industries and the rural social security are, the weaker the security function of land is, and thus the more likely farmers are to choose TCMR when they transfer out of cultivated land. As a result, for the transferee, the probability of choosing TCMR may increase accordingly.

(1) Circulation areas

The average households’ areas of transfer-in in the sample are about 7.25 hectares, with the minimum of 0.0067 hectares and the maximum of 866.67 hectares.

(2) Circulation uses

The most majority of large-scale households transfer into cultivated land for crop cultivation, accounting for about 87.15 percent of the total sample, only 6.73 percent for breeding and 5.61 percent garden & forestry managements. In addition, few used for other business.

(3) The characteristics of human capital of large-scale households

This paper chooses gender, age, educational attainment of the household heads and the number of labor forces to reflect the characteristics of human capital of large-scale households. Among the samples, 92.52% household heads are male and 7.48% are female. The average age of the household heads is about 47 years old. The smallest is 1 year old, and the largest is 94 years old. Moreover, the education attainment of household heads is mainly junior high school, accounting for 60.53%, followed by primary school, and high school & above education accounting for 26.02% and 12.02% respectively. Additionally, few have not attended school and the average number of labor forces per household is about 2.7, with the minimum of 0 and the maximum of 12.

Control variables include village topography (plains = 1; hilly areas = 2; mountainous areas = 3), township traffic conditions (whether there are highway exits and entrances, train stations, docks: none = 0; one of the three = 1; two of the three = 2; all of the three = 3), the main irrigation water source of the village (no = 0; yes = 1), the ratio of pension insurance coverage (the ratio of pension insurance coverage to village household registration population, namely RPI), and the proportion of output value of the primary industry of the province (city) taking up in GDP (RFI, hereinafter referred to as the proportion of the primary industry). Among them, the first three variables reflect the environmental conditions of cultivated land, and the last two variables reflect the security function of local land.

Owing to the database used in this paper have been desensitized, it is impossible to identify or infer which administrative village or township corresponds to the large-scale households in the sample, but it can identify whether they belong to the same prefecture-level city. Therefore, this paper uses the prefecture average level of topography, traffic, water and RPI among those control variables.

5.1. Model setting

The explained variable contains three unordered categories (TCMR = 0; lease = 1; shareholding = 2), so this paper adopts mlogit regression. According to the above theoretical analysis, the mlogit model is established as follows: (1)

In function (1), ML indicates the logit predicted value or Log Odds of households choosing different modes of farmland transfer-in. x_i represents each explanatory variable; z_i represents each control variable; μ is a random error item; β₀, β_i, δ_i are parameters to be estimated.

Since the circulation areas have abnormal observations, the logarithm of circulation areas is taken in the model.

5.2. Results and discussions

Using STATA for statistical analysis, mlogit regression analysis is carried out on the data with the minimum value as reference and the results are shown in Table 2. Moreover, In order to deal with the heteroscedasticity in the linear probability model, Table 2 reported robust standard error and statistics, and Prob > chi2 = 0.0000 indicates that overall explanatory variables and control variables have significant impacts on explained variables.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Model regression results.

https://doi.org/10.1371/journal.pone.0287022.t002

This paper also examines the interaction effect between explanatory variables. In order to avoid multicollinearity, a middle strategy is used for interaction items (except for virtual variables and less classified categorical variables). The results show that there is a significant interaction effect only between the age of household heads and the number of labor forces. Therefore, the interaction term of age and labor, namely (age-mean)#(labor-mean), is introduced into Model 2. The impacts of these variables are discussed as follows.

5.3. Robustness test

Considering that models may have endogenous problems and the results are not robust, in this paper, we carry out the robustness tests in model 3 and model 4. On the one hand, according to Lu & Wang’s [66] practice of removing special samples, the special samples of transferring into cultivated land for other purposes including multiple uses are excluded in Model 3. The results show that in contrast to transferring into cultivated land for crop cultivation, the probability of those whose purpose is garden & forestry management choosing lease is relatively higher, but it is insignificant, and the possibility of choosing shareholding increases significantly. These correspond to the results of Model 1 and Model 2. Meanwhile the symbols of other variables are also consistent with those of Model 1 and Model 2, and the changes of coefficient and significance are little.

On the other hand, in Model 4, the educational attainment of the heads of large-scale households is treated as a continuous variable. It is shown that among the choice of three circulation modes, the probability of choosing lease and shareholding increases significantly with the improvement of educational attainment, while the latter increases faster, which is consistent with the results of Model 1 and Model 2. In addition, the symbols of other variables also correspond to those of Model 1 and Model 2, and the coefficient and significance change little. Consequently, in this study, the estimation results are robust.