EN中文

Journal of Geographical Sciences >

2023 , Vol. 33 >Issue 8: 1702 - 1724

DOI: https://doi.org/10.1007/s11442-023-2149-z

Special Issue: Human-environment interactions and Ecosystems

Impact of labor transfer differences on terraced fields abandonment: Evidence from micro-survey of farmers in the mountainous areas of Hunan, Fujian and Jiangxi

  • XIE Hualin , 1 ,
  • WU Qing , 2, * ,
  • LI Xiubin 3
Expand
  • 1. Institute of Ecological Civilization, Jiangxi University of Finance and Economics, Nanchang 330013, China
  • 2. School of Business, Jiangxi University of Science and Technology, Nanchang 330013, China
  • 3. Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
* Wu Qing (1991-), PhD, specialized in land resource economy. E-mail:

Xie Hualin (1979-), Professor, specialized in land resource economy and ecological economy. E-mail:

Received date: 2023-04-10

  Accepted date: 2023-05-25

  Online published: 2023-08-29

Supported by

National Natural Science Foundation of China(41930757)

National Natural Science Foundation of China(41971243)

Fold

Abstract

In recent years, the phenomenon of abandonment of cultivated land in mountainous areas has occurred frequently, and the problem of abandonment has become a focus of attention of government agricultural departments and academic circles. However, few studies have paid attention to the impact of differences in labor transfer on the abandonment behavior of farmers’ terraced fields. Based on this, this paper takes the terraced fields of Hunan, Fujian and Jiangxi provinces as the research area, combined with data from 1438 farmer households, and uses the Probit and Tobit models to analyzed the impact of the difference in the quantity, distance and quality of labor off-farm transfer on the decision-making and scale of terraced field abandonment of farmer households. The results show that: (1) The greater the quantity of labor transfer, the farther the transfer distance and the higher the quality of non-agricultural employment, can promote the decision of farmers to abandon terraced fields; (2) With the deepening of the degree of non-agricultural transfer, the scale of abandoned terraced fields by farmers in hilly and mountainous areas will also expand; (3) The distance and quality of labor transfer can strengthen the impact of labor transfer quantity on farmers’ decision to abandon land. To alleviate the phenomenon of abandoned terraced fields, the government should pay attention to the labor substitution role of agricultural service outsourcing and mechanization in mountainous areas. The government should actively promote the transformation of terraced fields into mechanized farming, improve the cultivated land transfer market, and encourage farmers to transfer terraced fields.

Key words： terraced fields abandonment; behavior mechanism of abandonment; labor transfer difference; hilly and mountainous area

Cite this article

XIE Hualin , WU Qing , LI Xiubin . Impact of labor transfer differences on terraced fields abandonment: Evidence from micro-survey of farmers in the mountainous areas of Hunan, Fujian and Jiangxi[J]. Journal of Geographical Sciences, 2023 , 33(8) : 1702 -1724 . DOI: 10.1007/s11442-023-2149-z

1 Introduction

China is a country rich in farmland, but a large proportion of it is in hilly and mountainous areas. Compared to the plains, topographical conditions limit the use of farmland resources in the hilly mountains. For a long period, farmers in the mountainous areas have built a large number of terraces in order to harvest more food for their livelihoods. Terraces reflect farmer’s wisdom in adapting to nature and embody the idea of “storing grain in the farmland and technology”. In the process of cultivating the terraces, unique production techniques, management experience and culture have been developed, with inspiration and significance for the development of sustainable agriculture (Yao and Wang, 1991; Chen et al., 2016; Zhang and Min, 2016).
Terrace, although a valuable traditional farming system, is facing gradual marginalization. With social and economic developments of China, a large number of rural labor forces shift to non-agricultural sectors in towns and cities. The outflow of rural labor force decreases the labor supply in terrace agriculture, resulting in a large number of terraces are abandoned (He, 2005; Xiang and Han, 2005; Zhang et al., 2011; Ma et al., 2018). For example, in 2016 the abandoned area of Longji terraced in Guangxi reached 11.13 hm2, in 2016 the abandonment rate of the Hakka terraces in Chongyi, Jiangxi was as high as 39% (Shao et al., 2011; Miao et al., 2018). The terraces are an integral system, with the fields closely linked by ditches and roads. If part of the concentrated terraced field area is abandoned and collapsed, it will affect the stability of the entire terrace system. Farmland abandonment has attracted the attention of the Chinese government, with the Ministry of Agriculture and Rural Development issuing the “Guidance on the Coordinated Use of Abandoned Land for the Development of Agricultural Production” in 2021, requiring departments at all levels to strengthen the tracking and supervision of farmland abandonment, and to guide farmers to actively replant abandoned farmland. Therefore, it is urgent to conduct in-depth research on the driving factors of terrace abandonment by farmers in mountainous areas.
Existing research has conducted a lot of exploration on the reasons for households’ abandonment of farmland, and the results indicate that limiting the supply of agricultural labor due to the off-farm transfer of is an important factor affecting the farmland abandonment (Aide and Grau, 2004; Zhang et al., 2014a; Shi et al., 2018; Zhang et al., 2019; Li et al., 2021). Lieskovský et al. (2013) found that aging of labor force, precipitation and farm size are the main factors affecting abandonment. Bezu et al. (2014) found that the majority of rural youth labors in southern Ethiopia opted for non-agricultural employment in order to separate from agriculture. Huang et al. (2021) points out that the agricultural transfer of young rural male labor is the direct driver of farmland abandonment. Zhao et al. (2018) found that the spatial distribution of abandoned farmland is affected by factors such as rural labor transfer. Zhang et al. (2019) proposed that the formation of China’s abandonment spatial pattern is positively correlated with regional economic development and labor force transfer, among which low agricultural income and insufficient labor force are common factors. Li et al. (2016) pointed out that the changes in socio-economic factors caused by labor migration are also one of the driving forces affecting farmland abandonment.
It has also been argued that the shift of agricultural labor does not necessarily lead to the withdrawal of farming and the abandonment of farmland. As Cai suggests, there are two sides to the impact of labor transfer on households’ agricultural decisions, with the negative effect of causing farmers to abandon their land and the positive effect of promoting agricultural production (Cai, 2001; 2018). In agricultural production, agricultural machinery and labor input can replace each other. Increasing the input of agricultural machinery can alleviate the insufficient agricultural labor input caused by labor transfer (Huang et al., 2018). Therefore, rural labor transfer can promote the development of agricultural mechanization (Ma et al., 2004; Sheng, 2007). At the same time, Huang and Li (2019) suggest that the agricultural labor transfer will increase the proportion of grain crops in agriculture production, which confirms that labor transfer may not have a significant negative impact on the actual area of cultivated land (Tian et al., 2010; Liu, 2011).
Overall, there is no consensus in existing research on the impact of agricultural labor transfer on farmers’ abandonment behavior, and there is little systematic theoretical research and empirical testing on the mechanism of terrace abandonment behavior of farmers in hilly and mountainous areas. In the context of the continuous development of modern agriculture, the abandonment mechanism of farmland in plain areas cannot fully explain the internal reasons for the continuous abandonment of valuable terrace. Reviewing existing research, it is found that when analyzing the impact of agricultural transfer of labor force on abandonment behavior, current research often regards the transfer of individual labor as the complete departure of the entire family from agriculture (Xie and Huang, 2022), ignoring the stage and part-time characteristics of labor transfer in farmer’s families (Cai, 2001), and lacks analysis of the impact of different transfer degrees on abandonment behavior from the perspective of differences in labor transfer among farmers. Paying attention to the differences in rural labor transfer is crucial for analyzing farmers’ terrace abandonment behavior.
This is because in rural households, the transfer of a single labor force does not represent the withdrawal of the entire family from agricultural production activities. In different transfer stages and situations, there are differences in the quantity and distance of labor transfer between rural households, and the impact of labor transfer quantity and distance on farmers’ decision to abandon terrace is not consistent. The differences are mainly reflected in the following two aspects. Firstly, most of the transfer of rural labor in China is not permanent, and the agricultural labor transfer presents incomplete and phased characteristics. Most farmers in China choose to engage in both agriculture and non-agriculture, which also results in different labor transfer quantities, transfer distances, and situations for rural households. Different levels of labor transfer represent different non-agricultural transfer costs, and the impact on farmers’ abandonment decisions is not consistent. Secondly, there are differences among individual labor forces, as the human capital contained in different labor forces and the labor intensity they can withstand vary. Due to limitations in their own conditions, the majority of transferred agricultural labor is concentrated in labor-intensive, low-tech, and easily replaceable industries and positions. The quality of non-agricultural employment is not high, and they face higher uncertainty and risks. This part of labor transfer does not necessarily lead to terrace abandonment. Based on the above reasons, in hilly and mountainous areas facing dual pressures of resource endowment constraints and labor transfer, further research is needed on how the differences in the quantity, distance, and quality of agricultural labor transfer affect households’ decision to abandon terrace.
Given this, this article elaborates on the differences in rural labor transfer from agriculture to non-agriculture, examining the impact of transfer quantities, transfer distances, and transfer quality on farmers’ terrace abandonment decision-making and scale. By analyzing the relationship between the characteristics of non-agricultural employment of farmers and terrace abandonment, this research explores the impact mechanism of rural labor transfer differences on farmers’ terrace abandonment decision-making and scale. This study can provide suggestions for the development of targeted measures for the utilization and protection of terrace, with the aim of providing valuable reference for China’s future development of farmland protection policies.

2 Theoretical analysis and hypothesis

With the rapid advancement of industrialization and urbanization in China, farming is no longer the only way for farmers’ families to allocate labor resources and obtain income. The labor in rural households faces choices from both the agricultural and non-agricultural sectors, there are structural differences in the quantity, distance, and quality of agricultural labor transfer among different rural households. When analyzing farmers’ abandonment decisions from the perspective of rural labor transfer, it is not only necessary to analyze whether the rural labor transfer occurred, but also the differences in the quantity, distance, and quality of labor transfer among rural households.
(1) Constraint effect of Agricultural Labor Transfer quantity
From the perspective of labor transfer quantity, the number of labors in a rural household is limited. As the number of agricultural labor transfers, it will inevitably reduce the input of agricultural labor in agricultural production. As farmers’ non-agricultural employment opportunities increase, especially as non-agricultural employment income continues to increase, under the condition that the total amount of labors remains unchanged, the labor input into agriculture will gradually decrease. Eventually, the agricultural labor of a household will be lower than the minimum labor limit required for agricultural production. Agricultural production will face a series of problems with insufficient labor input. At this time, households’ production decisions will change due to insufficient agricultural labor. For example, rural household will reduce the multiple cropping index and abandon some farmland to alleviate insufficient agricultural labor input.
(2) Constraint effect of agricultural labor transfer distance
The rural labor transfer has significant spatial distance differences, and while analyzing the impact of labor transfer on agricultural production, it is necessary to comprehensively study the constraint effect of labor transfer distance on agricultural production input. Most rural labor in China allocate their time in the form of “half farming and half working”, the agricultural labor transfer does not necessarily lead to the loss of agricultural labor, as there are significant spatial and distance differences in the agricultural labor transfer among farmers’ families (Zhu and Luo, 2020). Moreover, due to the limitations of labor transfer costs, some rural households with relatively low stock of human capital have lower non-agricultural employment time and income, these farmers are more likely to engage in part-time activities as a temporary worker in the local area, neither “leaving the farmland” nor “leaving the agriculture” (Tan et al., 2019). As shown in Figure 1, if a farmer’s labor transfer distance is relatively close and only works locally, then only a corresponding reduction in labor input for agricultural production is needed to simultaneously take into account non-agricultural employment and agricultural production, and the possibility of abandoning farmland is less.
Figure 1 Impact of non-agricultural employment distance on farmers’ terrace abandonment decision
When farmers meet the definition of long-distance labor transfer, and the farther the transfer distance, the less likely the labor is to engage in agricultural production at the same time. Rural families face greater rigid constraints on their labor and have to reduce production scale or farmland use intensity. Therefore, the farther the transfer distance of agricultural labor, the greater the possibility of farmers abandoning their farmland. However, regardless of the situation of labor transfer, it is necessary to consider the differences in human capital among farmers (Hou, 1999), which requires consideration of quality differences in labor transfers and non-agricultural employment caused by individual differences among farmers.
(3) Constraint effect of agricultural labor transfer quality
While considering agricultural labor transfer distance, it is also necessary to further analyze the quality and stability of farmers’ non-agricultural employment. Firstly, rural labor transferred to non-agricultural employment is mainly composed of young and middle-aged people with higher education levels, while women, the elderly and children are left at home to undertake small-scale agricultural production activities, reflecting selective transfer of agricultural labor (Li and Guo, 2011). With the gradual transfer of rural labor, the main labor and capital (including human capital) of rural households have withdrawn from agricultural production. the proportion and working time of the elderly and female population participating in agricultural production have significantly increased, and the labor participating in agricultural production is showing an aging and feminization trend.
Secondly, many studies point out that the rural labor transfer in China is constrained by farmers human capital. The flow and transfer of rural labor are like migratory birds, where they go out to work during leisure time and return home to continue farming during busy farming seasons (Huang and Zhang, 2005; Zhang et al., 2020). However, with the development of urbanization and the improvement of education level of rural labor in recent years, the human capital of rural labor has been improved. Some young and middle-aged labor with high human capital has gradually shifted from migrant workers to citizens, and the stability of non-agricultural employment has greatly improved. Some labor no longer participates in agricultural production, resulting in a further decline in the quality of labor engaged in agricultural production (Li and Guo, 2011; Zhao et al., 2020). This leads to insufficient effective labor input in agricultural production, resulting in the abandonment of some farmland. Therefore, when considering the impact of agricultural labor transfer number and distance on terrace abandonment, it is also necessary to consider the impact of the labor transfer quality and stability.
In summary, this article systematically analyzes the impact mechanism of rural labor transfer differences on farmers’ terraced field abandonment from three aspects: the number of non-agricultural employments, the distance of non-agricultural employments and the quality of non-agricultural employments. The theoretical analysis framework of the impact of rural labor transfer differences on terrace abandonment is shown in Figure 2. Based on this, this article proposes the following two research hypotheses to be verified.
Figure 2 Analysis of the impact of rural labor transfer difference on terrace abandonment
Hypothesis 1: The number, distance and quality of rural labor transfer have a positive impact on farmers’ terrace abandonment decisions and abandonment scale.
Specifically, the quantity, distance, and quality of rural labor transfer alone will have a positive impact on farmers’ terrace abandonment decisions and abandonment scale. This is manifested in the fact that as the quantity, distance, and quality of rural labor transfer increases, and the more inclined farmers are to abandon terrace. But the rural labor transfer is not achieved overnight but phased (Cai, 2001), this article proposes another hypothesis to be verified.
Hypothesis 2: The distance and quality of rural labor transfer will exacerbate the impact of the quantity of rural labor transfers on households’ terrace abandonment decisions.
Specifically, the number of rural labor transfer constrains the supply of labor required for agriculture. However, this constraint is influenced by distance and quality of rural labor transfer. According to the on-site investigation, most rural households have undergone labor transfer, the transfer proportion of labor is relatively high, and the transfer distance is relatively close. By reducing the labor input to agricultural production, both non-agricultural employment and agricultural production can be balanced. Therefore, the possibility of terrace abandonment is low.
Similarly, if there is a high transfer proportion of labor with low non-agricultural employment quality, the farmers’ non-agricultural income and time to participate in non-agricultural employment will be less. To analyze the impact of labor transfer quantity on households’ abandonment decisions, the interaction term between transfer quantity and transfer distance, as well as the interaction term between transfer quantity and transfer quality will be introduced to the empirical model.

3 Data and methods

3.1 Data source

The data comes from the field survey of rural households in Hunan, Jiangxi and Fujian provinces conducted by the research group. To obtain the data of rural households, the first step is to determine a suitable survey location with the following process: (1) Select Hunan, Jiangxi, and Fujian with large amount of terrace as the survey destinations; (2) Based on stratified sampling method, determine the number of samples based on the population, total farmland, and total terrace of the three provinces, and select prefecture level cities; (3) Randomly select 2-3 counties (including county-level cities and districts) from the selected prefecture level cities, and selects 1-2 townships from each county; (4) Randomly select 3-4 administrative villages with terrace from each township, and complete survey questionnaires of 10 households in each administrative village.
Field research was conducted in July, August, and November 2020. To ensure the quality of questionnaire, the investigators were uniformly trained to familiarize themselves with the content and connotation of the questionnaire before the formal investigation. When communicating with farmers, questionnaire Q&A and family interviews is adopted, the interviewer fills out the questionnaire and records interview contents on the spot. A total of 1750 questionnaires were distributed during the survey, and 1733 questionnaires were collected finally.
Due to the large number of questionnaires in some townships, based on the quantity and distribution of the questionnaires, this article excludes some farmers who participated in the rotation, fallow and returning farmland to forest projects, and deletes some poor quality and stacked samples. As shown in Figure 3, the final sample size is 1438.
Figure 3 Sample distribution

3.2 Methods

To achieve the research objectives, econometric models (Probit + Tobit) were used to probe the relationships between terrace abandonment and agricultural labor transfer differences.

3.2.1 Model for the impact of agricultural labor transfer differences on households’ terrace abandonment decision

Probit model can analyze whether rural household abandon terrace. In Probit model, 1 indicating the presence of abandoned terrace, and 0 indicating the absence of abandoned terrace. Probit model is suitable for regression analysis with the binary dependent variable and is an ideal model for analyzing the willingness and decision-making of micro-individuals (Zhang et al., 2011). The formula of Probit model is as follows.
${{Y}_{i}}=\alpha +\beta labo{{r}_{i}}+\delta {{X}_{i}}+\varepsilon $
where Yi represents the binary dependent variable that whether the rural household abandons terrace or not. If Y equals 1, terrace abandonment occurred, Y equals 0, terrace abandonment not occurred. labori represents characteristic of agricultural labor transfer difference; Xi represents labor characteristic, substitution extent of labor input, income characteristic, terrace endowment characteristic and regional characteristic of rural household; ε represents random disturbance term.

3.2.2 Model for the impact of agricultural labor transfer differences on households’ terrace abandonment scale

The Tobit model is used to further explore the impact of agricultural labor transfer differences on household terrace abandonment scale. The formula of Tobit model is as follows.
$Y_{i}^{*}=\alpha +\beta labo{{r}_{i}}+\delta {{X}_{i}}+{{\varepsilon }_{i}},\text{ }\left\{ \begin{matrix} Y_{i}^{*}={{Y}_{i}}~~if~~Y_{i}^{*}>0 \\ Y_{i}^{*}=0\text{ }if\text{ }Y_{i}^{*}\le 0 \\ \end{matrix} \right.$
where Y* I represents terrace abandonment rate of the i-th rural household, which is calculated as the proportion of abandoned terrace to the total terrace; labori represents characteristics of agricultural labor transfer difference; Xi represents labor characteristics, labor substitution degree, income characteristics, terrace endowment characteristics and regional characteristics of rural household; ε represents random disturbance term.

4 Impact of agricultural labor transfer differences on rural households’ terrace abandonment

4.1 Summary statistics of variables

This research selects “abandon terrace or not” and “terrace abandonment rate of rural household” as dependent variables to analyze the impact of agricultural labor transfer differences on terrace abandonment. Referring to existing research, the concept of agricultural labor transfer in this research is labor of rural household transfer from the agricultural sector to the non-agricultural sector and engage in non-agricultural work, and quantity, distance and quality of agricultural labor transfer are selected as core independent variables.
Agricultural labor transfer quantity refers to the amount of non-agricultural employment labor in a rural household; agricultural labor transfer distance refers to the distance between the place where a non-agricultural labor works and home; agricultural labor transfer quality refers to the stability of non-agricultural employment, which is measured by non-agricultural labor time.
Control variables representing labor characteristics, income characteristics, terrace endowment characteristics, labor substitution degree and regional characteristics of rural household are selected. Table 1 shows the interpretation and descriptive statistics of each variable.
Table 1 Definition and summary statistics of variables
Variables Description Mean Standard deviation
Dependent variables
Terrace abandonment behavior Whether household abandon terrace (0=no;1=yes) 0.405 0.491
Terrace abandonment rate Proportion of abandoned terrace to the total farmland (%) 0.170 0.271
Independent variables
Agricultural labor transfer differences Transfer quantity Proportion of non-agricultural employment labor to the total labor (%) 0.678 0.324
Transfer distance Weighted processing of non-agricultural employment 1.797 1.244
Transfer quality Working times of non-agricultural employment (Month) 18.593 13.114
Control variables
Labor
characteristic		 Amount (Number) 2.797 1.203
Average age Average age of labor (Year) 48.164 9.512
Education Labors with middle school education level (Number) 1.060 0.969
Health status Proportion of labor with poor health to the total labor (%) 0.126 0.233
Elderly labor proportion Proportion of elderly labor to the total labor (%) 0.149 0.277
Village cadre or not Are there village cadre among labors
(0=no; 1=yes)		 0.153 0.368
Income
characteristic		 Total income Processed with logarithm
(yuan, 1 USD≈6.8974 yuan)		 4.804 0.424
Terrace
endowment
characteristic		 Farmland transfer Whether transfer farmland (0=no; 1=yes) 0.516 0.5
Farmland plots (Number) 33.004 46.361
Farmland area (Mu, 1 Mu≈0.067 ha) 5.307 4.012
Farmland soil quality (1=poor; 2=medium; 3=good) 2.016 0.654
Substitution degree of labor input Mechanization degree of agricultural production Assign values from whether terrace can be plowed, seeded, and harvested mechanically 0.445 0.454
Regional
characteristic		 Regional variable N/A
No of samples 1438
(1) In terms of agricultural labor transfer differences among rural households, proportion of non-agricultural employment labor to the total labor reflects the transfer quantity; labors degree of apart from agriculture reflects the transfer distance, and divide labors’ non-agricultural employment into 5 types according to the distance between the place where a non-agricultural labor works and home: within township, outside township and within county, outside county and within city, outside city and within province and outside province, labors degree of apart from agriculture = proportion of non-agricultural employment within township ×1 + proportion of non-agricultural employment outside township and within county ×2 + proportion of non-agricultural employment outside county and within city ×3 + proportion of non-agricultural employment outside city and within province ×4 + proportion of non-agricultural employment outside province ×5 (Tan et al., 2019), the higher labors degree of apart from agriculture, the less time spent on agriculture and the higher possibility that abandon terrace; average working times of non-agricultural employment reflects the transfer quality (Lu and Hu, 2017), the more non-agricultural working times, the more labor shift to non-agricultural activities and the higher possibility that abandon terrace.
(2) Labor characteristics of rural household include total labor amount, average age of labor, education level of labor, health status of labor, elderly labor proportion and village amount. As the important productive factors, the more available labor for agriculture activities, the less possibility that abandon terrace. The proportion of elderly labor reflects labor quality, if a rural household has a high proportion of elderly labor, the quality will be relatively low as elderly labors unfit high-intensity agricultural production and terrace will be abandoned.
To education level and health status of rural household, the higher average education level means stronger ability expand agricultural production through advanced agricultural technology. On the other hand, the higher education level, the higher human capital endowment, which means transferred labor will have more non-agricultural employment opportunities and lead to terrace abandonment. Similarly, labor proportion with poor health affects agricultural operations lead terrace abandonment too.
(3) Substitution degree of labor input is reflected by assign values from whether terrace can be plowed, seeded, and harvested mechanically. Replacing labor with machines can reduce the manpower required for agricultural production to some extent, enabling labors to maintain agricultural production while engaging in non-agricultural work. Therefore, higher labor input substitution degree means less possibility that abandon terrace. This research measured mechanization degree from whether terrace can be plowed, seeded, and harvested mechanically, and used factor analysis method to combine mechanization degree of different production processes into one value.
(4) Terrace endowment characteristic is reflected by area, plots, soil quality and whether transfer farmland. Soil quality is an important factor in farmers decision-making of terrace abandonment (Zhang et al., 2014b). The worse the terrace soil quality, the more likely to be abandoned. The amount of terrace reflects the fragmentation and mechanization difficulty. Fragmentation of terrace in mountainous and hilly areas is more obvious than in plain areas, which increases agricultural costs. Fragmentation of terrace is not conducive to intensive land use. In mountainous areas, farmers need to spend more time and energy shuttling between different slopes and altitudes. Therefore, the higher fragmentation degree of terrace, the more likely to be abandoned. In terms of farmland transfer, household which transfer farmland are less likely to abandon terrace.

4.2 Baseline model

4.2.1 Impact of agricultural labor transfer differences on household’s terrace abandonment decision

Based on the econometric models’ settings in the previous section, we used Stata 16.1 to run the Probit regressions, and used Robust regression to deal with potential heteroscedasticity. According to the model results, Endogenous Wald χ2 values are significant at the 1% level, indicates fitting results of the models are good and subsequent analysis can be performed. Table 2 shows the regression results of the Probit model.
Table 2 Probit model regression results of the impact of labor transfer differences on households’ terrace abandonment decision
Variables Dependent variable: Whether household abandon terrace (0=no;1=yes)
Probit model 1-1 Probit model 1-2 Probit model 1-3
Coefficients Robust standard error dy/dx Coefficients Robust
standard error		 dy/dx Coefficients Robust standard error dy/dx
Core independent variables
Agricultural labor transfer differences Labor transfer quantity 0.284* 0.139 0.934
Labor transfer distance 0.111*** 0.034 0.037
Labor transfer quality 0.015*** 0.005 0.005
Control variables
Labor characteristic Amount -0.017 0.043 -0.006 -0.042 0.043 -0.014 -0.134** 0.056 -0.044
Average age -0.001 0.006 -0.000 0.003 0.006 0.001 0.0005 0.006 0.000
Education level -0.037 0.044 -0.012 -0.034 0.044 -0.011 -0.022 0.044 -0.007
Health status 0.112 0.165 0.037 0.079 0.164 0.026 0.085 0.164 0.028
Elderly labor proportion 0.277 0.199 0.091 0.198 0.197 0.065 0.208 0.198 0.069
Village cadre or not 0.068 0.098 0.022 0.126 0.099 0.041 0.021 0.101 0.007
Income characteristic Total income -0.124 0.118 -0.041 -0.133 0.117 -0.044 -0.166 0.121 -0.054
Terrace endowment characteristic Farmland transfer -0.115 0.072 -0.038 -0.127* 0.072 -0.042 -0.118* 0.072 -0.039
Farmland plots 0.002 0.001 0.001 0.002 0.001 0.001 0.002 0.001 0.001
Farmland area 0.025** 0.012 0.009 0.027** 0.012 0.009 0.028** 0.012 0.009
Farmland soil quality -0.631*** 0.06 -0.208 -0.627*** 0.06 -0.205 -0.631*** 0.06 -0.208
Substitution degree of labor input Mechanization degree of agricultural production -0.469*** 0.086 -0.155 -0.469*** 0.086 -0.154 -0.463*** 0.086 -0.152
Regional characteristic Regional variable Yes Yes Yes
Wald chi2 224.045 228.594 225.200
Prob > chi2 0.000 0.000 0.000
No of samples 1438 1438 1438
As result of Model 1-1 shown in Table 2, coefficient of non-agricultural employment proportion on whether abandon terrace is 0.284, with a significant positive impact at the 10% level. This confirms that the more labor transfer to non-agricultural activities, the less participate in agricultural production, and the more constrained agricultural production scale and output will be. According to the marginal effects analysis, with every 1 increase in labor transfer to non-agricultural employment, the probabilities of rural households abandon terrace will increase by 0.934%.
The results of Model 1-2 show that the variable labor transfer distance has a positive effect on terrace abandonment decision and the estimated coefficients are statistically significant at the 1% level. Results of Model 1-2 indicates that with every 1 unit increase in labor transfer distance, the probabilities of rural households abandon terrace will increase by 0.037%. Thus it can be seen that the farther labor transfers distance, the higher the opportunity cost for rural labors to combine agricultural production and non-agricultural employment. If the work distance is far, rural labors are more inclined to engage in full-time non-agricultural employment, the probabilities of rural households abandon terrace will increase
The results of Model 1-3 show the significant positive impact of labor transfer quality on terrace abandonment decision at 1% level, and the coefficient is 0.015. The results are consistent with the theoretical analysis, as non-agricultural working time increases, agricultural labor transfer quality improves, and labor and time input into agricultural production will be greatly reduced.
In addition, sample data shows that the majority of transfer labors are young and middle-aged, while the aging trend of agricultural labor is obvious. In order to address the problem of insufficient household agricultural labor force, farmers will adjust planting scale to achieve the optimal land management scale and labor allocation, which will lead to the emergence of farmland abandonment. According to the marginal effects analysis, with every 1 month increase in non-agricultural working times, the probabilities of rural households abandon terrace will increase by 0.005%.
Compared results of models 1-1, 1-2 and 1-3, among control variables, terrace endowment characteristic (including farmland transfer situation, contracted farmland area, and soil quality) and substitution degree of labor input have significant impact on rural households’ abandonment decisions.
From the regression results, the degree of labor substitution has a significant negative impact on farmers’ abandonment decisions and passed the test at a significance level of 1%. For every additional unit of labor substitution degree, the possibility of farmers abandoning their land decreases by 0.15%. The degree of labor substitution reflects the mechanization level of a rural household. The lower the mechanization degree, the higher the probabilities of rural households abandon terrace. Similarly, if the soil quality is low, the farmland output will be less, and the probabilities of abandonment will be higher. From the perspective of farmland rent, the more farmland transfer, the lower the probabilities of terrace abandonment, result shows that farmland rent can effectively decrease terrace abandonment.
The degree of labor substitution has a significant negative impact on household’s terrace abandonment decision. Using agricultural machinery can reduce the labor needs of agricultural production and alleviate the constraints of agricultural labor supply and thus reduce terrace abandonment.
Overall, explanatory variables reflecting labor transfer differences all have a significant positive impact on farmers’ abandonment decisions. The Probit regression results verify Hypothesis 1. Besides core independent variables, rural households’ labor characteristic also has a significant impact on household’s terrace abandonment decision. On the whole, factors affecting terrace abandonment decisions in hill and mountain area mainly from two aspects: labor transfer and farmland endowment.
From 3 variables of labor transfer difference, then significance of labor transfer quantity is lowest. In order to verify theoretical hypothesis, this research introduces cross terms of labor transfer quantity and quality and cross terms of labor transfer quantity and distance into regression model to further determine impact of labor transfer quantity.
As results of Model 1-4 shown in Table 3, coefficient for cross term between labor transfer quantity and distance is 0.12, with a significant positive impact on household terrace abandonment decision at the 1% level. This result indicates labor transfer distance will exacerbate impact of labor transfer quantity on farmer’s abandonment decision, when determining the impact of labor transfer quantity on abandonment decision, it is necessary to consider influence of labor transfer distance on labor transfer quantity.
Table 3 Analysis of households’ terrace abandonment decisions by introducing cross-terms
Variables Dependent variable: Whether household abandon terrace (0=no;1=yes)
Probit model 1-4 Probit model 1-5
Coefficients Robust standard error Coefficients Robust standard error
Core independent variables
Labor transfer quantity 0.012 0.178 0.029 0.203
Labor transfer quantity × Labor transfer distance 0.120*** 0.042
Labor transfer quantity × Labor transfer quality 0.350** 0.202
Controls Yes Yes
Control regional characteristic Yes Yes
Prob > chi2 0.000 0.000
No of samples 1438

Note: Statistical significance at the 0.1, 0.05 and 0.01 levels is denoted by *, ** and * **, respectively.

When the labor of rural households is mostly working locally, it is easier to balance agricultural production and non-agricultural work, and the probability of terrace abandonment is lower. On the opposite, when working distance is far, participating agricultural production will reduce non-agricultural labor income, making it difficult for farmers to balance agricultural production and non-agricultural employment. Therefore, labors with closer transfer distance are less constrained than labors with farther transfer distance.
As results of Model 1-5 shown in Table 3, consistent with labor transfer distance, labor transfer quality also exacerbates the impact of labor transfer quantity on household terrace abandonment decision. The coefficient for cross term between labor transfer quantity and quality is 0.35, with a significant positive impact on household terrace abandonment decision at the 5% level.
When labor transfer occurs, rural household that spend more time on non-agricultural employment input less labor in agricultural production, and are more constrained by labor shortage and less likely to balance agricultural production and non-agricultural employment. On the contrary, households with labor transfers but spend less time on non-agricultural employment are less constrained because of the unstable non-agricultural employment. Mostly these farmers only do odd jobs during leisure time, which will not reduce agricultural labor input and abandon terrace.
According to the model results after introducing cross terms, Hypothesis 2 that labor transfer distance and labor transfer quality exacerbate the impact of the labor transfer quantity on households’ terrace abandonment decisions was verified.
When agricultural labor transfer happens, differences in transfer distance and quality will cause different level of labor quantity constraint on agricultural production. If we only analyze the impact of whether labor transfer happens or the labor transfer proportion on the rural households’ abandonment decision, while ignoring the differences in transfer distance and transfer quality, results will be unilateral and biased. To some extent, this research explains why different scholars have inconsistent research conclusions when analyzing impact mechanism of labor transfer on households’ abandonment decision.

4.2.2 Impact of agricultural labor transfer differences on households’ terrace abandonment scale

As method stated above, we used Tobit models 2-1, 2-2 and 2-3 to verify the impact of agricultural labor transfer differences on rural households’ terrace abandonment scale.
Table 4 shows estimation results for impact of agricultural labor transfer differences on terraced field abandonment rate. From the fitting results of the model, it can be seen that the LR values are significant at the 1% level, indicating a good fit of the model and reliable explanation.
Table 4 Tobit model regression results of the impact of labor transfer differences on households' terrace abandonment scale
Variables Dependent variable: Households’ terrace abandonment rate
Tobit model 2-1 Tobit model 2-2 Tobit model 2-3
Coefficients Robust standard error Coefficients Robust standard error Coefficients Robust standard error
Core independent variables
Agricultural labor transfer differences Labor transfer quantity 0.218*** 0.065
Labor transfer distance 0.068*** 0.015
Labor transfer quality 0.010*** 0.003
Control variables
Labor characteristic Amount -0.016 0.020 -0.034* 0.019 -0.100*** 0.027
Average age -0.002 0.003 -0.001 0.003 -0.002 0.003
Education level -0.016 0.021 -0.013 0.021 -0.005 0.021
Health status 0.107 0.077 0.079 0.077 0.083 0.077
Elderly labor proportion 0.198** 0.096 0.146 0.093 0.148 0.094
Village cadre or not -0.035 0.043 0.003 0.043 -0.068 0.044
Income characteristic Total income -0.108* 0.059 -0.099* 0.057 -0.132** 0.059
Terrace endowment
characteristic		 Farmland transfer -0.111*** 0.034 -0.119*** 0.034 -0.113*** 0.034
Farmland plots 0.001*** 0.0004 0.001** 0.0004 0.001** 0.0003
Farmland area 0.008* 0.004 0.009** 0.005 0.009** 0.004
Farmland soil quality -0.245*** 0.027 -0.24*** 0.026 -0.245*** 0.027
Substitution degree of labor input Mechanization degree of agricultural production -0.244*** 0.043 -0.244*** 0.042 -0.240*** 0.043
Regional characteristic Regional variable Yes Yes Yes
F 16.23 17.35 16.56
Prob > F 0.000 0.000 0.000
No of samples 1438 1438 1438

Note: Statistical significance at the 0.1, 0.05 and 0.01 levels is denoted by *, ** and * **, respectively.

Consistent with the result of probit model, core explanatory variables of the quantity, distance and quality of agricultural labor transfer have a significant positive impact on households’ terrace abandonment, the higher the quantity, distance, and quality of agricultural labor transfer among rural households, the higher the terrace abandonment rate, this result further confirming research hypothesis 1.
Coefficient of agricultural labor transfer quantity on terrace abandonment rate is 0.218, with a significant positive impact at the 1% level. Agricultural labor transfer quantity not only affects households’ terrace abandonment decision but also has significant positive impact on their abandonment scale, the higher the proportion of non-agricultural employment in rural households, the larger the terrace abandonment scale.
Coefficient of agricultural labor transfer distance on terrace abandonment rate is 0.068, with a significant positive impact at the 1% level. This result confirms that the increase in agricultural labor transfer distance will increase the opportunity cost of engage in agricultural production, terraces will be abandoned and as the distance increases, the terrace abandonment scale will expand. Coefficient of agricultural labor transfer distance on terrace abandonment rate is 0.010, with a significant positive impact at the 1% level. This result confirms that the quality of non-agricultural employment also affects terrace abandonment scale. Farmers with high quality non-agricultural employment no longer rely on agricultural income, so they will abandon more terraces. Compared with Tobit models 2-1 2-2 and 2-3, control variables as elderly labor proportion, total income, farmland transfer situation, terrace plots, farmland area, farmland soil quality and substitution degree of labor input have an impact on farmer’s abandonment scale.
Based on the estimation results of labor characteristics variables, correlations between elderly labor proportion and terrace abandonment rate are significant at 5% level. The abandonment rate of rural households with more elderly labor is higher, which confirms that the aging of rural household labor decrease available labor of rural household, constraining labor amount that invest in agricultural production, and enlarge terrace abandonment scale.
Based on the estimation results of farmland transfer variable, negative correlation between farmland transfer and terrace abandonment rate are significant. Households with farmland transfer abandoned less terrace, which shows that farmland transfer can effectively decrease rural households’ abandonment behavior.
The substitution degree of labor input has a significant negative impact on terrace abandonment. The use of agricultural machinery can reduce labor input in agricultural production, alleviate the supply constraints of agricultural labor, and enable farmers to cultivate more farmland. However, in hilly and mountainous areas with complex terrain, mechanical input cannot completely replace the role of labor input in farmland production (Sun, 2021), which also reflected in the results of the variable of terrace endowment characteristic. Variables such as terrace plots, soil quality and contracted farmland area explain the impact of farmers’ terrace endowment characteristic on abandonment decision. In recent years, the wages of non-agricultural employment have rapidly increased. In order to maximize income, rural households have increased the quantity of agricultural labor transfers, and the time and quality of non-agricultural employment have also become increasingly stable.
However, in hilly and mountainous areas where terrace are concentrated, the investment and difficulty of farmers in utilizing agricultural machinery and participating in agricultural socialized services are limited by the terrace endowment characteristic, making it insufficient to replace the labor input required for terrace production. Therefore, the agricultural labor transfer greatly increases the possibility of terrace abandonment.
In summary, the results of Tobit and Probit models are consistent, the quantity, distance and quality of agricultural labor transfer have a significant positive impact on rural households’ terrace abandonment scale. The higher the quantity, the farther the distance, and the higher the quality of agricultural labor transfer, the higher the terrace abandonment rate, which further confirms the research Hypothesis 1.

4.3 Endogeneity and robustness

4.3.1 Endogeneity

When identifying impact of rural households’ agricultural labor transfer on terrace abandonment decision and scale, it is necessary to consider possible endogeneity problems in models. In this research, the labor allocation decisions of rural households often occur with the abandonment decisions simultaneously, and there may be potential interactions between labor transfer and rural abandonment (Lu, 2020). Therefore, it is necessary to consider potential endogeneity and effective way to deal with endogeneity is to introduce appropriate instrumental variables into the models (Yuan, 2018).
Theoretically, an appropriate instrumental variable should be highly correlated with endogenous explanatory variables and not correlated with the random Error term of the model. Based on the new economic theory of labor migration, agricultural labor transfer not only aims to increase absolute income for family but also to reduce relative poverty within certain group or reference system (Stark, 1985). Generally speaking, farmer’s non-agricultural employment decision will be influenced by employment situation of the same group in the local area, as non-agricultural employment of left-behind elderly and women farmers will be influenced by local workers or relatives from the same village (Liu and Liang, 2021). Refers to Lu et al. (2017), this research selected the mean time of non-agricultural employment of other rural households in their village as the instrumental variable of a certain rural household. If there are N households in Village X, the instrumental variables for household Li-th are as follows.
$V{{T}_{X{{L}_{i}}}}=\underset{L\ne {{L}_{i}}}{\mathop \sum }\,{{T}_{L}}\div \left( N-1 \right)$
As shown in Table 5, the Wald test results for the original hypothesis are significant at 5% level. It can be seen that the core variables and instrumental variables are endogenous explanatory variables. If a general Probit or Tobit model is used for estimation at this time, the endogeneity will be ignored.
Table 5 Measurement results of instrumental variable models
Variables Model 3-1 Model 3-2 Model 3-3
IV-Probit IV-Tobit IV-Probit IV-Tobit IV-Probit IV-Tobit
Instrumental variable
Labor transfer quantity 1.750***
(0.567)		 0.984***
(0.312)
Labor transfer distance 0.746**
(0.309)		 0.376***
(0.129)
Labor transfer quality 0.053**
(0.20)		 0.027***
(0.008)
Wald-test 5.53** 7.02** 5.62** 7.22** 4.09** 4.51**
P-value of Wald endogeneity test 0.0186 0.0080 0.0178 0.0072 0.0431 0.0337
Constants 0.475 0.499 -0.014 0.202 2.269 1.435
Weak instrumental variable test
Weakiv-wald 6.45*** 8.35*** 5.8** 7.16*** 6.96*** 9.00***
Weakiv-AR 7.11*** 9.23*** 7.17*** 9.12*** 7.2*** 9.25***
No of samples 1438 1438 1438
As baseline regression of Probit model shown in Table 2, three core independent variables have a significant positive impact on households’ terrace abandonment decision. The coefficient for labor transfer quantity is 0.284 and is significant at the 10% level; the coefficient for labor transfer distance is 0.111 and is significant at the 1% level; the coefficient for labor transfer quality is 0.015 and is significant at the 1% level.
However, Endogeneity leads to underestimation of positive impact of core independent variables on households’ abandonment decisions. As baseline regression of Tobit model shown in Table 4, three core independent variables have significant positive impact on households’ terrace abandonment rate at 1% level, which is consistent with IV-Tobit regression results in Table 5. From Tobit model baseline regression, the coefficient for labor transfer quantity is 0.218, the coefficient for labor transfer distance variable is 0.068 and coefficient for labor transfer quality variable is 0.010. Similarly, endogeneity leads to underestimation of positive impact of core independent variables on households’ abandonment rate.
Table 5 also gives results for weak instrumental variable test where Wald and AR values both pass significance level test, which means rejecting original hypothesis that “endogenous variable and instrumental variable are not correlated”. This means that instrumental variable chosen in this research is not weak instrumental variable. In summary, regression results for instrumental variable show that coefficients for labor transfer quantity, distance and quality have positive significant impact on households’ terrace abandonment decision and scale. This proves that the quantity, distance, and quality of agricultural labor transfer have a causal relationship with the decision and scale of terrace abandonment.
The results of baseline regression and instrumental variable models further confirm the research hypothesis, the greater the quantity, the farther the distance, and the longer the time of households’ agricultural labor transfer, the greater the probability and scale of terrace abandonment.

4.3.2 Robustness

Previous section estimated impact of labor transfer quantity, distance and quality on terrace abandonment. Baseline estimation results for Probit model and Tobit model are basically consistent that differences in quantity, quality and distance of agricultural labor transfer have certain impact on terrace abandonment. Instrumental variable method solved possible endogeneity problem and confirmed that empirical results are credible.
In order to further confirm reliability of regression results, this research attempts to confirm robustness by replacing core independent variables and dependent variables. In the baseline regression, this research uses whether household abandon terrace and the terrace abandonment rate as the dependent variables to characterize farmers’ terrace abandonment behavior.
In order to verify robustness of model, this section replace dependent variable with terrace abandonment area to examine impact of agricultural labor transfer differences on terrace abandonment scale.
Tobit models 5-1 5-2 5-3 results are shown in Table 6, where terraced field abandonment area replaces terraced field abandonment rate.
Table 6 Model robustness test of core variable replacement
Variables Dependent variable: Whether household abandon terrace (0=no;1=yes) Replaced dependent variable: Households’ terrace abandonment area
Probit model 4-1 Probit model 4-2 Tobit model 5-1 Tobit model 5-2 Tobit model 5-3
Coefficients Coefficients Coefficients Coefficients Coefficients
Replaced core independent variables
Labor transfer quantity 0.106*
(0.057)		 0.386**
(0.167)
Labor transfer quality 0.254*
(0.140)		 0.725*
(0.400)
Original core independent
variables
Labor transfer distance 0.365***
(0.100)
Control variables
Labor characteristics Amount -0.099*
(0.056)		 -0.047
(0.043)		 -0.364**
(0.164)		 -0.160
(0.123)		 -0.155
(0.121)
Average age -0.001
(0.006)		 -0.002
(0.006)		 -0.008
(0.019)		 -0.012
(0.018)		 0.004
(0.018)
Education level -0.036
(0.044)		 -0.041
(0.044)		 -0.091
(0.131)		 -0.112
(0.131)		 -0.081
(0.130)
Health status 0.098
(0.164)		 0.128
(0.165)		 0.280
(0.482)		 0.360
(0.486)		 0.199
(0.480)
Elderly labor proportion 0.263
(0.198)		 0.247
(0.198)		 0.961*
(0.571)		 0.905
(0.568)		 0.756
(0.563)
Village cadre or not 0.065
(0.098)		 0.083
(0.098)		 0.038
(0.277)		 0.095
(0.275)		 0.244
(0.278)
Income characteristic Total income -0.114
(0.117)		 -0.121
(0.120)		 -0.418
(0.346)		 -0.411
(0.353)		 -0.454
(0.344)
Terrace endowment characteristic Farmland transfer -0.117
(0.072)		 -0.103
(0.073)		 -0.408*
(0.210)		 -0.373*
(0.211)		 -0.441**
(0.210)
Farmland plots 0.002
(0.001)		 0.002
(0.001)		 0.007*
(0.004)		 0.007*
(0.004)		 0.007*
(0.004)
Farmland area 0.026**
(0.012)		 0.028**
(0.012)		 0.136***
(0.037)		 0.140***
(0.038)		 0.138***
(0.037)
Farmland soil quality -0.630***
(0.060)		 -0.625***
(0.060)		 -1.617***
(0.199)		 -1.599***
(0.198)		 -1.597***
(0.197)
Substitution degree of labor input Mechanization degree of agricultural production -0.474***
(0.086)		 -0.480***
(0.086)		 -1.374***
(0.275)		 -1.401***
(0.276)		 -1.354***
(0.274)
Regional characteristics Regional
variable		 Yes Yes Yes Yes Yes
Chi-square /F value 223.857 223.892 7.249 7.230 7.352
Prob > chi2/ Prob > F 0.000 0.000 0.000 0.000 0.000
No of Samples 1438 1438 1438 1438 1438
As shown in Table 6, to core independent variables, another measurement indicator was adopted referring to existing research to test robustness. Number of households’ transferred labor was used to measure labor transfer quantity instead of transferred labor proportion, proportion of households’ non-agricultural income was used as measurement indicator labor transfer quality instead of average working times of non-agricultural employment. Because of no suitable replacement index for labor transfer distance, the explained variable was replaced in Tobit model to test robustness.
From the results of baseline regression and robustness test, it can be seen that significance level and influence direction for core independent variables are consistent. Among control variables, elderly labor proportion, farmland transfer situation, farmland plots, contracted farmland area, soil quality and substitution degree of labor input have an impact on households’ terrace abandonment decision and terrace abandonment scale, which fits well with baseline regression results and confirms robustness.

5 Conclusion and policy implications

In rural households, the agricultural transfer of a single labor does not represent the withdrawal of the entire family from agricultural production activities. Existing researches regard the individual labor transfer as the complete departure of households from agriculture, ignoring the differences in agricultural transfer of farming households. This research uses 1438 survey data from rural households in Hunan, Fujian and Jiangxi provinces to comprehensively consider the differences in the quantity, distance and quality of agricultural labor transfer, analyzes the households’ decision and scale of terrace abandonment and draws the following conclusions.
(1) The quantity, distance and quality of agricultural labor transfer of rural households all have a significant positive effect on households’ decision and scale of terrace abandonment. Consistent with the research conclusions of Xie and Huang, (2022) the agricultural labor transfer quantity has a significant positive impact on households’ decision-making on terrace abandonment, with every 1 increase in labor transfer to non-agricultural employment, the probabilities of rural households abandon terrace will increase by 0.934%. Agricultural labor transfer distance has a significant positive effect on terrace abandonment decision and scale at the 1% level. The farther the agricultural labor transfer distance, the higher the opportunity cost for labors to take into agricultural production, thereby increasing the occurrence of terrace abandonment. Similarly, the higher the quality of agricultural labor transfer, the higher the probability that terrace will be abandoned and the larger the abandonment scale, with every 1 month increase in non-agricultural working times, the probabilities of rural households abandon terrace will increase by 0.005%.
(2) The distance and quality of agricultural labor transfer will exacerbate the impact of labor transfer quantity on rural households’ abandonment decisions. Different from previous researches, this research introduces cross terms of labor transfer quantity and quality and cross terms of labor transfer quantity and distance to analyze the accurate impact of the number of labor transfers on farmers’ abandonment decisions. The results show that the distance and quality of agricultural labor transfer will intensify the impact of the quantity of labor transfers on households’ decision-making on terrace abandonment. When agricultural labor transfer happens, differences in transfer distance and quality will cause different level of labor quantity constraint on agricultural production. If we only analyze the impact of whether labor transfer happens or the labor transfer proportion on the rural households’ abandonment decision, while ignoring the differences in transfer distance and transfer quality, results will be unilateral and biased. To some extent, this research explains why different scholars have inconsistent research conclusions when analyzing impact mechanism of labor transfer on households’ abandonment decision.
Based on the above conclusions, this research obtains the following policy implications.
Firstly, the government can promote commercialized agricultural service system in hilly and mountainous area. The reason why agricultural labor transfer induces terrace abandonment is that labor transfer has a strong constraint on agricultural labor supply in mountainous areas, agricultural service system and mechanization can replace labor input and alleviate terrace abandonment caused by seasonal agricultural labor transfer. The government can actively support agricultural professional cooperation organizations, develop agricultural service organizations and markets to provide field management links such as seedling, breeding and pest control service, therefore terrace can be transferred through agricultural professional cooperative organizations, thereby utilizing abandoned terraces and reducing abandonment.
Secondly, the government can promote the secondary development and construction of terrace through mechanization. The complex terrain conditions and fragmentation of terrace increase the difficulty of agricultural production and limit mechanized operations, resulting in a much lower level of agricultural mechanization in in hilly and mountainous areas than in plain areas. The high difficulty and cost of terrace planting led farmers lack motivation to utilize terrace. To fundamentally improve farmers’ motivation in terrace planting, the government should improve the machine plowing path and mechanical operation conditions of terrace, reduce the difficulties for farmers in planting terrace, and enable farmers to feel the “happiness” and “technology” of agricultural planting, thereby effectively alleviating the terrace abandonment in hilly and mountainous area. On the premise of ensuring farmers’ well-being, the government should take into account the actual situation of the village to avoid ineffective investment in “hollow villages” with severe population loss. Ecological restoration can further restore vegetation and reduce ecological problems such as soil erosion caused by terrace abandonment. Especially for some villages with high altitude, poor agricultural conditions, and severe labor loss, ecological restoration measures should be adopted as the main measure to deal with terrace abandonment.
Thirdly, the governments should establish an information accessible platform and establish market for farmland transfer, encourage farmers who have abandoned or have low willingness to agricultural production to transfer terrace, and help large farmers transfer into terrace and expand their business scale, thereby achieving large-scale cultivation of terrace and forming a contiguous terrace landscape. In addition, improving farmers’ understanding of farmland transfer, reducing their loss sense towards farmland transfer, and enhancing their gain sense towards farmland transfer, thereby weakening the dependency of some farmers on farmland, will help promote terrace transfer in hilly and mountainous areas, and reduce terrace abandonment.
Finally, the government should promote the development of agricultural industries in hilly and mountainous areas. With the development of the economy, it is foreseeable that more and more rural households will completely transfer to urban and non-agricultural sectors. To alleviate terrace abandonment caused by the permanent labor transfer, it is necessary to increase more non-agricultural employment opportunities and effectively increase farmers’ income.
On the one hand, we should actively support new agricultural economic entities, help introduce advanced varieties such as “colored rice” and “giant rice” with ornamental value, promote the integration of crops and terraced landscape, integrate agricultural production processes into terraced landscape, and combine agricultural planting with rural experiential tourism to form a “pre-production” humanistic experience and “mid-production” terraced landscape for terraced farming The agricultural industry chain of “post-production” agricultural value enhances agricultural production efficiency, drives the development of local terraced agricultural industry, provides local employment opportunities for farmers, slows down the trend of labor outflow, and thus alleviates farmers’ abandonment.
With the development of social economy, it can be predicted that more and more farming households will completely transfer to towns and non-agricultural sectors in a family-based migration mode. To alleviate terrace abandonment due to permanent labor transfer, it is necessary to increase more non-agricultural employment opportunities and effectively increase farmers’ income. The government should actively support new agricultural economic entities, help introduce advanced planting crops with ornamental value such as colored rice and giant rice, promote the integration of crops and terraced landscape, and integrate agricultural production processes into terrace landscape recreation. By combining agricultural production with rural tourism, a value chain of humanistic experience tourism during the irrigation phase, terrace landscape sightseeing during the crop growth phase, agricultural product sales during the harvest phase will form, which can promote the development of local agricultural industry and provide local employment opportunities for farmers, thus slow down the trend of labor transfer and alleviate terrace abandonment.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Aide T M, Grau H R, 2004. Ecology enhanced: Globalization, migration, and Latin American ecosystems. Science, 305(5692): 1915-1916. (in Chinese)

DOI

[2]
Bezu S, Holden S, 2014. Are rural youth in Ethiopia abandoning agriculture? World Development, 64(12): 259-272.

DOI

[3]
Cai F, 2001. Two processes of labor migration and its institutional obstacles. Sociological Studies, (4): 44-51. (in Chinese)

[4]
Cai F, 2018. Has China’s labor mobility exhausted its momentum? Chinese Rural Economy, (9): 2-13. (in Chinese)

[5]
Chen D, Wei W, Chen L D et al., 2016. Progress of the ecosystem services and management of terraces. Mountain Research, 34(3): 374-384. (in Chinese)

[6]
He Z H, 2005. An analytic framework of concurrent business of farmers. China Rural Survey, (1): 2-9, 78. (in Chinese)

[7]
Hou C P, 1999. The new trend in the adjustment of employment structure: A preliminary study on the phenomenon of part-time employment. Population Journal, (2): 17-22. (in Chinese)

[8]
Huang L J, Zhang D Q, 2005. The institutional basis of benign transfer of rural surplus labor force. Rural Economy, (1): 120-122. (in Chinese)

[9]
Huang M L, Li X Y, 2019. The impacts of rural labor price rising on crop structure among provinces. Economic Geography, 39(6): 172-182. (in Chinese)

[10]
Huang M L, Li X Y, You Z L et al., 2018. Impact of agricultural labor investment on grain production and its elasticity of substitution. Journal of Huazhong Agricultural University (Social Sciences Edition), (2): 37-45, 156. (in Chinese)

[11]
Huang M Q, Li Y B, Ran C H et al., 2021. Dynamic changes and transformation of agricultural landscape pattern in mountainous areas: A case study in the hinterland of the Three Gorges Reservoir area. Acta Geographica Sinica, 76(11): 2749-2764. (in Chinese)

DOI

[12]
Li S F, Li X B, 2016. Progress and prospect on farmland abandonment. Acta Geographica Sinica, 71(3): 370-389. (in Chinese)

DOI

[13]
Li Y L, Ma W Q, Jiang G H et al., 2021. The degree of cultivated land abandonment and its influence on grain yield in main grain producing areas of China. Journal of Natural Resources, 36(6): 1439-1454. (in Chinese)

DOI

[14]
Li Z J, Guo J X, 2011. The influence of selective labor transfer on the change of production preference of rural family population. China Rural Survey, (3): 40-49. (in Chinese)

[15]
Lieskovský J, Bezák P, Špulerová J et al., 2013. The abandonment of traditional agricultural landscape in Slovakia: Analysis of extent and driving forces. Journal of Rural Study, 32(2): 356-365.

[16]
Liu H Y, 2011. Economic growth effect of rural labors’ employment in non-agricultural industry in China. Population & Economics, (2): 23-27. (in Chinese)

[17]
Liu J W, Liang J Y, 2021. Life rationality: the action logic of rural women in industrial poverty alleviation with national characteristics based on the experience of “Jinxiu Plan” project in Ceheng county, Guizhou province. China Rural Survey, (2): 15-27. (in Chinese)

[18]
Lu C, 2020. Does household laborer migration promote farmland abandonment in China? Growth and Change, 51(4): 1-33.

DOI

[19]
Lu H, Hu H, 2017. Off-farm labor supply: Does land fragmentation work? Based on the perspective of Lewis turning-point. Economic Review, (1): 148-160. (in Chinese)

[20]
Ma L, Long H L, Zhang Y N et al., 2018. Spatio-temporal coupling relationship between agricultural labor changes and agricultural economic development at county level in China and its implications for rural revitalization. Acta Geographica Sinica, 73(12): 2364-2377. (in Chinese)

DOI

[21]
Ma Z D, Zhang W M, Liang Z et al., 2004. Labor migration as a new determinant of income growth in rural China. Population Research, (3): 2-10. (in Chinese)

[22]
Miao J Q, Wang Z Q, Ma Y Q et al., 2018. Comprehensive evaluation of sustainable development of the Hakka terrace ecosystem of Chongyi. Acta Ecologica Sinica, 38(17): 6326-6336. (in Chinese)

[23]
Shao H, Huang J, Zuo T Y, 2011. Analysis of balanced water supply and demand in Longji terraced field regulation of Longsheng county, Guilin. Chinese Agricultural Science Bulletin, 27(14): 227-232. (in Chinese)

[24]
Sheng L Y. 2007. The economic impact of rural labor flow. Statistical Research, (10): 15-19. (in Chinese)

[25]
Shi T, Li X, Xin L et al., 2018. The spatial distribution of cropland abandonment and its influential factors at the township level: A case study in the mountainous area of China. Land Use Policy, 70(1): 510-520.

DOI

[26]
Stark O, Bloom D E, 1985. The new economics of labor migration. American Economic Review, 75(2): 173-178.

[27]
Sun X T, 2021. The transfer of rural labor and improvement of farmland efficiency from the perspective of disequilibrium. Economic Survey, 38(5): 41-50. (in Chinese)

[28]
Tan Z P, Hong W J, Luo B L, 2019. The transfer effect of agricultural labor force and grain-oriented planting structure. Reform, (7): 111-118. (in Chinese)

[29]
Tian Y J, Li X B, Ma G X et al., 2010. Influences of labor emigration from agriculture on the production abandonment of cultivated land in ecological sensitive areas. China Land Science, 24(7): 4-9. (in Chinese)

[30]
Xiang G C, Han S F, 2005. Farmers’ part-time industrialization: An analysis from the perspective of division of labor. Chinese Rural Economy, (8): 4-9, 16. (in Chinese)

[31]
Xie H L, Huang Y Q, 2022. Impact of non-agricultural employment and land transfer on farmland abandonment behaviors of farmer: A case study in Fujian-Jiangxi-Hunan mountainous areas. Journal of Natural Resources, 37(2): 408-423. (in Chinese)

DOI

[32]
Yao Y F, Wang L X, 1991. Formation and development of terrace in China. Soil and Water Conservation in China, (6): 56-58. (in Chinese)

[33]
Yuan W, 2018. Endogenous test methods, procedures and Stata applications of binary selection model. Statistics & Decision, 34(6): 15-20. (in Chinese)

[34]
Zhang B L, Yang Q Y, Yan Y et al., 2011. Characteristics and reasons of different households’ farming abandonment behavior in the process of rapid urbanization based on a survey from 540 households in 10 counties of Chongqing municipality. Resources Science, 33(11): 2047-2054. (in Chinese)

[35]
Zhang K Z, Gao Z B, Zhang L N et al., 2020. Migratory-bird flow or permanent migration? Public services accessibility and population migration options based on the perspective of social integration. Research on Economics and Management, 41(7): 112-133. (in Chinese)

[36]
Zhang X Z, Zhao C S, Dong J W et al., 2019. Spatio-temporal pattern of cropland abandonment in China from 1992 to 2017: A meta-analysis. Acta Geographica Sinica, 74(3): 411-420. (in Chinese)

DOI

[37]
Zhang Y, Li X B, Song W, 2014a. Determinants of cropland abandonment at the parcel, household and village levels in mountain areas of China: A multi-level analysis. Land Use Policy, 41(11): 186-192.

DOI

[38]
Zhang Y, Li X B, Song W et al., 2014b. Effect of agricultural laborer on cropland abandonment under land circulation at different levels in Wulong county, Chongqing city. Progress in Geography, 33(4): 552-560. (in Chinese)

[39]
Zhang Y X, Min Q W, 2016. A review of conservation of rice terraces as agricultural heritage systems. Chinese Journal of Eco-Agriculture, 24(4): 460-469. (in Chinese)

[40]
Zhao S C, Xu Q, Liu J, 2020. Labor migration, capital deepening, and agricultural land circulation in China. Journal of Agrotechnical Economics, (3): 4-19. (in Chinese)

[41]
Zhao W Y, Li Y B, Li Y Y et al., 2018. Evolution and landscape pattern of abandoned land in center region of Three Gorges Reservoir area: A case study in Pingan and Zhuyuan township of Fengjie county, Chongqing municipality. Bulletin of Soil and Water Conservation, 38(3): 256-262. (in Chinese)

[42]
Zhu W J, Luo B L, 2020. Labor migration, gender differences, farmland transfer and its contract selection. China Population, Resources and Environment, 30(1): 160-169. (in Chinese)

Options
Outlines

/

〈 〉