The current water shortage crisis has necessitated an increased focus on improving the irrigation efficiency in groundwater overdraft areas. Consequently, the Chinese government has supported small farmers in installing community-based water-saving technologies (WSTs) providing high irrigation efficiency. Based on the data collected from 620 households located in the groundwater overdraft area of Hebei, North China Plain, this study conducts a stochastic frontier analysis to measure farmers' irrigation water use efficiency (WUE) and analyzes the impact of land fragmentation and WST types on their WUE. The results show that the average WUE of groundwater irrigation is 0.606. The WUE between community-based and household-based WSTs differs based on the degree of land fragmentation. A high degree of land fragmentation restricts community-based WSTs from efficiently improving farmers' WUE, whereas household-based WSTs perform better and are easier to adopt. For high land fragmentation, the WUE of the community-based WST is 9.12% lower than that of the household-based WST. However, the WUE of the community-based WST is 12.55% higher than that of the household-based WST when the degree of land fragmentation is low. Therefore, the government should pay attention to small farmers' adaptability toward WST and promote community-based WSTs on a low degree of land fragmentation.

  • This study calculated farmers’ irrigation water use efficiency (WUE) and analyzes the determinants of it.

  • This study constructs a comprehensive insight to explain the phenomenon that community-based (water-saving technology) did not perform well in improving WUE.

  • In addition to top-down financial support, more attention should be paid to the needs of small farmers and their actual agricultural production conditions.

Hebei Province experiences severe water shortage with an average annual rainfall of 532 mm and a total amount of water and per capita water resources of 307 m3. This is far below the internationally recognized 500 m3 of ‘extreme water shortage standard.’ The total annual water supply of Hebei Province is approximately 19 billion m3, 75% of which is collected from groundwater exploitation. The agricultural sector accounts for 70% of the water use. Hebei Province with a total irrigation area of approximately 4.49 million ha is one of the major grain-producing provinces in China, and 63% of the irrigated area is irrigated by groundwater (Yu et al., 2020). The agricultural use of water for a long time has led to excessive groundwater exploitation, resulting in the deterioration of water quality, surface subsidence, seawater intrusion, and other ecological and environmental problems.

In addition, groundwater vulnerability is affected by climate change. Climate change can directly alter groundwater vulnerability through interactions with surface water, net recharge, and groundwater levels (Hua et al., 2015; Zhou et al., 2020). The scourge of climate change evidenced and driven by global warming will decrease groundwater recharge, storage, and levels (Amanambu et al., 2020). With the available groundwater resources becoming scarce with an increasing temperature, there is an urgent need to improve irrigation efficiency (Gou et al., 2020). Farmers also need to take adaptive measures by increasing irrigation frequency and expanding irrigated areas to mitigate production risks resulting from extreme drought events (Zhang et al., 2017; Wang et al., 2018, 2019a, 2019b; Zhang et al., 2020). Rapid urbanization and industrialization will increasingly use more water and decreasing water resources will force agriculture to reduce its water usage. Therefore, improving the irrigation efficiency in China under climate change is of utmost priority.

Policymakers in China have been focused on reducing groundwater overuse. The North China Plain has a large proportion of groundwater exploitation, and agriculture is the largest sector of water use; therefore, it is particularly important to reduce the water consumption for irrigation and improve the efficiency of agricultural irrigation in this area. The State Council issued the ‘National Water Saving Irrigation Plan,’ which pointed out that by 2020, the irrigation area with a water-saving irrigation project should reach 80% of the effective irrigation area of the country. The policy supports farmers in adopting community-based water-saving technologies (WSTs) (such as sprinklers and micro-irrigation). However, at present, the utilization rate of community-based WST is significantly low, less than 10% of the irrigated land area (Wang et al., 2020c).

Land fragmentation has existed in China since the implementation of the household contract responsibility system in the 1980s. The average area per household is approximately half a hectare (Wang & Li 2020). Community-based WST not only requires larger investments, but also covers large areas and cannot be adopted by individual farmers. The implementation of community-based WST for small farmers relies mainly on top-down government support and investment. Due to land fragmentation, farmers' irrigation needs are heterogeneous. In this context, there is a difference in the impact of land fragmentation on water use efficiency (WUE). Community-based WST is very important for increasing irrigation water consumption. It is interesting to understand the performance of community-based WSTs in the context of land fragmentation.

Some scholars believe that land fragmentation is an important factor hindering the improvement of WUE. Manjunatha et al. (2013) used a stochastic frontier model to analyze the impact of land fragmentation on the irrigation efficiency of 90 groundwater irrigation farms in the hard rock area of southern India and found that land fragmentation had a significant negative impact on irrigation efficiency. Compared with small farmers, land fragmentation has a greater impact on the efficiency loss of large-scale farmers. Tang et al. (2015) studied farmers' surface water irrigation behavior in the Guanzhong Plain and found that land fragmentation has a negative impact on WUE. Pereira & Marques (2017) reviewed and analyzed the literature on agricultural irrigation WUE and concluded that land fragmentation is not conducive to improving WUE. Hassan (2017) studied the water usage of farmers in the Nile Delta and found that land fragmentation increased farmers' irrigation time and negatively affected their WUE. Based on the stochastic frontier model, Liu et al. (2018) studied WUE and its influencing factors in wheat farmers in Guanzhong and found that the number of land plots negatively affects the WUE. Sedat & Zeki (2018) studied the situation of agricultural land in Turkey and found that small, broken, and scattered agricultural lands have led to a decline in agricultural productivity. Ali et al. (2019) studied the corn production behavior of 510 farmers in the Khyber Pakhtunkhwa area, Pakistan, and analyzed the technical efficiency using a stochastic frontier model. It was found that land fragmentation has a significant negative impact on the technical efficiency of farmers' production. Dogaru et al. (2019) studied corn planting in the lower Danube plain of southern Romania over the past 15 years and found that irrigation water consumption increases in the case of land fragmentation. Based on the data obtained from 350 households in Cangxian County, Hebei Province, Wang & Li (2020) studied the impact of land fragmentation on grain production efficiency and concluded that land fragmentation negatively affects the grain production efficiency of farmers. Xu & Chen (2020) studied the influencing factors of groundwater irrigation WUE in Tianjin and Hebei and found that the number of land plots negatively affects the WUE.

It is the main consensus to improve irrigation WUE by adopting efficient WST. According to Wang et al. (2020a), the low adoption rate of community-based WST can be attributed to low water prices and the lack of farmers' incentives to save water. Xu & Liu (2013) believe that land fragmentation also inhibits farmers from adopting the appropriate WST.

As mentioned above, the government promotes the construction of water-saving facilities from top to bottom to control for groundwater over-exploitation. In our research area, the adoption of community-based WST is supported by the government project system rather than initiative in the case of small farmers; hence, the impact of land fragmentation on the performance of community-based WST among farmers is not clear. Current cases of land fragmentation are the result of the household responsibility system. This system requires plots to be divided equally according to fertility and size; land fragmentation is an exogenous factor that is difficult to change quickly. Can the community-based WST supported by the government be suitable for small farmers with fragmented land?

The overall aim of this study is to answer the questions mentioned above and fill the existing gap in the literature. To pursue this goal, we define three specific objectives. First, we employ stochastic frontier analysis (SFA) models to calculate farmers' WUE. Second, we identify the impact of land fragmentation on WUE, focusing on two types of WSTs. Third, we compare the direct and indirect effects of land fragmentation on WUE to ensure the rationality of the model setting. Finally, the research question was discussed with greater insight. And several important policy implications are highlighted for promoting improved WUE and water saving.

Stochastic frontier model

At present, data envelopment models (DEA) and SFAs are commonly used in production efficiency analysis. SFA considers the influence of random factors. Agricultural production is easily affected by the weather, natural disasters, environment, and other random events. The estimation of agricultural production efficiency using SFA is more stable. The SFA model is suitable for micro data, whereas the DEA model is more suitable for macro data estimation. Therefore, this study estimates the WUE of farmers based on the SFA model. To study the technical efficiency of production using the stochastic frontier model, it is necessary to set the production function. The common production functions are the Cobb Douglas, generalized production, transitional, and translog functions. Referring to Tang et al.'s (2013) study on the measurement of agricultural irrigation WUE, this study employed a stochastic frontier production model in the form of a translog. It can be expressed as follows:
formula
(1)
where is wheat output. The four inputs include , water use; , labor ; , fertilizer ; and , agriculture machine use . Here, is the standard error term, and the non-negative error term is , which follows a half-normal distribution reflecting the shortfall of a farmer's output from the production frontier due to technical inefficiency.
Under the same output, the optimal condition is that the technical inefficiency item is zero, and the water use is optimal; WUE can be expressed as .
formula
(2)
WUE can be expressed through Equations (3) and (4):
formula
(3)
formula
(4)
From Equations (1)–(4), we get (5) as follows:
formula
(5)
From Equation (5), for the individual farmer i can be obtained as follows:
formula
(6)
formula
(7)
The output elasticity of farmers' water use can be expressed via Equation (8); the output elasticity of other inputs can be expressed via Equation (9); and the farmers' returns to scale can be expressed via Equation (10).
formula
(8)
formula
(9)
formula
(10)

Specification of the econometric model

To better identify the impact of land fragmentation and WST on WUE, the following econometric model was developed:
formula
(11)

In model (11), indicates the WUE of the individual farmer i measured by using the SFA model; as the actual water input is always greater than or equal to the potential optimal input, its value must lie between 0 and 1.

The right-hand side of the model includes the following independent variables. is the key independent variable; we are interested in land fragmentation, which is measured by the reciprocal of the average plot size of the farmer. is a dummy variable that measures whether a farmer adopts community-based WST (1 = community-based WST; 0 = household-based WST). is the set of control variables.

Data and variables

The data used in this study were collected from a survey conducted by our research team in 2018. The survey collected information on agricultural production; farmers' income and expenses; and details on irrigation, such as construction, frequency, and cost. The survey used two questionnaires: one for the village level and one for the households. The former was mainly filled in by a person, such as the village head or director, familiar with the situation prevalent in the village. It covered basic information on the village, tubewell, irrigation and water conservancy, infrastructure, and public services. Due to the continuous shutdown of tubewells by the government to control the over-exploitation of groundwater in recent years, the number of tubewells in each village has become limited. Data on tubewells, such as the depth and head of the pumps, were mainly obtained in this part of the study, which is critical for the calculation of water consumption. The latter questionnaire was filled in by household heads, who were randomly selected by the interviewer; it covered basic information on the number of family members and their living environment, land and house property, irrigation, consumption, expenses, and so on. This survey covered a wide range of issues, and only the data relevant to this study were analyzed.

In this study, Nanpi County and Xian County of Cangzhou City, Cheng'an County of Handan City, and Yuanshi County of Shijiazhuang city were selected as the research areas. The areas are highly dependent on groundwater irrigation. A random stratified sampling method was used. Additionally, 6–10 townships were randomly selected from each county, and 2–6 villages were randomly selected from each township. In each sample village, the investigators asked the village leaders to provide a list of names for all the household heads. The first household was selected randomly, and the other households were selected according to the calculated interval distance (total number of households/number of households to be selected). Approximately 5–15 farmers were randomly selected from each village. A total of 620 valid questionnaires were obtained from our research samples.

The average land owned by farmers in the survey area was 7.08 mu (1 mu = 0.067 ha), and the number of plots was 3.23. The plot number distribution map of the research area is shown in Figure 1. The irrigation water consumption distribution map of wheat is shown in Figure 2.

Fig. 1.

The distribution of land plots owned by rural households in the survey area.

Fig. 1.

The distribution of land plots owned by rural households in the survey area.

Close modal
Fig. 2.

Irrigation water for wheat in the survey area (m3/mu, 1 mu = 0.067 ha).

Fig. 2.

Irrigation water for wheat in the survey area (m3/mu, 1 mu = 0.067 ha).

Close modal
Fig. 3.

Distribution of WUE.

Fig. 3.

Distribution of WUE.

Close modal

More than half of the households in the survey area had more than three plots, and 15% of the households had more than five plots. The proportion of micro-irrigation or sprinkler irrigation was only 18.06%, and the overall adoption of community-based WST was not high. To estimate the WUE, it is necessary to determine the input–output index in formula (1). The variables are presented in Table 1. (1) The output is expressed as the average wheat yield per mu (1 mu = 0.067 ha). (2) Water consumption of groundwater irrigation is mainly pumped from motor wells by electric pumps, and there is a relationship between water consumption and electricity charges. In this study, the water consumption of farmers is calculated by the electricity charges of irrigation pumping consumption, expressed as the average water consumption per mu (m3/mu). (3) Labor input measured in man-days/mu a day is calculated after 8 h. (4) Fertilizers are homogeneous, and the unit costs are the same. The fertilizer input is calculated from the cost of using means per mu (Yuan/mu). (5) The cost of using machinery services by farmers is calculated from the average cost of using machinery per mu (Yuan/mu).

Table 1.

Descriptive statistics of the input–output indicators.

Community-based WST
Household-based WST
VariableUnit of measurementMeanSDMeanSD
Yield (Ykg/mu 506.976 91.839 480.197 92.633 
Water (Wm3/mu 207.629 61.449 232.051 69.743 
Labor (Lman-days/mu 1.722 0.971 1.833 1.121 
Fertilizer (FYuan/mu 249.611 64.288 252.169 68.301 
Machine (MYuan/mu 121.472 42.039 117.652 44.620 
Community-based WST
Household-based WST
VariableUnit of measurementMeanSDMeanSD
Yield (Ykg/mu 506.976 91.839 480.197 92.633 
Water (Wm3/mu 207.629 61.449 232.051 69.743 
Labor (Lman-days/mu 1.722 0.971 1.833 1.121 
Fertilizer (FYuan/mu 249.611 64.288 252.169 68.301 
Machine (MYuan/mu 121.472 42.039 117.652 44.620 

To avoid the bias caused by omitting the variables and obtaining unbiased and consistent estimates, we added the control variables related to WUE and the adoption of community-based WST, such as family characteristics and the variables affecting farmers' production and management. It mainly includes the distance between the cultivated land and the driven well. If the farmers are far away from the driven well, there may be relatively more water loss in the process of water conveyance and also an increased irrigation cost, which may reduce the efficiency of groundwater irrigation. Soil type may affect farmers' agricultural production and output. The soil types in the survey area mainly consisted of sand, loam, and clay. In the case of farmers, the age of the household head, education level, and nonagricultural income ratio are also controlled. If farmers perceive a scarcity of water resources, they may affect the efficiency of water use. When farmers realize that groundwater is scarce, they may improve their water efficiency to cope with water scarcity. The change in the groundwater level affects the water efficiency of farmers. When the groundwater level drops, the difficulty and cost associated with procuring water will increase, which may negatively impact the WUE. However, the decreasing groundwater level may also make farmers aware of water shortages, create a sense of crisis in them, and consequently improve WUE to cope with the decline in the water table level. Farmers' knowledge of WST and water-saving policies has a positive impact on their water-saving awareness and behavior; this increases their WUE. We set these variables as the control variable Z, as summarized in Table 2.

Table 2.

Definitions and descriptive statistics of the variables.

VariableIndicatorDescription and definitionMeanSD
Land fragmentation Frag Reciprocal of average plot size of farmer plots/mu 0.444 0.351 
Adoption of community-based WSTs Save 1 = Community-based WST; 0 = Household-based WST 0.181 0.385 
Water scarcity perception Sca 1 = Very sufficient; 2 = sufficient; 3 = just enough; 4 = a little nervous; 5 = severe deficiency 3.11 1.303 
Awareness of water table perception Table 1 = More abundant; 2 = basically the same; 3 = a significant decrease 2.69 0.593 
Awareness of water-saving policy WSP 1 = Completely ignorant; 2 = not very familiar with; 3 = basic understanding; 4 = better understanding; 5 = very well understood 1.97 1.206 
Water-saving technologies WST 1 = Completely ignorant; 2 = not very familiar with; 3 = basic understanding; 4 = better understanding; 5 = very well understood 2.22 1.169 
Distance from the well Distance between land and well (m) 145.7 114.015 
Nonagricultural income ratio Nonagricultural income (Yuan)/agricultural income (Yuan) 0.473 0.384 
Age Age Age of household head 55.54 10.187 
Education Edu 1 = Below primary school; 2 = primary school; 3 = junior high school; 4 = senior high schools; 5 = college or above 2.88 0.921 
Soil type Soil 1 = Sand; 2 = loam; 3 = clay; 4 = other 2.37 0.997 
VariableIndicatorDescription and definitionMeanSD
Land fragmentation Frag Reciprocal of average plot size of farmer plots/mu 0.444 0.351 
Adoption of community-based WSTs Save 1 = Community-based WST; 0 = Household-based WST 0.181 0.385 
Water scarcity perception Sca 1 = Very sufficient; 2 = sufficient; 3 = just enough; 4 = a little nervous; 5 = severe deficiency 3.11 1.303 
Awareness of water table perception Table 1 = More abundant; 2 = basically the same; 3 = a significant decrease 2.69 0.593 
Awareness of water-saving policy WSP 1 = Completely ignorant; 2 = not very familiar with; 3 = basic understanding; 4 = better understanding; 5 = very well understood 1.97 1.206 
Water-saving technologies WST 1 = Completely ignorant; 2 = not very familiar with; 3 = basic understanding; 4 = better understanding; 5 = very well understood 2.22 1.169 
Distance from the well Distance between land and well (m) 145.7 114.015 
Nonagricultural income ratio Nonagricultural income (Yuan)/agricultural income (Yuan) 0.473 0.384 
Age Age Age of household head 55.54 10.187 
Education Edu 1 = Below primary school; 2 = primary school; 3 = junior high school; 4 = senior high schools; 5 = college or above 2.88 0.921 
Soil type Soil 1 = Sand; 2 = loam; 3 = clay; 4 = other 2.37 0.997 

Stochastic frontier model

The analysis was performed using STATA 16, and the estimates are reported in Table 3.

Table 3.

Estimated translog production function.

VariableCoefficientSE
lnW 1.1847** 0.5607 
lnL 0.2052** 0.1037 
lnF 0.0763* 0.0451 
lnM 0.2994* 0.1759 
(lnW)2 −0.1194** 0.0594 
(lnL)2 0.0079 0.0731 
(lnF)2 −0.0798 0.0731 
(lnM)2 −0.0039 0.0143 
lnW*lnL −0.077* 0.0458 
lnW*lnF 0.0955 0.1168 
lnW*lnM −0.0883 0.0749 
lnL*lnF −0.0039 0.0445 
lnL*lnM 0.0508** 0.0224 
lnF*lnM 0.0348 0.0624 
 0.02169  
 0.0171  
Log-likelihood 153.5968  
LR test 57.04***  
VariableCoefficientSE
lnW 1.1847** 0.5607 
lnL 0.2052** 0.1037 
lnF 0.0763* 0.0451 
lnM 0.2994* 0.1759 
(lnW)2 −0.1194** 0.0594 
(lnL)2 0.0079 0.0731 
(lnF)2 −0.0798 0.0731 
(lnM)2 −0.0039 0.0143 
lnW*lnL −0.077* 0.0458 
lnW*lnF 0.0955 0.1168 
lnW*lnM −0.0883 0.0749 
lnL*lnF −0.0039 0.0445 
lnL*lnM 0.0508** 0.0224 
lnF*lnM 0.0348 0.0624 
 0.02169  
 0.0171  
Log-likelihood 153.5968  
LR test 57.04***  

Standard errors in parentheses.

***p < 0.01, **p < 0.05, *p < 0.1.

The LR test results indicate that it is necessary and reasonable to use the stochastic frontier model. The ratio indicates that technical efficiency contributes 55.9% of the total variance to the output. The output elasticities of wheat yield for each input are listed in Table 4. The highest elasticity is observed in fertilizer inputs (0.4642), followed by Machinery (0.1899), and Labor (0.1605). The elasticity of water is 0.0402, indicating that a 1% increase in irrigation water leads to only a 0.0402% increase in output. The sum of elasticities for the four inputs is 0.8548, indicating a slightly diminishing return to scale.

Table 4.

Output elasticities.

InputWaterLaborFertilizerMachineryReturn to scale
Elasticity 0.0402 0.1605 0.4642 0.1899 0.8548 
InputWaterLaborFertilizerMachineryReturn to scale
Elasticity 0.0402 0.1605 0.4642 0.1899 0.8548 

The estimated distributions of WUE are shown in Figure 3. The mean value of WUE is 0.6057, indicating that given the current technology and keeping other inputs constant, the same output can be produced using 39.43% less water. This implies that a large proportion of irrigation water is wasted. However, this also indicates a substantial savings potential.

Econometric model

Table 5 presents the regression results. The estimation results showed that land fragmentation had a significant negative effect on WUE. The adoption of community-based WST has a significant positive effect on WUE. The interaction term between fragmentation and the adoption of community-based WST is significantly negative, indicating that water efficiency improved by the adoption of community-based WST depends on the degree of land fragmentation. Farmers with higher levels of education use water more efficiently. The distance from the well had a significant negative impact on WUE. The farther the distance, the lower the WUE. The farther the distance from the well, the greater the possibility of leakage-induced water loss, thus reducing the WUE of the farmers. When farmers realize that water is scarce, they increase the irrigation efficiency in response to water scarcity. When realizing a fall in the local water table, farmers' WUE may increase or decrease. As mentioned above, the results do not indicate a clear and significant direction of impact. When the groundwater level drops, the difficulty of getting water and the cost of getting water will increase, which may negatively impact the WUE. However, the decrease in the groundwater level may also make farmers aware of water shortages, create a sense of crisis, and then improve the WUE to cope with the decline in water level. The degree of understanding of WST and water-saving policies has a positive impact on WUE. The more knowledgeable farmers are about WST, the more efficient is the use of irrigation water.

Table 5.

Regression results of the determinants of WUE.

WUE
VariableModel 1Model 2Model 3Model 4Model 5Model 6
Frag −0.0941*** −0.0914*** −0.0751*** −0.0846*** −0.0839*** −0.0719*** 
 (0.0196) (0.0196) (0.0197) (0.0214) (0.0214) (0.0221) 
Save  0.0397** 0.1269***  0.0328* 0.108** 
  (0.0197) (0.0352)  (0.0174) (0.0410) 
Frag*Save   −0.2219***   −0.200** 
   (0.0746)   (0.0853) 
Sca    0.0579** 0.0548* 0.0529* 
    (0.0285) (0.0286) (0.0285) 
Table    −0.298 −0.286 −0.301 
    (0.184) (0.184) (0.183) 
WSP    0.0595* 0.0588* 0.0542 
    (0.0338) (0.0338) (0.0337) 
WST    0.113** 0.104** 0.0878* 
    (0.0483) (0.0486) (0.0490) 
D    −4.80 × 10−5−4.67 × 10−5−4.55 × 10−5
    (2.56 × 10−5(2.56 × 10−5(2.55 × 10−5
R    0.000799 0.000816 0.000835 
    (0.000981) (0.000980) (0.000976) 
Age    0.00156 −0.00156 −0.00159 
    (0.000835) (0.000834) (0.000831) 
Edu    0.0562* 0.0577* 0.0548 
    (0.0335) (0.0335) (0.0334) 
A set of dummy variables of soil and County    Control Control Control 
Constant   0.6319 3.542** 3.541** 3.592** 
    (1.633) (1.631) (1.625) 
R2 0.0403 0.0474 0.0626 0.136 0.140 0.148 
WUE
VariableModel 1Model 2Model 3Model 4Model 5Model 6
Frag −0.0941*** −0.0914*** −0.0751*** −0.0846*** −0.0839*** −0.0719*** 
 (0.0196) (0.0196) (0.0197) (0.0214) (0.0214) (0.0221) 
Save  0.0397** 0.1269***  0.0328* 0.108** 
  (0.0197) (0.0352)  (0.0174) (0.0410) 
Frag*Save   −0.2219***   −0.200** 
   (0.0746)   (0.0853) 
Sca    0.0579** 0.0548* 0.0529* 
    (0.0285) (0.0286) (0.0285) 
Table    −0.298 −0.286 −0.301 
    (0.184) (0.184) (0.183) 
WSP    0.0595* 0.0588* 0.0542 
    (0.0338) (0.0338) (0.0337) 
WST    0.113** 0.104** 0.0878* 
    (0.0483) (0.0486) (0.0490) 
D    −4.80 × 10−5−4.67 × 10−5−4.55 × 10−5
    (2.56 × 10−5(2.56 × 10−5(2.55 × 10−5
R    0.000799 0.000816 0.000835 
    (0.000981) (0.000980) (0.000976) 
Age    0.00156 −0.00156 −0.00159 
    (0.000835) (0.000834) (0.000831) 
Edu    0.0562* 0.0577* 0.0548 
    (0.0335) (0.0335) (0.0334) 
A set of dummy variables of soil and County    Control Control Control 
Constant   0.6319 3.542** 3.541** 3.592** 
    (1.633) (1.631) (1.625) 
R2 0.0403 0.0474 0.0626 0.136 0.140 0.148 

Standard errors in parentheses.

***p < 0.01, **p < 0.05, *p < 0.1.

From Figure 4, we can see that land fragmentation has different effects on WUE. There are differences between community-based WST and household-based WSTs. When the land fragmentation was 0.8 (approximately 1 SD above the mean), the WUE of community-based WST was 0.518, and the WUE of household-based WST was 0.570. When the degree of land fragmentation was 0.1 (approximately 1 SD below the mean), the WUE of community-based WST was 0.709, and the household-based WST was 0.620.

Fig. 4.

Effects of land fragmentation on groundwater use efficiency in different WSTs.

Fig. 4.

Effects of land fragmentation on groundwater use efficiency in different WSTs.

Close modal

It can be seen that regardless of the household-based WST and community-based WST, land fragmentation will negatively affect WUE, but community-based WST is more likely to be affected by the negative impact of land fragmentation than household-based WST. One possible explanation is that the cost of the household-based WST is low. Using surface pipes for irrigation, which is one of the most extensive household-based WSTs, was flexible because it is easier to pull pipelines in fragmented farmland. The scattered land can still be flexibly dealt with, but a high degree of land fragmentation will still weaken the WUE of household-based WSTs. Sprinklers and drip irrigation are the main community-based WSTs in the surveyed area. Community-based WST can play a better role when the degree of land fragmentation is low. Therefore, it has certain requirements for the scale of land and is more vulnerable to the impact of land fragmentation. When the degree of land fragmentation is high, household-based WST performs better at WUE than community-based WST because of its low cost and flexibility. Hence, it is relatively suitable for high land fragmentation. When the degree of land fragmentation is low, community-based WST can significantly improve the irrigation WUE.

According to the results presented in Tables 6 and 7, the adoption of community-based WST has a significant positive impact on WUE. Land fragmentation negatively affects the WUE. It also negatively affects the adoption of community-based WSTs.

Table 6.

Estimation results of the mediating effect.

WUECommunity-based WST
Land fragmentation −0.0820*** −0.0893** 
 (0.0176) (0.0406) 
Community-based WST 0.0328*  
 (0.0174)  
Control variables Controlled  
WUECommunity-based WST
Land fragmentation −0.0820*** −0.0893** 
 (0.0176) (0.0406) 
Community-based WST 0.0328*  
 (0.0174)  
Control variables Controlled  

***p < 0.01, **p < 0.05, *p < 0.1.

Table 7.

Estimation of direct and indirect effects.

Effectsp-value
Indirect effect −0.003* 0.084 
Direct effect −0.0820** 0.00 
Total effect −0.0849*** 0.00 
Effectsp-value
Indirect effect −0.003* 0.084 
Direct effect −0.0820** 0.00 
Total effect −0.0849*** 0.00 

***p < 0.01, **p < 0.05, *p < 0.1.

However, the direct effect of land fragmentation on WUE is more than the indirect effect. The direct and indirect effects observed here prove the rationality of the model set, indicating that land fragmentation and adoption of community-based WST were two exogenous variables; the correlation between them is very weak. This is due to the installation of community-based WST mainly being supported by the government and not the farmers, according to their own land and production conditions.

Small farmers are less likely to take the initiative to adopt community-based WST because land fragmentation fails to give full play to the role of community-based WST in saving labor and water. Most smallholder households using community-based WSTs are supported and invested in by the government; hence, land fragmentation has a weak impact on WUE based on its impact on the adoption of community-based WST. Land fragmentation increases farmers' irrigation costs and management time, making it impossible for farmers to concentrate efficiently on irrigation. Therefore, it has a direct negative impact on farmers' WUE.

Land fragmentation has a significant negative impact on WUE, which is consistent with the existing literature. Fragmented land cannot be scaled up; it is difficult to reduce production costs, and it is not conducive to the efficient use of input factors. Land fragmentation in groundwater over-extraction area reduces the irrigation WUE of farmers. The performance of community-based WSTs is more likely to be affected by land fragmentation. When the degree of land fragmentation is high, the performance of community-based WST is worse than that of household-based WST; community-based WST can only play a role in efficient water-saving when the degree of land fragmentation is low. The indirect path of land fragmentation affecting the WUE through the adoption of community-based WSTs is weak. In our research area, we found that an important reason behind this is that the community-based WST of small farmers is supported by the government rather than an initiative undertaken by small farmers themselves. Since land fragmentation is not conducive to the collective actions of farmers (Wang et al. 2020b), it is more important for small farmers to cooperate and reach a consensus on the use and maintenance of community-based WST, which may also be a reason behind the poorer performance of community-based WST under land fragmentation.

To realize agricultural water-saving and promote community-based WST with high irrigation efficiency, the central government of China has allocated various agricultural support funds from top to bottom through the project system. To complete the task of promoting WST issued by the superior government, basic-level governments have installed community-based WST in some places with a relatively high degree of land fragmentation, which is not good for using community-based WST. In addition to installing efficient water-saving facilities, managing and maintaining them is necessary; land fragmentation makes this difficult. However, due to its low cost and flexibility, household-based WST can adapt to the situation of high land fragmentation and does not need to be maintained under public rural irrigation facilities. Top-down administrative tasks do not consider the diversity of the irrigation needs of small farmers, generated from land fragmentation; this leads to a situation where community-based WST is not suitable for the fragmented land, which is not conducive to the realization of water-saving. Therefore, in addition to top-down financial support, more attention should be paid to the needs of small farmers and their actual agricultural production conditions during the actual implementation of water-saving facilities. The government should attach importance to the support of small-scale farmers in applying water-saving agricultural technology and constantly improve the public facilities management level at the village level and the level of mutual assistance and cooperation among farmers.

Land fragmentation is an exogenous factor that is difficult to change quickly. Land fragmentation in the surveyed area is the main reason why the WUE of farmers cannot be improved. Due to land fragmentation, farmers often have to perform irrigation on their fragmented land plots, which leads to a decline in irrigation efficiency due to variable time and roundtrip paths. Appropriate water-saving irrigation facilities should be equipped, and irrigation management should be performed uniformly. Collective unified irrigation may compensate for the loss of WUE caused by land fragmentation.

Governments could promote the use of reclaimed wastewater to help farmers with the uncertainty associated with climate change (Lavee, 2010). Long-term educational tools should be used to raise farmers' awareness of water-saving. Increasing water tariffs are an ineffective tool (Lavee et al., 2013). Strengthening farmers' understanding of water-saving policies can improve their WUE. Therefore, it is necessary to increase government publicity and support for water-saving, enhance farmers' awareness of water-saving, and promote their understanding of the importance of water conservation and water-saving policies and WST, which is conducive to improving their WUE.

The author would like to express their gratitude to all the reviewers for their valuable recommendations. The work was supported by Joint Research and Development Project Under the Sino-Thai Joint Committee on Science and Technology Cooperation (NO. 2017YFE0133000).

Data cannot be made publicly available; readers should contact the author for details.

Ali
I.
,
Huo
X. X.
,
Khan
I.
,
Ali
H.
&
Khan
S. U.
, (
2019
).
Technical efficiency of hybrid maize growers: a stochastic frontier model approach
.
Journal of Integrative Agriculture
18
(
10
),
2408
2421
.
Amanambu
A. C.
,
Obarein
O. A.
,
Mossa
J.
,
Li
L.
,
Ayeni
S. S.
,
Balogun
O.
,
Oyebamiji
A.
&
Ochege
F. U.
, (
2020
).
Groundwater system and climate change: present status and future considerations
.
Journal of Hydrology
.
https://doi.org/10.1016/j.jhydrol.2020.125163
.
Dogaru
D.
,
Balteanu
D.
,
Mauser
W.
,
Probst
E.
,
Laura
L.
&
Boroneant
C.
, (
2019
).
Spatio-temporal crop yield variability and water use efficiency under diverse land use patterns and climatic conditions: a case study of maize in the Lower Danube Plain, south of Romania
.
Geophysical Research Abstracts
21
,
1
.
Gou
Q.
,
Zhu
Y.
,
Horton
R.
,
H.
,
Wang
Z.
,
Su
J.
,
Cui
C.
,
Zhang
H.
,
Wang
X.
,
Zhen
J.
&
Yuan
F.
, (
2020
).
Effect of climate change on the contribution of groundwater to the root zone of winter wheat in the huaibei plain of China
.
Agricultural Water Management
240
(
2020
),
106292
.
Hassan
W.
, (
2017
).
Irrigation management assessment from the land fragmentation perspective in the Nile Delta
.
Irrigation and drainage
67
,
4
.
Lavee
D.
, (
2010
).
The effect of water supply uncertainty on farmers’ choice of crop portfolio
.
Agricultural Water Management
97
(
11
),
1847
1854
.
Lavee
D.
,
Danieli
Y.
,
Beniad
G.
,
Shvartzman
T.
&
Ash
T.
, (
2013
).
Examining the effectiveness of residential water demand-side management policies in Israel
.
Water Policy
15
(
4
),
585
597
.
Liu
W.
,
Chang
M.
&
Wang
X.
, (
2018
).
Irrigation water efficiency based on stochastic production frontier and influencing factors: an empirical study of wheat in the Guanzhong region, Shaanxi
.
Chinese Journal of Eco-Agriculture
26
(
9
),
1407
1414
.
Sedat
K.
&
Zeki
G.
, (
2018
).
Land consolidation practices in Turkey
.
Current Trends in Natural Sciences
7
(
14
),
76
80
.
Tang
J.
,
Folmer
H.
,
Arno
J.
&
Xue
J.
, (
2013
).
The impacts of management reform on irrigation water use efficiency in the Guanzhong plain, China
.
Papers in Regional Science
93
(2).
Wang
X.
&
Li
X.
, (
2020
).
Impacts of land fragmentation and cropping systems on the productivity and efficiency of grain producers in the North China Plain: taking Cangxian County of Hebei Province as an example
.
Journal of Resources and Ecology
11
(
6
),
580
588
.
Wang
Y.
,
Huang
J.
,
Wang
J.
&
Findlay
C.
, (
2018
).
Mitigating rice production risks from drought through improving irrigation infrastructure and management in China
.
Australian Journal of Agricultural & Resource Economics
62
,
161
176
.
Wang
J.
,
Jiang
Y.
,
Wang
H.
,
Huang
Q.
&
Deng
H.
, (
2019a
).
Groundwater irrigation and management in northern China: status, trends, and challenges
.
International Journal of Water Resources Development
36
,
1
27
.
Wang
J.
,
Yang
Y.
,
Huang
J.
&
Adhikari
B.
, (
2019b
).
Adaptive irrigation measures in response to extreme weather events: empirical evidence from the north China plain
.
Regional Environmental Change
19
(
4
),
1009
1022
.
Wang
J.
,
Zhu
Y.
,
Sun
T.
,
Huang
J.
,
Zhang
L.
&
Guan
B.
2020c
Forty years of irrigation development and reform in china
.
Australian Journal of Agricultural and Resource Economics
64
(
1
),
126
149
.
Xu
L.
&
Chen
L.
, (
2020
).
Analysis of water efficiency and influencing factors of agricultural irrigation in over-exploitation area of groundwater
.
Yellow River
42
(
07
),
145
150
.
Xu
L.
&
Liu
J.
2013
Study on factors influencing farmers’ choice behavior for water-saving irrigation techniques based on survey data from Mengyin, Shandong Province
.
China Rural Survey
(
006
),
45
51
.
Yu
L.
,
Ling
M.
,
Chen
F.
,
Ding
Y.
&
Lv
C.
, (
2020
).
Practices of groundwater over-exploitation control in Hebei Province
.
Water Policy
22
(
4
),
591
601
.
Zhang
T.
,
Wang
J.
&
Teng
Y.
, (
2017
).
Adaptive effectiveness of irrigated area expansion in mitigating the impacts of climate change on crop yields in Northern China
.
Sustainability
(
5
).
doi:10.3390/su9050851
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY-NC-ND 4.0), which permits copying and redistribution for non-commercial purposes with no derivatives, provided the original work is properly cited (http://creativecommons.org/licenses/by-nc-nd/4.0/).