Data Envelopment Analysis (DEA)

The technique of linked processing of numerous comparable indicators and effective queuing of decision-making units (DMUs) is known as DEA [22]. In this paper, the input-oriented BCC model is used for analysis. In the presence of variable payoffs to scale, the BCC model is presumed to have a variety of DMUs, which are applied for evaluating both technical and scale efficiency. The expression is:

where k is the number of DMUs; in addition, there are m input variables and r output variables for each DMU. x_nj is the n^th input index value of the j^th DMU. y_nj is the n^th output index value of the j^th DMU. X_nj0 is the n^th input index value of the j₀^th DMU. Y_nj0 is the n^th output index value of the j₀^th DMU. λ_j is the weight of the input and output indicators; θ is the efficiency evaluation index; S_n^– and S_n⁺ are the relaxation variables of the input and output indexes; and ε is the non-Archimedes infinities.

Malmquist Index Method

Malmquist, a Swedish economist, first introduced the Malmquist index in 1953. The Malmquist index, which is frequently used to quantify the change in total factor production level from period t to period t+1, was transformed to an empirical index by Fare et al. [23] Since the single BCC method can measure only the AWRUE in the study area from a static perspective, the combination of the Malmquist index method can dynamically describe the changes in WRUE in different periods, thus compensating for the defects of the BCC model [24]. Its expression is:

Fig. 1.

Geographical location of the MGPA of the MRYR.

where Tfpch stands for the total factor productivity change index, Techch for the technology change index, and Effch for the technical efficiency change index. Effch can be further divided into Pech and Sech, where Pech represents the change in pure technical efficiency and Sech represents the change in scale efficiency. When Tfpch>1, compared with the previous period, this indicates that the total factor productivity level has increased; otherwise, it has decreased.

Exploratory Spatial Data Analysis

ESDA refers to a group of data analysis methods and procedures that are now widely used in spatial statistics research to reveal the visual phenomenon of spatial dependence and spatial heterogeneity of data [25]. There are two primary forms of spatial autocorrelation coefficients used in this study: one measures whether an item of study has spatial correlation from an overall perspective and is called the global spatial autocorrelation index (Global Moran’s I); the other is the Local spatial autocorrelation index (Local Moran’s I), primarily assessing the spatial correlation of the study item from a local perspective and gauging the similarity between sample values from the local area and nearby areas. The specific formula is:

where S2=1n∑i=1n(xi−x¯), x¯=1n∑i=1nxi, X_i and X_j represent the water resource utilization efficiency of region i and region j, respectively. W_ij represents the adjacency space weight matrix; Z_i and Z_j are the normalizations of observed values over region i and region j. Global Moran’s I has a value range of [-1, 1]. A negative value means a negative correlation, and a positive value means a positive correlation. The local Moran’s I is similar.

Tobit Regression Analysis Model

The AWRUE is calculated by the DEA model with varying returns to scale, and the efficiency ranges within [0-1], which has the feature of truncation as the explained variable. Direct use of ordinary least squares regression analysis will result in inconsistent and erroneous parameter estimation, and thus will impact the accuracy and science of the findings [26]. Therefore, the Tobit model, an interception regression model, was selected, and the comprehensive AWRUE was taken as the explained variable AWRUE_it to investigate the influences of various factors on the AWRUE in the MGPA of the MRYR. The model is:

where AWRUE_it is the comprehensive AWRUE in year t of region i and inf_jit and β_j are the estimated values of each influencing factor index and its corresponding parameters, respectively. β₀ is the model constant term, and μ_i is the random disturbance term.

Influencing Factors of AWRUE

AWRUE is not only related to the labor force, capital, land, and other factors but is also affected by multiple factors. For example, these factors are natural resource endowment [29], water resource management level [30], irrigation technology level [31] and general agricultural input. This paper examined the influencing elements of AWRUE from five perspectives, considering the data that were available and the reality in the MGPA of the MRYR: water resource endowment, water resource management level, agricultural water intensity, agricultural planting structure, and agricultural economic level.

(1) In terms of water resource endowment, water resources per capita (CW) are chosen. Water resource endowment is a “double-edged sword”. On the one hand, sufficient water resources provide great convenience for agricultural irrigation, and this is conducive to promoting agricultural production increases and improving the AWRUE. However, sufficient water supplies cause farmers to have weak water awareness, making it simple for them to misuse and even waste water resources, lowering utilization efficiency. Thus, the expected role of water resource endowment is uncertain.

(2) In terms of the water resources management level, the effective irrigation rate (EI) was selected as the explanatory variable. EI refers to the percentage of effective irrigated area to the total cultivated land area. The waste of irrigation water resources can be decreased by increasing EI. As a result, it is anticipated that agricultural water resource usage will be more effective with more water resources being effectively managed.

(3) In terms of agricultural water intensity, the proportion of total water allotted to agricultural water consumption (AWC) represents agricultural water intensity. In general, traditional agricultural production methods consume greater amounts of water. Therefore, agricultural water intensity is expected to be negatively correlated with AWRUE.

(4) In terms of planting structure, the planting area of the main cash crops (MCA) was selected. Compared with food crops, cash crops consume more water in the whole growth process, but the output value per unit of water resources is higher. Therefore, the expected effect of planting structure is uncertain.

(5) In terms of agricultural economic level, disposable income per rural resident (DI) and primary industry’s added value (PI) are chosen. Farmers and governments will have more financial resources to invest in the management and conservation of agricultural water resources as the level of agricultural economic growth rises, and this will help increase the AWRUE. As a result, a positive impact on the agricultural economic level is anticipated.

Static Analysis of AWRUE

DEAP2.1 was used to calculate the comprehensive efficiency (crste), technical efficiency (vrste), and scale efficiency (scale) of AWRUE in the MGPA of the MRYR during 2006-2020. The efficiency results are shown in Table 2.

(1) crste analysis. The average crste of AWRUE in the MGPA of the MRYR was 0.732, demonstrating the need for increased AWRUE throughout the research area. The crste was higher than average in 45.16% of these regions, among which the comprehensive efficiency of Wuhan, Yichang, and Ezhou was 1.000. The crste in Jiujiang is 0.518, indicating that it is in urgent need of improvement. It is evident that there are vital distinctions in the AWRUE between 2006 and 2020, with a maximum difference of 0.482, which is possibly due to the variations in resource endowment and the unequal distribution of water resources over time and space in the MGPA of the MRYR.

(2) vrste analysis. The average vrste in the MGPS of the MRYR was 0.835, indicating high efficiency but not effective utilization. Among these values, the vrste in Xinyu, Yingtan, Wuhan, and the other 8 regions is 1.000, hence it is highest value. However, it is 0.548 in Jiujiang, indicating that the vrste is obviously very low. Therefore, there is an urgent need for water technology innovation to improve the machinery for and technical level of agricultural irrigation.

(3) scale analysis. The average scale is 0.884, and thus it is relatively high but not fully effective and needs to be further raised. Except for Wuhan, Yichang, and Ezhou, the rest of the regions did not reach the effective state of DEA, 38.71% of the regions were below the average, and even Yingtan was 0.691. Therefore, these regions should reasonably deploy production factors such as land, capital, and water resources and adjust the input scale to have a higher scale.

Dynamic Analysis of AWRUE

To further investigate the dynamic changes in AWRUE in the MGPA of the MRYR, the Malmquist index method was used, and its decomposition results are shown in Table 3. Then, the state of AWRUE in each region over time was studied.

(1) In terms of Effch, the average value is 0.976, with fluctuations in efficiency in different years and a slight overall decrease, indicating that Effch has not been improved. Disaggregate the Effch into Pech and Sech. The mean value of Pech is 0.985, with slight small fluctuations, and hence it is similar to that of Effch. Over the period of 16 years, there were 2 years in which the trend of Pech differed from the trend of Effch and 5 years in which the trend of Sech differed from the trend of Effch, indicating that Pech had a greater impact on Effch.

(2) Techch and Tfpch are both greater than 1 except in 2017, and the average annual growth rates of both are 10.3% and 8.3%, respectively, indicating that the overall AWRUE in the MGPA of the MRYR shows an increasing trend. The directions of Tfpch and Techch remain largely consistent. This is due to Techch being the main factor affecting Tfpch.

Fig. 2.

Total Factor Productivity Index of AWRUE (2006-2020).

The Tfpch of AWRUE in all regions, except Changde, was greater than 1 from 2006 to 2020, with an average value of 1.077 and an average annual growth of 7.7%, showing that the AWRUE in all regions has been improving gradually. The region with the highest Tfpch is Yichun, with an annual growth rate of 12.0%. The lowest region is Changde, with an index value of 0.953, reflecting a decrease in the AWRUE. Moreover, the AWRUE in the research area still has large spatial variances, as evidenced by the fact that the gap in growth rates between the provinces with the most and least growth rates was 16.7%. The Effch, Techch, Pech, and Sech in Wuhan, Yichang and Ezhou are all greater than 1, indicating that the factor inputs have brought about agricultural progress and good development of agricultural production. The trend of Tfpch is basically the same as the change in Techch, indicating that Tfpch is the key to improving AWRUE.

Overall Analysis of AWRUE by Region

In this respect, ArcGIS 10.5 was used to reveal the spatial distribution characteristics of AWRUE, and the results are as follows.

(1) From an overall perspective, the AWRUE in the study area is characterized by “high in the northwest and low in the southeast”, with the AWRUE decreasing from the northwest to the southeast. From 2006 to 2020, the AWRUE in the region experienced a trend of “improvement, decline, and rebound”. The overall efficiency improved more obviously from 2006 to 2010, and improved somewhat after 2015, but Yueyang and Changde still performed poorly.

(2) From a regional perspective, Hubei has the highest AWRUE at 0.826, followed by Hunan at 0.694, while Jiangxi Province has the lowest at 0.641 (according to Table 2). All three provinces showed a decreasing trend in AWRUE overtime, while Hunan and Jiangxi have declined significantly, so these two provinces still need to explore effective strategies to further strengthen AWRUE.

(3) From the cities’ perspective, agricultural water in Wuhan, Yichang, and Ezhou has always been used efficiently. The AWRUE in Jiujiang has been at an extremely low level, and in Ji’an and Yichun it is not high, while in Huangshi, Jingmen, and Changde it has been on a decreasing trend.

Local Analysis of AWRUE in the MGPA of the MRYR

The global correlation and spatial differences of AWRUE were studied based on the Global Moran’s I and Local Moran’s I (LISA aggregation map) (Table 5). The global Moran’s I varied from 0.158 to 0.284, and it satisfied the significance exams, showing that there was an aggregation phenomenon of AWRUE.

Since the Global Moran’s I cannot indicate the spatial leap between different regions, this paper marked the LISA aggregation maps of AWRUE in the MGPA of the MRYR in 2006, 2010, 2015 and 2020 (Fig. 4).

Fig. 3.

Spatial distribution of AWRUE in the MGPA of MRYR.

(1) The high-high aggregation area is mainly distributed in Xiangyang, Ezhou, Huangshi, Wuhan, and Huanggang in the northwestern part of the study area, belonging to Hubei Province, coinciding with the results of the AWRUE measurement and this area is considered a high-value aggregation area. These areas have high AWRUE, sufficient agricultural water resources, and mature agricultural water conservation technology and have a good diffusion effect.

(2) The low-low aggregation area is mainly distributed in Nanchang and Shangrao in the southeast from 2006 to 2010 and Yueyang and Yiyang in the southwest from 2015 to 2020, belonging to Jiangxi Province and Hunan Province. Jiangxi and Hunan have a lower level of economic development, unsophisticated water conservation techniques, and agricultural water conservancy facilities. Jiangxi is a traditional grain-producing province, and production of food consumes a large amount of water, but the economic output is low, so the AWRUE is low.

(3) The two low-high aggregation areas appearing are Jingzhou in 2006 and Tianmen in 2010. These two areas are adjacent to Wuhan and Ezhou high-high aggregation areas, improving the AWRUE through their positive spillover effects. The high-low aggregation areas are Yueyang and Changsha. Due to the adjacent low-low aggregation area, affected by its negative spatial and temporal spillover effect, Yueyang falls as a low value area in 2020.

Hidden Causes of AWRUE

Based on Eviews10.0 software, Tobit regression analysis was performed (Table 6). All six factors satisfied the significance tests, among which the CW, EI, AWC, MCA, and DI had significant negative effects, and the PI showed a positive effect.

Fig. 4.

LISA aggregation map of the AWRUE in MGPA of the MRYR.

(1) CW’s estimated coefficient value is -0.0000378, indicating a relative relationship with AWRUE. This tendency might be caused by the rise in CW, that has facilitated relatively easy access to water, and decreased farmers’ understanding of water conservation, resulting in the misuse of agricultural water.

(2) The estimated coefficient of EI is -0.135, demonstrating a negative correlation between the effective irrigation rate and AWRUE, contrary to the expected effect, and this is perhaps related to the rough irrigation practices in the region. Agricultural irrigation in major grain producing areas is one of the most water-consuming activities in the region, accounting for up to 70% of the water consumption, there inevitably wasting water resources [32]. Therefore, an important way to improve AWRUE at present is to modify the water-saving irrigation technology and improve the water-saving irrigation level.

(3) The estimated coefficient value of AWC is -0.475, and hence it significantly prevents the improvement of AWRUE. The greater the agricultural water intensity is, the higher the agricultural water consumption, reflecting unsophisticated agricultural production methods, poor agricultural water management, and inefficient water utilization.

(4) With an impact coefficient value of -0.000169, the MCA significantly hinders the improvement of AWRUE. As cash crops require an abundance of water to flourish, their AWRUE is low.

(5) The estimated coefficient of DI is -0.0000167, which, albeit being very low in value, indicates that this variable plays a hindering role with its very low level of AWRUE. The MGPA of MRYR are traditional Chinese grain-producing areas and the increase in farmers’ disposable income will prompt farmers to prioritize increases in water sources for irrigation, and these farmers will not invest funds in water conservation, thus failing to raise AWRUE.

(6) The coefficient of the PI is 0.000282, and hence it resembles the expected effect. Increased PI indicates that higher output is obtained with certain resource inputs, and the agricultural production field has more sources of funding to invest in agricultural water conservation technology and improve the AWRUE.

Research Comparison

In terms of the choice of research regions, studies at the national level [17, 33], watershed or economic belt [4, 27], and provincial scales [34] have been fruitful. This paper conducts a study on the MGPA of the MRYR, complementing the lack of previous regional studies at the mesoscopic scale. Geographically speaking, the paper has major locational benefits focusing on the hinterland of the Yangtze River Economic Belt that occupies an indispensable part in the upstream and downstream linkages as well as the east-west transmission of the Yangtze River.

In the choice of time span, this paper studied the time change in AWRUE in 16 years with years as the basic unit. Ekrem Mutlu and Asli Kurnaz [35] selected the months as the data collection unit to evaluated the seasonal changes in the heavy metals and physicochemical parameters of Sakiz Pond, and drew valuable conclusions, thus prociding important implications for this article. Agricultural production is also a seasonal activity, which may aggravate the seasonal imbalance of agricultural water resources utilization. The use of monthly data to study the seasonal change in AWRUE is more accurate and helps to make more targeted suggestions.

According to the results, the AWRUE of the three provinces in the MGPA of the MRYR ranked Hubei, Hunan, and Jiangxi, according to the economic degree of the three provinces. This conclusion was further supported by Tong et al., [4] who found that the lower Yangtze River region had a much greater AWRUE than the upper and middle reaches. Agricultural water use and management technology is continually innovating because of the advanced economy, and farmers have a strong concept of water conservation awareness, causing AWRUE to continuously advance. The frequent floods and droughts in Jiangxi Province have caused a serious mismatch between water supply and demand, resulting in low WRUE.

Regarding the influencing factors, water endowment is one of the primary variables determining WRUE, and many scholars characterize it using CW. Luo et al. [36] argued that CW has a favorable effect on AWRUE, and higher crop water demand can be met with abundant water. Sun et al. [17] argue that the growth in CW is not helpful for WRUE and that when water resources are abundant and easily available, wasting resources becomes easy. This study concluded that CW, in this case, are considered an obstacle to increasing the AWRUE, and farmers’ knowledge of water conservation declines as a result of ample water supplies, leading to waste of agricultural water and lower AWRUE.

Policy Implications

(1) Promote water-saving irrigation techniques and instruments and improve the degree of water-efficient agricultural output. Effch and Techch are the main reasons for the differences in AWRUE. Jiujiang, Fuzhou, and other places with low agricultural water utilization technology efficiency should comprehensively promote the renovation of agricultural water conservation facilities as well as the innovation of agricultural water conservation technology and vigorously promote efficient water conservation technologies such as undermembrane drip irrigation and rainwater harvesting and irrigation to establish a strong base of facilities and technical assistance for enhancing AWRUE.

(2) Strengthen agricultural water conservation education and publicity and increase knowledge of water-saving production among farmers. Every region in the area should actively carry out studies on the use of water conservation equipment and training on water conservation techniques to promote and guide farmers’ water use behavior. Knowledge of agricultural water conservation should be effectively popularized, and farmers should be guided to use irrigation water rationally and reduce water abuse and waste to improve the AWRUE.

(3) Eliminate the compartmentalized administrative system and achieve regional cooperation in water conservation technology. In the context of the “Rise of Central China” strategy and the “Yangtze River Economic Belt” development strategy, administrative barriers should be broken down among the three provinces in the MGPA of the MRYR, and cooperation in terms of the efficient utilization and conservation of agricultural water resources should be developed to realize the basin-wide sharing of advanced water conservation technologies.

(4) Develop appropriate and effective water conservation strategies and guide the effective use of regional water sources. The improvement of AWRUE relies on the guidance of policies. Local governments should widely learn from the beneficial experiences of Wuhan, Yichang, and Ezhou and formulate appropriate policies in the context of local realities to maintain the current level in areas of efficient utilization and shift to efficient utilization in areas of inefficient utilization, thus realizing effective AWRUE in the whole region. Strengthen regional water resources legislation, improve the system of laws and regulations, strengthen departmental supervision, and promote the implementation of regulations and effective implementation.