Water-Saving Irrigation Promotion and Food Security: A Study for China

Abstract

In response to water scarcity and food security, most governments in the world have adopted water-saving irrigation promotion policies by increasing the water-saving irrigation area. Many researches focus on water scarcity, but there is a lack of research on the food security effects of water-saving irrigation policies. A two-way fixed effect model was used to identify the effect of water-saving irrigation area on the production of food crops with panel data of 31 Chinese provinces from 2000 to 2019. The study shows: First, water-saving irrigation area not only can save agricultural water use, but also has a significant positive effect on production of food crops; Second, water-saving irrigation area affects production of food crops through agricultural input factors, such as sown area, chemical fertilizer, and mechanized power; Third, there is heterogeneity in the effect of water-saving irrigation area on production of food crops, in which water-saving irrigation area has a greater impact on production of food crops in areas where there is less rainfall, or lower water-saving irrigation rate. Therefore, the water-saving irrigation promotion has a significant role in promoting China’s production of food crops and has made a certain contribution to ensuring food security.

1. Introduction

Water is the most important natural resource to economic development. Even though approximately 70% of the earth is covered by water [1], only nearly 2.5% of the water is available to humans as fresh water resources [2,3]. Because a large portion of earth’s fresh water is in the form of glaciers, only about 0.26% is accessible for use [4]. Food production water accounts for a large proportion of total water consumption, and about 40% of the world’s food harvest comes from irrigated land [5]. Overuse of aquifers puts severe stress on water-based food production systems. Countries currently over-pumping aquifers include China, India, and the United States. China still faces issues of water scarcity [6,7]. However, 80% of China’s food crops are produced on irrigated farmland [8] and China’s water supply has been declining rapidly [6]. Irrigation has played a key role in producing enough food for China. However, water shortages are becoming a bottleneck. At the same time, rapid urbanization is creating serious conflicts between industrial and agricultural water use. Therefore, it may be necessary to reduce the allocation of freshwater resources to agriculture in order to ensure demand for fresh water in other sectors of economic growth. So, it is a major goal of Chinese agriculture to develop water-saving strategies and technologies, and improve water efficiency of crop production [5].

As a large country of agriculture production, China faces the situation that agriculture is still the largest water-using sector. So, water conservation in the agricultural sector has become an essential way to ease China’s water scarcity and ensure food security. In 1998, the government of China proposed to formulate policies to promote water conservation, and vigorously develop water-saving agriculture, and promote water-saving irrigation as a revolutionary measure, which has become a major policy guarantee for water conservation [9]. Since then, various agricultural water conservation policies have been introduced. Most importantly, in 2012, the State Council specifically proposed the “National Agricultural Water Conservation Program (2012–2020)” as an agricultural water conservation promotion policy, and emphasized the specific goal of promoting water-saving irrigation [10]. The water-saving irrigation includes sprinkler irrigation, micro-irrigation, low-pressure pipe irrigation, and channel impermeability. The micro-irrigation is the best for water-saving, because it can evenly and accurately transport water and nutrients needed for crop growth directly to the soil near the roots of the crop with a small flow. So, the water-saving irrigation can not only saving water, but also promote the growth of crops. An important aim of the existing agricultural water conservation policy is to promote sustainable use of water resources and guarantee the country’s food security.

For water-saving irrigation promotion policies, most existing researches have focused on the problems of the adoption of water-saving irrigation technologies [11,12,13,14,15] and agricultural water use efficiency [16,17,18,19,20]. However, relatively few studies have been conducted on the relationship between water-saving irrigation promotion and crop production. Emerick et al. [21] indicated that improved technologies that reduce risk by protecting production have the potential to increase agricultural productivity. Salazar et al. [22] found that irrigation technology adoption is a way to prevent from the effect of production risk among small-scale farmers. Deng et al. [23] indicated that greater attention should be paid to more efficient use of water for improving wheat production, facing global climate change which may result in drought. Wallace [24] also clarified that the increases in production need to be achieved by agricultural water-saving, facing the situation where agriculture is the largest single user of fresh water and 7% of the world’s population live in water scarcity areas. Liu et al. [25] found that ground cover rice production system (a water-saving technology) can significantly increase crop yield by on average 18% by doing experiment. Belder et al. [26] also found that water-saving irrigation can increase the yield of rice by doing experiment. In summary, most studies have focused on the adoption of agricultural water conservation technologies and agricultural water use efficiency, and only a few studies have theoretically agreed that agricultural water conservation is essential for food production, but little literature has been completed to empirically explain how agricultural water conservation can ensure food security, and further research is needed in this area.

In order to find whether and how water-saving irrigation promotion can ensure food security, this paper empirically analyzed the effect of water-saving irrigation area on the production of food crops, using panel data of 31 provinces and cities across China (excluding Hong Kong, Macao, and Taiwan) from 2000 to 2019 with a two-way fixed effect model. Further analysis also has been completed, by investigating the mechanism and heterogeneity of the effect of water-saving irrigation area on production of food crops. Therefore, this paper has two main objectives. The first one is to estimate the average effect of water-saving irrigation area on production of food crops in China. The second one is to identify the mechanism of the effect, mainly based on the role of water-saving irrigation promotion in agricultural production risk. The heterogeneity of the effect is also discussed.

The present work is organized as follows: Section 2 conducts theoretical analysis and proposes research hypotheses. In Section 3, the materials and methods are given. The results are introduced in Section 4. The discussion can be seen in Section 5. Finally, a summary is provided and some conclusions are highlighted in Section 6.

2. Theoretical Analysis and Research Hypothesis

Water resources are the most basic natural resources and are crucial for crop production. The water deficit of crops will have a close relationship with the yield of crops. The conventional irrigation methods cannot irrigate precisely and will affect the crop yield, because it cannot achieve the expected maximum yield. When water-saving irrigation is used, it will enable the crops to be irrigated precisely, which will have a compensatory effect on crops with limited water deficit. Therefore, the expansion of water-saving irrigation area can lead to an increase in crop yield even when no other factors are changed [25]. For food crops, an increase in water-saving irrigation area also leads to an increase in production. So, the water-saving irrigation promotion can directly increase the production of food crops by the adoption of water-saving irrigation technology.

From the perspective of agricultural production risks, the water-saving irrigation promotion can reduce the risk of drought in agricultural production, thereby increasing agricultural input, and finally achieving an increase in agricultural production. The improved technologies that reduce risk by protecting production have the potential to increase agricultural productivity [21]. The water-saving irrigation is also a technology improvement which can prevent from the agricultural production risk in drought [22]. So, the expansion of water-saving irrigation area can further secure the water demand of crops, and reduce the risk of drought in crop production, thus realizing the fundamental role of water resources in crop production.

As typical farmers, they often produce crops relying on rainfall, or using conventional irrigation methods. However, rainfall has uncertainty, while conventional irrigation has the uncertainty whether it can achieve the expected production. Thus, the farmers mostly do not invest or invest less in other factors of agricultural production because of the existence of the drought risk in production. However, the water-saving irrigation promotion can reduce the drought risk in production, because it can irrigate precisely with less water in production. With the water-saving irrigation promotion, the farmers, as production decision makers, will increase the input of other agricultural production factors to increase the amount of agricultural production after the most basic water demand for crop production is satisfied. So, water-saving irrigated promotion can indirectly increase the production of food crops through the intermediary effect of agricultural production factors by reducing the drought risk in production.

Combining the attributes of water-saving irrigation and the agricultural production decisions of typical farmers, we can infer the macroscopic results of a region. When the area of water-saving irrigation in a region expands, firstly the production of food crops can increase directly, and secondly the input of other agricultural production factors in the region will also increase, and then indirectly increase production of food crops. Based on the above analysis, the following two hypotheses can be put forward:

Hypothesis 1 (H1).

Water-saving irrigation area will positively affect the production of food crops.

Hypothesis 2 (H2).

Water-saving irrigated area will affect the production of food crops through the intermediary effect of agricultural production factors.

3. Materials and Methods

3.1. Model Specification

The water-saving irrigation promotion is a national policy promoted by the government in China, so the policy has exogeneity to some extent. However, in order to avoid the typical characteristics of various regions and the influence of external shocks in different periods, the two-way fixed effect model is used to study the effect of water-saving irrigation promotion on production of food crops. In order to test the mechanism, intermediary variable are used in econometrics model [16,27,28]. So, the model can be seen as follows:

$$FC{P}_{it}={\sigma }^1+{\alpha }_i^1+{\beta }_t^1+{\eta }^{1}WI{A}_{it}+{\gamma }^1{X}_{it}+{ϵ}_{it}^1$$

(1)

$$AP{F}_{it}={\sigma }^2+{\alpha }_i^2+{\beta }_t^2+{\eta }^{2}WI{A}_{it}+{\gamma }^2{X}_{it}+{ϵ}_{it}^2$$

(2)

$$FC{P}_{it}={\sigma }^3+{\alpha }_i^3+{\beta }_t^3+{\eta }^{3}WI{A}_{it}+{\gamma }^3{X}_{it}+\mu AP{F}_{it}+{ϵ}_{it}^3$$

(3)

where, i and t denote province and year, respectively. The serial numbers on parameters represent the different equations. $FC{P}_{it}$ is the production of food crops as a dependent variable. The $WI{A}_{it}$ is the water-saving irrigation area as an independent variable, which reflects the effectiveness of water-saving irrigation promotion. There are a series of control variables, denoted by $X_{it}$, which may be associated with water-saving irrigation area. The $AP{F}_{it}$ denotes the main agricultural production factors, which are the intermediary variables. In order to avoid the estimation bias, region, and time fixed effects are added to the model, which are denoted by ${\alpha }_i$ (${\alpha }_i^1$, ${\alpha }_i^2$, and ${\alpha }_i^3$) and ${\beta }_t$ (${\beta }_t^1$, ${\beta }_t^2$, and ${\beta }_t^3$). The ${\sigma }^1$, ${\sigma }^2$, and ${\sigma }^3$ are constant terms. The ${\eta }^1$, ${\eta }^2$ and ${\eta }^3$ are the coefficients of $WI{A}_{it}$. The ${\gamma }^1$, ${\gamma }^2$, and ${\gamma }^3$ are the coefficients of $X_{it}$. The $\mu$ is the coefficient of $AP{F}_{it}$. The ${ϵ}_{it}^2$, ${ϵ}_{it}^2$, and ${ϵ}_{it}^3$ are the residual terms.

The mediation model can be used to identify the effect of water-saving irrigation promotion on food security. Firstly, the Equation (1) is used to estimate the average effect of water-saving irrigation area on food crop production in China. If the ${\eta }^1$ is significant, then the average effect of water-saving irrigation area on food crop production in China exists. Secondly, the Equations (2) and (3) are used to test the mediation effect of the intermediary variables in the effect of water-saving irrigation area on food crop production. According to the mediating effect test procedure [29,30,31], if the ${\eta }^1$, ${\eta }^2$, ${\eta }^3$, and $\mu$ are all significant, then the mediating effect is also significant.

3.2. Variable Selection

The dependent variable is food crop production ($FCP$), and the agricultural water use ($AWU$) is also used for analyze the water-saving effect of water-saving irrigation promotion. The independent variable is water-saving irrigation area ($WIA$). The intermediary variables are agricultural production factors, including crop sown area ($CSA$), chemical fertilizer ($CF$), and mechanized power ($MP$). Control variables are rainfall ($RF$), drought area ($DA$), and effective irrigation area ($EIA$). The primary explanation of the selected variables can be seen in

Table 1. Explanation of the selected variables.

Variables	Meaning	Explanation	Unit
FCP	Food crop production	The sum of different food crops according to the conversion of China Statistics	Ten thousand tons
AWU	Agricultural water use	The amount of water use for agriculture	One hundred million cubic meters
WIA	Water-saving irrigation area	The irrigation area with water-saving irrigation technique according to China Statistics	One thousand hectares
CSA	Crop sown area	The area of all land sown or transplanted for harvesting food crops	One thousand hectares
CF	Chemical fertilizer	The amount of fertilizer used in agricultural production according to the conversion of China Statistics	Ten thousand tons
MP	Mechanized power	The power used for agricultural production with machine	Ten thousand kilowatt-hours
RF	Rainfall	The depth of the water layer that accumulates on the water surface from rainfall	Ten millimeters
DA	Drought area	The crops sown area that suffers from severe drought	One thousand hectares
EIA	Effective irrigation area	The area of cultivated land with certain water source which can be with conventional irrigation or water-saving irrigation	One thousand hectares

, and the detail can be seen as follows.

3.2.1. Dependent Variable and Independent Variable

The dependent variable is food crop production ($FCP$). The food crops, in China, refer to staple food crops, which includes cereal crops (wheat, rice, corn), potato crops (sweet potatoes, potatoes, etc.), and legumes (soybeans, broad beans, peas, mung beans, etc.), and the amount of food crops is the sum of different food crops according to the conversion of China Statistics. In order to analyze the water-saving effect of water-saving irrigation, the independent variable also includes agricultural water use ($AWU$), which is the amount of water use for agriculture. The independent variable is water-saving irrigation area ($WIA$), which is promoted by agricultural water-saving policy. The purpose of agricultural water-saving policy is to obtain as much crop output as possible with as little water input as possible. The water-saving irrigation area is the irrigation area with water-saving irrigation technique according to China Statistics, which includes sprinkler irrigation area, micro-irrigation area, low-pressure pipe irrigation area, channel impermeability area, and other water-saving irrigation area.

3.2.2. Intermediary Variable

The intermediary variables are agricultural production factors, and we select the main agricultural production factors, which include crop sown area, chemical fertilizer, and mechanized power. The crop sown area ($CSA$) refers to the area of all land (cropland or non-cropland) sown or transplanted for harvesting crops during the calendar year, which reflects, to a certain extent, the land factor input for crop production. The chemical fertilizer ($CF$) is the actual amount of fertilizer used in agricultural production according to the conversion of China Statistics, including nitrogen, phosphorus, potassium and compound fertilizer. The mechanized power ($MP$) refers to the power used for agricultural production with machine.

3.2.3. Control Variable

The water-saving irrigation promotion policy will be correlated with regional drought conditions, so the water-saving irrigation area will be correlated with rainfall and drought area. In addition, the water-saving irrigation promotion policy is also dependent on water infrastructure, which means that it is correlated with the effective irrigation area. Therefore, the control variables considered in the estimation model are rainfall, drought area, and effective irrigation area. The rainfall ($RF$) is the depth of the water layer that accumulates on the water surface from rainfall falling from the sky to the ground, which reflects, to a certain extent, the degree of regional drought. The rainfall can have a positive effect on food crop production, however too much rainfall can delay the planting and can negatively affect production. Considering the complex impact of rainfall on agricultural production, quadratic and cubic terms of rainfall will also be considered as control variables. The drought area ($DA$) refers to the sown area of crops that suffer from severe drought and result in yield reduction. The effective irrigation area ($EIA$) refers to the area of cultivated land with certain water source, relatively flat land, irrigation engineering or equipment, and can be irrigated normally in a normal year, which is an important indicator of China’s farmland water conservancy construction. The effective irrigation area can be with conventional irrigation or water-saving irrigation.

3.3. Data Sources

The dependent variable, food crop production ($FCP$), is calculated by summing up all the main food crops according to the conversion of China Statistics, and the data are mainly obtained from the National Bureau of Statistics of China [32]. The independent variable, water-saving irrigation area ($WIA$) and agricultural water use ($AWU$), is mainly obtained from the China Water Conservancy Yearbook [33]. The data of crop sown area ($CSA$) are obtained from the National Bureau of Statistics of China [32]. The chemical fertilizer ($CF$) is calculated by the method of “discounted pure amount”, which is obtained from the China Rural Statistical Yearbook [34]. The data of mechanized power ($MP$) are also obtained from the China Rural Statistical Yearbook [34]. The data of rainfall ($RF$) are obtained from Ministry of Water Resources of the People’s Republic of China [35]. The data of drought area ($DA$) are obtained from the China Rural Statistical Yearbook [34]. The data of effective irrigation area ($EIA$) are mainly obtained from the China Water Conservancy Yearbook [33].

The descriptive statistics of the above variables are specified in

Table 2. Descriptive statistics of the selected variables.

Variables	Mean	Std. Dev.	Min	Max	N
CFP	1794.80	1548.60	28.76	7615.80	620
AWU	120.10	100.65	3.70	561.70	620
WIA	848.14	853.55	2.54	4181.48	620
CSA	3087.94	2455.88	46.52	11,219.55	620
CF	169.63	138.68	2.50	716.10	620
MP	2730.99	2675.10	94.00	13,353.00	620
RF	89.14	52.77	3.66	267.89	620
DA	584.00	813.40	0.00	6500.00	620
EIA	1964.42	1519.42	103.92	6208.23	620

. The final sample size for the study is 31 provinces for 20 years, so there are 620 observations in the data. The Hong Kong, Macao, and Taiwan are excluded due to difficulty of data collection. The maximum of dependent variable, food crop production ($FCP$), is 761,580 thousand tons, the minimum is 2876 thousand tons, and the mean and standard deviation are 179,480 thousand tons and 154,860 thousand tons. The maximum of independent variable, water-saving irrigation area ($WIA$), is 4181.48 thousand hectares, the minimum is 2.54 thousand hectares, and the mean and standard deviation are 848.14 thousand hectares and 853.55 thousand hectares. Therefore, the standard deviations of the explanatory variable and the explained variable are relatively large and close to the means.

4. Results

4.1. The Water-Saving Effect of Water-Saving Irrigation

In order to avoid the bias caused by the omitted variables, and correctly estimate the water-saving effect of water-saving irrigation, the control variables are mainly considered; those variables that are related to the water-saving irrigated area, mainly including the drought area, effective irrigated area and rainfall. Considering the complex impact of rainfall on food crop production, quadratic and cubic terms of rainfall will also be considered in the estimation model. The effective irrigated area is the main confounder in the effect of water-saving irrigation on agricultural water use. Based on this, the model only with effective irrigated area is first considered, then different control variables are put in order, and finally all the control variables are put in together, which specifically forms three models, and the estimation results are shown in

Table 3. The results of water-saving effect of water-saving irrigation.

Model	(1)	(2)	(3)
Variable	AWU	AWU	AWU
$WIA$	−0.0046 **	−0.0044 **	−0.0046 **
	(−2.4362)	(−2.3314)	(−2.4635)
$EIA$	0.0356 ***	0.0359 ***	0.0357 ***
	(24.6396)	(24.5618)	(24.5919)
$DA$		0.0009	0.0005
		(1.1942)	(0.6671)
$RF$			0.1689
			(1.4179)
$R{F}^2$			−0.0019 **
			(−1.9846)
$R{F}^3$			0.0001 *
			(1.8842)
$Region$	Y	Y	Y
$Time$	Y	Y	Y
$Constant$	18.7598 ***	16.5035 ***	16.1795 **
	(3.2454)	(2.7147)	(2.0055)
N	620	620	620
$R^2$	0.9901	0.9901	0.9903
$Adj\text{-}{R}^2$	0.9892	0.9892	0.9894

Note: The t values in parentheses; * p < 0.1; ** p < 0.05; *** p < 0.01.

. It should be noted that the models in Table 3 have added region and time fixed effects.

Table 3 reflects the water-saving effect of water-saving irrigation. Only fixed effects and effective irrigated area are included in Model (1), and the regression results indicate that water-saving irrigated area negatively affects agricultural water use at the 5% significance level, as shown by the fact that for every thousand hectares increase in water-saving irrigated area, agricultural water use will decrease by 0.0046 hundred million cubic meters. In Model (2), the drought area is also included as a control variable. With the inclusion of drought area as a control variable, the effect of water-saving irrigated area on agricultural water use remains negative at the 5% significance level. In Model (3), the rainfall, squared rainfall and cubic rainfall, are newly added as control variables, water-saving irrigated area still has a negative effect on agricultural water use at the 5% significance level. The estimated coefficients of water-saving irrigated area on agricultural water use from Model (1) to Model (3) are roughly the same. Therefore, based on the estimation results of Model (3) with all control variables, it can be seen that each thousand hectares increase in water-saving irrigated area can decrease agricultural water use by 0.0046 hundred million cubic meters. In summary, the water-saving irrigation area has a significant negative impact on agricultural water use, so the promoting water-saving irrigation area indeed can save agricultural water use.

Table 3 reflects not only the effect of water-saving irrigated area on agricultural water use, but also the effect of other important control variables on agricultural water use. Models (1)–(3) both show that effective irrigated area has a positive effect on agricultural water use at the 1% significance level. Models (2) and (3) both show that drought area has a positive effect on agricultural water use, but the effect is not significant. Model (3) indicates that rainfall has a non-linear effect on agricultural water use at the 10% significance level. So, effective irrigated area and rainfall are also important factors for agricultural water use.

4.2. The Effect of Water-Saving Irrigation Area on Food Crop Production

In order to avoid the bias caused by the omitted variables, and correctly estimate the effect of water-saving irrigation area on the amount of food crop production, the control variables also include the drought area, effective irrigated area and rainfall, and the quadratic and cubic terms of rainfall will also be considered in the estimation model. Based on this, in the baseline model, the model without control variables is first considered, then different control variables are put in order, and finally all the control variables are put in together, which specifically forms five models, and the estimation results are shown in

Table 4. The results of baseline model.

Model	(4)	(5)	(6)	(7)	(8)
Variable	FCP	FCP	CFP	FCP	FCP
$WIA$	0.7243 ***	0.6488 ***	0.3076 ***	0.7296 ***	0.2629 ***
	(10.7602)	(10.1228)	(5.9963)	(10.883)	(5.5070)
$DA$		−0.2361 ***			−0.1862 ***
		(−8.4812)			(−9.1800)
$EIA$			0.9256 ***		0.8717 ***
			(23.3699)		(23.6852)
$RF$				13.7433 ***	8.0602 ***
				(3.0365)	(2.6659)
$R{F}^2$				−0.1189 ***	−0.0955 ***
				(−3.1995)	(−3.8702)
$R{F}^3$				0.0003 ***	0.0002 ***
				(2.8676)	(3.9009)
$Region$	Y	Y	Y	Y	Y
$Time$	Y	Y	Y	Y	Y
$Constant$	2434.5588 ***	2790.8703 ***	−694.7284 ***	2011.6238 ***	−298.0514
	(20.5742)	(23.4080)	(−4.3868)	(8.9232)	(−1.4558)
N	620	620	620	620	620
$R^2$	0.9384	0.9453	0.9686	0.9400	0.9737
$Adj\text{-}{R}^2$	0.9330	0.9404	0.9658	0.9344	0.9711

Note: The t values in parentheses; *** p < 0.01.

. It should be noted that the models in Table 4 have also added region and time fixed effects.

Table 4 reflects the effect of water-saving irrigation area on food crop production. Only region and time fixed effect are included in Model (4), and the regression results indicate that water-saving irrigated area positively affects food crop production at the 1% significance level, as shown by the fact that for every thousand hectares increase in water-saving irrigated area, food crop production will increase by 0.7243 million tons. To avoid potential estimation bias, the inclusion of control variables was considered for the remaining models. In Model (5), in addition to the inclusion of region and time fixed effects, the drought area is also included as a control variable. With the inclusion of drought area as a control variable, the effect of water-saving irrigated area on food crop production remains significant at the 1% level, but the estimated coefficient decreases relative to Model (4). In Model (6), effective irrigated area was added as a control variable. In this case, water-saving irrigated area still has a positive effect on food crop production at the 1% significance level, but the estimated coefficient decreases relative to both Models (4) and (5). In Model (7), besides including region and time fixed effects, when rainfall, the squared rainfall term and the cubic term are newly added as control variables, water-saving irrigated area still has a positive effect on food crop production at the 1% significance level, and the estimated coefficient increases relative to Models (4)–(6). In Model (8), when all the control variables (drought area, effective irrigated area, rainfall, squared and cubic terms of rainfall) are considered, water-saving irrigated area still has a positive effect on food crop production at the 1% significance level, but the estimated coefficient decreases relative to Models (4)–(7), showing that models which do not include the control variables will overestimate the effect of water-saving irrigated area on food crop production. Therefore, based on the estimation results of Model (8) with all control variables, it can be seen that each thousand hectares increase in water-saving irrigated area can increase food crop production by 0.2629 million tons. In summary, regardless of whether control variables are added, water-saving irrigation area has a significant positive impact on food crop production, so the empirical results support the research hypothesis H1. Therefore, the water-saving irrigation promotion has a significant role in promoting the production of food crops and has made a certain contribution to ensuring food security.

Table 4 reflects not only the effect of water-saving irrigated area on food crop production, but also the effect of other important control variables on food crop production. Models (5) and (8) both show that drought area has a negative effect on food crop production at the 1% significance level. Models (6) and (8) both show that effective irrigated area has a positive effect on food crop production at the 1% significance level. Models (7) and (8) both indicate that rainfall has a non-linear effect on food crop production at the 1% significance level. Therefore, drought area, effective irrigated area, and rainfall are also important factors for food crop production, which cannot be ignored in formulation of food security policies.

4.3. Mediating Effect Analysis

The benchmark regression model in Table 4 shows that water-saving irrigation area has a significant positive impact on food crop production, but its impact mechanism needs further analysis and verification. In order to verify that the water-saving irrigation area affects food crop production through the main agricultural production factors (crop sown area, chemical fertilizer and mechanized power), the mediation effect model is used to analyze it. Based on the test procedure of the mediation effect model [29,30,31], the next step of estimation is required after the benchmark model regression, which can be seen in

Table 5. Results of mediation effect model.

Model	(9)	(10)	(11)	(12)
Variable	CSA	CF	MP	FCP
$WIA$	0.2945 ***	0.0225 ***	0.2289 **	0.0822 **
	(5.6233)	(5.3646)	(2.2205)	(2.0242)
$EIA$	0.2051 ***	0.0244 ***	0.7749 ***	0.7104 ***
	(5.0806)	(7.5279)	(9.7490)	(21.5431)
$DA$	−0.0435 *	−0.0040 **	−0.1622 ***	−0.1542 ***
	(−1.9567)	(−2.2330)	(−3.7044)	(−9.1935)
$RF$	4.8619	−0.1242	−6.0174	6.3173 **
	(1.4658)	(−0.4670)	(−0.9216)	(2.5502)
$R{F}^2$	−0.0657 **	0.0001	0.0240	−0.0672 ***
	(−2.4261)	(0.0431)	(0.4512)	(−3.3145)
$R{F}^3$	0.0002 **	0.0000	0.0000	0.0002 ***
	(2.4931)	(0.0103)	(−0.2905)	(3.2881)
$CSA$				0.4450 ***
				(13.5267)
$CF$				1.8875 ***
				(4.3763)
$MP$				0.0310 *
				(1.7648)
$Region$	Y	Y	Y	Y
$Time$	Y	Y	Y	Y
$Constant$	5585.3013 ***	188.6106 ***	1774.8928 ***	−3194.7413 ***
	(24.8658)	(10.4725)	(4.0141)	(−13.0112)
N	620	620	620	620
$R^2$	0.9874	0.9746	0.9589	0.9826
$Adj\text{-}{R}^2$	0.9862	0.9721	0.9549	0.9808

Note: The t values in parentheses; * p < 0.1; ** p < 0.05; *** p < 0.01.

. It should be noted that the models in Table 5 have added region and time fixed effects, and added all control variables, because the estimation bias can be avoided by adding these control variables.

Table 5 demonstrates the effect of water-saving irrigation area on the main agricultural production factors. The results of Model (9) show that there is a positive effect of water-saving irrigation area on crop sown area at 1% significance level, specifically, for every thousand hectares increase in water-saving irrigation area, the crop sown area increases by 0.2945 thousand hectares. The results of Model (10) show that water-saving irrigation has a positive effect on chemical fertilizer at the 1% significance level, specifically, for every thousand hectares of water-saving irrigation, chemical fertilizer increases by 0.0225 million tons. The results of Model (11) show that water-saving irrigation area has a positive effect on mechanized power at 5% significance level, specifically, for each thousand hectares increase in water-saving irrigation area, mechanized power will increase by 0.2289 million kilowatt-hours. In summary, with the promotion of the national water-saving irrigation policy, water-saving irrigated area positively affects the main agricultural production factors (crop sown area, chemical fertilizer and mechanized power), which also illustrates the fundamental role of water resources in agricultural production.

According to the estimation results in Table 5, it is possible to reflect the mediating effect of water-saving irrigated area on food crop production through agricultural production factors (crop sown area, chemical fertilizer, and mechanized power). On the one hand, Models (9)–(11) reflect that water-saving irrigated area significantly affects the intermediary variables (crop sown area, chemical fertilizer and mechanized power). On the other hand, Model (12) reflects that the intermediary variables (crop sown area, chemical fertilizer and mechanized power) also significantly affect food crop production. Specifically, each increase of one thousand hectares of crop sown area will increase food crop production by 0.4450 million tons; each increase of ten thousand tons of chemical fertilizer used will increase food crop production by 1.8875 million tons; and each increase of ten thousand kilowatt-hours of mechanized power will increase food crop production by 0.0310 million tons. Therefore, according to the test steps of the mediating effect model, it can be found that water-saving irrigation area will affect food crop production through the mediating effect of agricultural production factors (crop sown area, chemical fertilizer, and mechanized power). In summary, the research hypothesis H2 is verified.

Model (8) in Table 4 reflects the total effect of water-saving irrigation area on food crop production, specifically, for each thousand hectares of water-saving irrigation area, the amount of food crop production will increase by 0.2629 million tons. Model (12) in Table 5 also reflects the direct effect of water-saving irrigation area on food crop production, specifically, for each thousand hectares of water-saving irrigation area, food crop production will directly increase by 0.0822 million tons, which also reflects the precision irrigation attribute of water-saving irrigation. Combined with Model (8) in Table 4 and Model (12) in Table 5, the indirect effect of water-saving irrigation area on food crop production can be explored. Combining the direct and total effects of water-saving irrigation area on food crop production, it can be found that the direct effect of water-saving irrigation area on food crop production is only 31.27%, while the indirect effect of water-saving irrigation area on food crop production by increasing agricultural production factors (crop sown area, chemical fertilizer and mechanized power) is as high as 68.73%. This shows that the water-saving irrigation promotion is essential in increasing the food crop production by reducing drought risk to keep agricultural production factors input.

4.4. Heterogeneity Analysis

The estimation results of the baseline model reflect the average effect of water-saving irrigated area on food crop production, but cannot reflect the heterogeneity of the effect. The following analysis provides heterogeneity analysis from two perspectives: rainfall and water-saving irrigation rate. The rainfall reflects average drought condition, while water-saving irrigation rate reflects popularity of water-saving irrigation promotion, which is the ratio of water-saving irrigation area to effective irrigation area.

From the perspective of rainfall, the samples can be divided into three groups, which can be called subsamples of small rainfall, middle rainfall and large rainfall, respectively. The small rainfall (S-RF) subsample is the set of the observations in which the rainfall is below 400 mm (arid and semi-arid areas), the middle rainfall (M-RF) subsample is the set of the observations, in which the rainfall is above 400 mm and below 800 mm (semi-humid areas), large rainfall (L-RF) subsample is the set of the observations, in which the rainfall is over 800 mm (humid areas). From the perspective of water-saving irrigation rate, the samples can be divided into three groups, which can be called subsamples of small water-saving irrigation rate (S-WIR), middle water-saving irrigation rate (M-WIR) and large water-saving irrigation rate (L-WIR), respectively. The samples can be divided into three equal groups according to the ranking of the magnitude of water-saving irrigation rate, which reflects the difference in the popularity of water-saving irrigation promotion. The estimation results of the heterogeneity analysis are shown in

Table 6. The results of heterogeneity analysis.

Model	(13)	(14)	(15)	(16)	(17)	(18)
Variable	FCP	FCP	FCP	FCP	FCP	FCP
$WSI$	1.0883 ***	0.3912 ***	0.3197 ***	1.2718 ***	0.6967 ***	0.3454 ***
	(8.6776)	(5.3680)	(3.9198)	(4.2861)	(6.2723)	(2.9366)
$EIA$	−0.7765 ***	0.9856 ***	0.7943 ***	0.6628 ***	1.0338 ***	0.4293 ***
	(−5.0730)	(23.1518)	(11.3380)	(6.3737)	(26.2278)	(3.9416)
$DA$	−0.1047 **	−0.1598 ***	−0.0748 ***	−0.1227 ***	−0.0752 ***	−0.0963 ***
	(−2.0020)	(−5.5622)	(−2.6203)	(−3.3632)	(−2.8464)	(−3.2938)
$RF$	−83.7938	279.2595 *	−11.0209	0.4170	8.9858 **	6.7732 *
	(−1.6207)	(1.6805)	(−1.2000)	(0.0740)	(2.4323)	(1.6588)
$R{F}^2$	3.1373	−4.9708 *	0.0490	−0.0411	−0.0715 **	−0.0683 *
	(1.3983)	(−1.7395)	(0.8389)	(−0.9685)	(−2.3635)	(−1.7807)
$R{F}^3$	−0.0341	0.0288 *	−0.0001	0.0001	0.0002 **	0.0002
	(−1.1048)	(1.7898)	(−0.6323)	(1.3183)	(2.0797)	(1.5173)
$Region$	Y	Y	Y	Y	Y	Y
$Time$	Y	Y	Y	Y	Y	Y
$Constant$	5588.8385 ***	−5753.7261 *	726.9029	−32.5735	−774.2303 ***	−259.8695
	(8.3157)	(−1.7583)	(1.4442)	(−0.0723)	(−4.2282)	(−1.3714)
$Subsample$	S-RF	M-RF	L-RF	S-WIR	M-WIR	L-WIR
N	105	223	292	207	207	206
$R^2$	0.9785	0.9874	0.9801	0.9809	0.9914	0.9856
$Adj\text{-}{R}^2$	0.9651	0.984	0.9765	0.9758	0.9888	0.9818

Note: The t values in parentheses; * p < 0.1, ** p < 0.05, *** p < 0.01; S-RF, M-RF, and L-RF mean small rainfall, middle rainfall, and large rainfall, respectively; S-WIR, M-WIR, and L-WIR mean small water-saving irrigation rate, middle water-saving irrigation rate, and large water-saving irrigation rate, respectively.

.

The heterogeneity of the effect of water-saving irrigated area on food crop production can be seen in Table 6. Models (13), (14), and (15) are the regression results under the subsample with less than 400 mm of rainfall (arid semi-arid areas), 400–800 mm of rainfall (semi-humid areas) and more than 800 mm of rainfall (humid areas), respectively. The results show that the effect of water-saving irrigated area on food crop production will be greater in the areas with less rainfall. Models (16), (17), and (18) show the regression results for the first 1/3, middle 1/3, and last 1/3 observations sorted by water-saving irrigation rate, respectively. The estimation results indicate that the effect of water-saving irrigation area on food crop production is greater in areas with lower water-saving irrigation rates. In summary, in areas with low rainfall (arid and semi-arid areas) or low water-saving irrigation rate (low promotion of water-saving irrigation), the expand of water-saving irrigation area is more beneficial to the increase in food crop production.

5. Discussion

Previous studies [25,26] have confirmed that the effect of water-saving irrigation on crop yield growth by crop physiological experiments. However, this study verifies the positive effect of water-saving irrigation promotion on food crop production based on an econometric approach. The experimental approach can only capture the direct effect of water-saving irrigation on agricultural production, while the econometric approach can capture not only the direct effect but also further analyze the indirect effect. In this study, the direct effect accounts for 31.27% of the overall effect, while the indirect effect accounts for 68.72% of the overall effect. The indirect effect is mainly through main agricultural production factors (crop sown area, chemical fertilizer, and mechanized power) to affect food crop production. The fundamental mechanism of water-saving irrigation promotion policy affecting food crop production is that water-saving irrigation policy can reduce drought risk in production, then achieve effective input of agricultural production factors based on the basic role of water in agricultural production.

Emerick et al. [21] found that improved technologies that reduce risk by protecting production have the potential to increase agricultural productivity. As an improved technology, water-saving irrigation can also increase agricultural productivity by reducing drought risk in agricultural production [22]. In this paper, we clarify that water-saving irrigation can increase food crop production through agricultural input factors, such as crop sown area, chemical fertilizer, and mechanized power. Because of the availability of data, the chemical fertilizer and mechanized power are all indicators of agriculture, not just food crop production. Though there is a difference between agriculture indicators and food crop production indicators, there will still be a high correlation between them. Therefore, the effect of water-saving irrigation area on the chemical fertilizer and mechanized power of food crop production will be less than the results in Table 5, and the magnitude of the intermediary effect is likely to decline, but the effect must exist. Further research on the mechanism of the role of water-saving irrigation promotion policies in food crop production can be carried out by analyzing the production behavior of farmers.

Some studies [25,26] discussed the effect of water-saving irrigation on crop yield growth, and Emerick et al. [21] discussed the affecting path of technology improvement on production, but the existing studies rarely analyzed the heterogeneity of the effect of water-saving irrigation promotion on agricultural production, which would ignore potential differentiated policy implementation of water-saving irrigation promotion. In this paper, we considered the possible heterogeneous effects of drought conditions and promotion stages of water-saving irrigation policy. The need to promote water-saving irrigation is even greater in arid areas. Its impact on increasing food crop production is greater at the beginning of the promotion of water-saving irrigation, in which the impact will gradually lessen as the promotion proceeds.

6. Conclusions

Using panel data from 31 provinces and cities across the country from 2000 to 2019, a two-way fixed model was used to empirically analyze the effect of water-saving irrigation area on food crop production, and further study the mediating effect of water-saving irrigation area on food crop production. A heterogeneity analysis was carried out, and the following conclusions were obtained:

(1)

Water-saving irrigation area not only can save agricultural water use, but also has a significant positive impact on food crop production. Overall, an average increase of one thousand hectares of water-saving irrigation area can increase food crop production by 0.2629 million tons. This shows that the promotion of water-saving irrigation area is conducive to the increase of food crop production, thereby ensuring national food security.

(2)

The expansion of water-saving irrigation area has direct and indirect effects on food crop production. The direct effect accounts for 31.27% of the overall effect. The indirect effect is mainly through main agricultural production factors (crop sown area, chemical fertilizer, and mechanized power) to affect food crop production, which accounts for 68.72% of the overall effect. The fundamental mechanism of water-saving irrigation promotion policy affecting food crop production is that water-saving irrigation policy can reduce drought risk in production, then achieve effective input of agricultural production factors based on the basic role of water in agricultural production.

(3)

The effect of water-saving irrigation area on food crop production is heterogeneous. The water-saving irrigation area has a higher impact on food crop production in areas with less rainfall than that with higher rainfall. The impact of water-saving irrigation area on food crop production is higher in areas with less water-saving irrigation rate than that with higher water-saving irrigation rate. So, water-saving irrigation promotion policy will be greater for food security in the beginning of policy implementation, or in area where there is relatively common drought risk.