Does agricultural water-saving policy improve food security? Evidence from the Yellow River Basin in China

According to Kinda et al. (2022), three dimensions of food security can be identified: availability, accessibility and utilization. The availability of food refers to the quantity of food supplied within domestic production and imports from foreign countries. Access to food through trade is vulnerable to international political conditions. Therefore, from the perspective of food availability, the food security of a country depends more on domestic production, which is also the focus of this paper. The past half-century has seen great growth in the amount of food production, but there is still more than one in seven people today who do not have enough food for consumption (Godfray et al., 2010). The threat to food security in developing countries can easily lead to social conflict (Martin-shields & Stojetz, 2019). Therefore, improving food security is still an essential issue over the world.

There are lots of research to identify the impact of various factors on food security, and the factors include agroecology (Bezner Kerr et al., 2021), climate change (Schmidhuber & Tubiello, 2007; Brown & Funk, 2008; Wheeler & von Braun, 2013; Campbell et al., 2016), COVID-19 (Amare et al., 2021; O'Hara & Toussaint, 2021), culture (Briones Alonso et al., 2018), management strategies (Bairagi et al., 2020; Bazzana et al., 2022), land rush (Kinda et al., 2022), trade (Dithmer & Abdulai, 2017), food price (Amolegbe et al., 2021), organic agriculture (Meemken & Qaim, 2018), precision agriculture (Gebbers & Adamchuk, 2010) and agricultural policy (Rosegrant & Cline, 2003; Bureau & Swinnen, 2018; Bizikova et al., 2020; Sibhatu et al., 2022). Besides, Rosegrant & Cline (2003) indicate that the crop yield may fall in many areas because of increasing water scarcity. The competition for water between agriculture and industry will affect the ability to produce food (Godfray et al., 2010). The efficient use of irrigation water may be an essential way to ensure food security (Cole et al., 2018). Among the studies of the impact of agricultural policies on food security, there are still few literatures analysing the impact of agricultural water-saving policies on food security. While water resource is an important factor in food production, the impact of agricultural water-saving policy on food security cannot be ignored. In addition, water-saving agriculture mainly refers to a way to promote agricultural production through the construction of water conservancy facilities and the promotion of water-saving irrigation (Wang et al., 2002).

The significant scarcity of water resources is also a problem in China, and the mismatch between supply and demand is still the fundamental obstacle to sustainable growth. The Chinese government introduced the National Agricultural Water-Saving Outline as a national strategy in 2012 to maintain food security and sustainable economic growth. The construction of water conservation projects and the promotion of modified irrigation technology that uses less water are the primary components of this policy. For water conservancy projects, the policy requires the continued construction of irrigation facilities in major grain-producing areas with severe water shortages and areas with fragile ecological environments and strives to solve the problems of ageing support facilities and large leakage losses. For the promotion of modified irrigation technology, the policy requires local governments to build a number of large-scale promotion projects for high-efficiency irrigation technology in areas with good agricultural production conditions.

The construction of water conservation projects and the promotion of high-efficiency irrigation technology will be beneficial to promote food security (Wang et al., 2002; Yang & Gao, 2021). In order to better understand whether the agricultural water-saving policy affects food security, it is now possible thanks to China's National Agricultural Water-Saving Outline, which was introduced in 2012. The implementation of the policy is mainly aimed at reducing irrigation water, which accounts for the largest agricultural water use and the increase in irrigation water relies on the land area of crop planting. As a result, the amount of exogenous suitable sown area will determine how densely the agricultural water-saving policy is implemented throughout Chinese prefectures (Nunn & Qian, 2011). In order to study the impact of agricultural water-saving policies, Xu & Yang (2022) use the size of suitable arable land to measure the regional differences of the agricultural water-saving policy. This provides the basis for analysis using the empirical strategy of the continuous Difference-in-Differences (DID).

The DID approach reveals that China's agricultural water-saving policy may increase food security. After considering other water-related policies, the above conclusions still hold. When further analysing the potential mechanism of the impact, we find that the policy can promote water productivity in agricultural production and the grain sown area, so we indicate that the policy may increase other input factors to promote food security through improving the effective utilization of water resources and thus ensuring the water demand for agricultural production. From the perspective of heterogeneous impacts, it can be found that, in the upper, middle and lower reaches of the Yellow River Basin in China, there are regional differences in the impact of the agricultural water-saving policy on grain production. In the upper reach of the basin, the impact is relatively small; in the middle reach, the impact is almost equivalent to the overall average impact; and the impact is greatest in the lower reach. The reason may be due to the differences in the geographical characteristics and land potential in the upper, middle and lower reaches of the basin. The slope and elevation of the upper reach are larger than those in the middle and lower reaches of the basin, while the land potential in the middle and lower reaches of the basin is greater than that in the upper basin. This conclusion can be also empirically tested by heterogeneity analysis.

This research adopts an empirical technique of DID to evaluate the impact of the agricultural water-saving policy on food security using data from prefectures in nine provinces of China's Yellow River Basin from 2000 to 2020. The rest is structured as follows: Section 2 describes the methods and data sources. Section 3 contains the key findings of this investigation. Section 4 contains the discussion. Section 5 highlights the study's findings.

The selected control variables include geographic, meteorological, agricultural and political variables. The geographic variables are slope and elevation, and the data come from the Resource and Environment Science and Data Center in China. The meteorological variables are precipitation and sunshine, and the data come from the National Meteorological Science Data Center. The agricultural variable is land potential; the data come from Liu et al. (2015). The political variable is a dummy variable, which will be equal to 1 if the prefecture is a provincial capital or sub-provincial city and 0 otherwise. In order to further analyse the potential mechanism of the impact of the policy on grain yield, the data in this paper also includes agricultural water productivity and the grain sown area. Agricultural water productivity means the real value added in agricultural production per unit of agricultural water use, in which the real value added in agricultural production is the real value compared to the first year. The data of value added in agricultural production and grain sown area come from the China Statistical Yearbook. The data of agricultural water use are from the China Water Resources Bulletin. The basic information of all the data is shown in Table 1.

Table 2 lists the initial findings from the empirical DID methodology used to calculate the effect of the agricultural water-saving policy on grain yield. The dependent variable is the grain yield, all columns have been added with year and prefecture fixed effects. Column 1 does not include any additional control variables. The estimated results show that the land area suitable for cultivation can increase grain yield at a significance level of 1% after the implementation of the policy, suggesting that the policy may promote grain yield. Columns 2–5 gradually add geographic, meteorological, agricultural and political control variables. The Columns 2 through 5 agree with the actual findings in Column 1 and the estimated coefficients have risen. Based on the selected specification incorporating all control variables, one increase of suitable cultivated land by 1% can lead to an increase of grain yield by 203,300 tonnes after the agricultural water-saving policy. Because the policy still increases grain yield at a significance level of 1% even with the addition of control variables, it may be beneficial for food security.

To ensure the conclusion that the grain yield can be increased due to the agricultural water-saving policy, it is necessary to further test whether it satisfies the parallel trend hypothesis. Therefore, flexible estimation is used for the further analysis below, as shown in Table 3. The fixed effects of year and prefecture are added to all columns. Column 1 just controls the fixed effects and Columns 2–5 gradually add geographic, meteorological, agricultural and political control variables. Different from the baseline estimation, the independent variable becomes the interaction terms between the suitable cultivated land area and year dummy variables, and the estimation results are compared with the first year (Nunn & Qian, 2011). The estimated results in Column 1 show that before 2012, the suitable cultivated land area had no effect on grain yield, and only after 2012 does it have an effect on grain yield at a 10% significance level. This is in accordance with the time node in 2012 when the policy went into effect. When the policy is not implemented, the suitable cultivated land area has no significant impact on grain yield. But only after the policy is implemented, the suitable cultivated land area has a significant impact on grain yield. The estimation results can indicate that the increase in grain yield is indeed due to the policy. Column 2 further adds geographic variables, and the estimated results indicate that there is a significant effect of the policy on grain yield simultaneously in the same year when the policy is put forward, but there is still no significant effect before 2012. Columns 3–5 gradually add meteorological, agricultural and political variables, and the results are in line with the first column. According to the estimated results in the Column 5, the significant impact of the policy on grain yield appears in the first year at a 10% significance level and appears in the third year at a 5% significance level. Therefore, the results of the flexible estimation confirm the parallel trend hypothesis, so the policy can increase grain yield, thereby promoting food security.

Other water resource policies, such as the Water Rights Reform (WRR) and the South-to-North Water Diversion Project (SNWDP), were implemented in China around 2012. The reform of the WRR mainly involves some prefectures in Gansu, Henan, Inner Mongolia and Ningxia, which was implemented in 2014. The SNWDP involves some prefectures in Henan, Shandong, which were implemented in 2013. The increase in grain yield may be also due to these two water-related policies. Thus, to further examine the effect of the agricultural water-saving policy on grain production, the two policies will also be included in the following analysis listed below. This will help to eliminate estimation bias brought by other water-related policies.

Table 4 presents the estimated results considering other policies. Columns 1–2 are the estimation results considering the WRR. After the WRR is considered, the suitable cultivated land area still significantly affects the grain yield after the policy, but the WRR has no significant effect on the grain yield. Columns 3–4 are the estimation results considering the SNWDP. After controlling the SNWDP, the suitable cultivated land area still significantly affects grain yield after the agricultural water-saving policy, and the SNWDP significantly negatively affects grain yield after adding all control variables. Columns 4–5 are the estimation results considering both the WRR and the SNWDP. After simultaneously controlling the two water-related policies, the suitable cultivated land area still significantly increases grain yield after the policy. According to the results in Column 6, increasing suitable cultivated land area by 1% increases grain yield by 202,100 tonnes, which just decreases slightly compared to the baseline estimation. Besides, the WRR can reduce the grain yield, but it is not significant, while the SNWDP can significantly reduce grain yield at a 5% significance level. Therefore, the agricultural water-saving policy can still promote grain yield even after addressing the contemporaneous water-related policies.

The findings of this study are in line with the body of previous literature (Wang et al., 2002; Yang & Gao, 2021; Xu & Yang, 2022). According to Wang et al. (2002), water-saving irrigation is crucial for ensuring food security. With empirical data, the causal link between water-saving irrigation and agricultural output has been established by Yang & Gao (2021). In addition, according to Xu & Yang (2022), China's agricultural water-saving policy, which promoted water-saving irrigation subsidies, not only did not cut agricultural water extraction but instead enhanced the extraction of agricultural water, which will benefit food production. But, why does the agricultural water-saving policy promote grain yield? The theoretical mechanism is that the agricultural water-saving policy assures grain production water demand through water-saving infrastructure and water-saving irrigation subsidies (Emerick et al., 2016). Subsidies for water-saving irrigation in agriculture will boost agricultural water use and production. In the Yellow River Basin, after ensuring the water need for grain production, other production factors will also increase, thereby promoting grain production. Among the many production factors, the input of land resources is the key production factor. Therefore, the following analysis will first focus on whether the agricultural water-saving policy has enhanced water productivity in agriculture, followed by a discussion of whether the agricultural water-saving policy has expanded grain-planted area.

Table 5 presents the estimated results of the influence of the agricultural water-saving policy on water productivity in agriculture, which is defined as the real value added in agricultural production per unit of water use in agriculture. It should be noticed that year fixed effects and prefecture fixed effects are included in every column. Other control factors are not introduced in Column 1, whereas political, agricultural, meteorological, and geographic control variables were gradually added in Columns 2–5. The results of Columns 1–5 show that the policy can raise water productivity in agriculture at a significant level of 1%. Therefore, the agricultural water-saving policy may ensure that water resources are used effectively in agricultural production. The policy may boost water productivity in both grain crops and cash crops at the same time, but there is no method to discern between grain production and others owing to data-gathering issues.

Table 6 shows the estimated findings of the influence of the agricultural water-saving policy on the area of grain sown, with year fixed effects and prefecture fixed effects included in all columns. Column 1 does not include any additional control variables, and the estimated findings demonstrate that the policy significantly increases the area of grain sown. The geographic, meteorological, agricultural and political control variables are subsequently introduced in Columns 2–5. The estimated findings in Columns 2–5 are consistent with Column 1, even though the estimated coefficients have risen in Columns 2–5. They all continue to demonstrate that the policy can enhance the grain sown area at a significance level of 1%. Therefore, the policy may promote grain yield by increasing the grain sown area. The agricultural water-saving policy may also increase grain production by increasing other production factors, but due to data acquisition problems, there are no other input factor data for grain production to verify.

Because there may be heterogenous in the influence of the agricultural water-saving policy on grain output in the upper, medium and lower reaches of the Yellow River Basin, we separate the samples of the basin into three subsamples of the higher, middle and lower reaches. The upper reach includes the prefectures in Qinghai, Gansu, Ningxia and Sichuan, the middle reach includes the prefectures in Inner Mongolia, Shaanxi and Shanxi, and the lower reach includes the prefectures in Henan and Shandong. Subsample-based estimations can be seen in Table 7, where all columns include various control variables as well as fixed effects for year and region.

The estimated results of the overall sample are presented in Column 1, which can be used as the comparison of the subsample estimation. Column 2 is the estimated result of the subsample in the upper basin. The estimated coefficient is 3.82, which is only 16% of the overall estimated result. Therefore, in the upper reach of the basin, although the policy can increase grain yield, the impact is relatively small. Column 3 shows the estimated results of the subsample in the middle reach of the basin. The estimated coefficient is almost equivalent to the average estimated coefficient in the overall sample of the basin. Column 4 is the estimated result of the subsample in the lower basin. The estimated coefficient is 78.17, which is 3.8 times the average estimated coefficient in the overall sample of the basin. Therefore, the agricultural water-saving policy can better promote grain production in the lower basin to ensure food security. Column 5 shows the estimated results of the subsamples in the middle and lower reaches of the basin, and the estimated coefficient is also more than 1.5 times the average estimated coefficient in the overall sample of the basin. Therefore, in the middle and lower reaches of the basin, the agricultural water-saving policy has a better effect on promoting grain yield.

The following is a further analysis of the heterogeneity of each characteristic variable, which can further verify the reasons for the regional heterogeneity, as shown in Table 8. All columns are added with year fixed effects, prefecture fixed effects and all control variables. To test the heterogeneity of characteristic variables, the independent variables are the interactions between the ln Land Area × Post and the slope, elevation, precipitation, sunshine, land potential and whether it is a provincial capital or a sub-provincial city. Column 1 shows that in prefectures with a higher slope, the impact of the policy on grain yield is relatively small. Column 2 shows that the higher the elevation, the smaller the impact of the policy on grain yield. Column 3 shows that the difference in precipitation does not affect the impact of the policy on grain yield. The possible reason is that the differences in water supply between prefectures caused by differences in precipitation are alleviated due to the policy. Column 4 shows that the longer the sunshine, the greater the impact of the policy on grain yield. Column 5 shows that prefectures with greater land potential have a greater impact on grain yields from the policy. Column 6 shows that the impact of the policy on grain yield is not significantly different between provincial capitals or sub-provincial regions and other prefectures. Therefore, the differences between the upper, middle and lower reaches of the Yellow River Basin may be due to geographical features and land potential. The slope and elevation of the upper basin are higher than those in the middle and lower reaches, and the land potential of the middle and lower reaches of the basin is higher than that of the upper basin.

This paper provides empirical evidence for studying the impact of the agricultural water-saving policy on food security. The implementation of the National Agricultural Water-Saving Outline of China in 2012 provides a quasi-experiment for our empirical analysis. Besides, the Yellow River Basin is one of the major areas of grain production in China. Based on this, we use the data of prefectures in nine provinces in the basin from 2000 to 2020 and employ a continuous DID to estimate the impact of the agricultural water-saving policy on food security as indexed by grain yield. In the model estimation, the treatment variable is the suitable cultivated land area, and its calculation process mainly considers major food crops such as wheat, corn, rice, bean and potato. According to Nunn & Qian (2011), the density in the policy will be consistent with regional differences in exogenous suitable cultivated land area. The estimated results show that the suitable cultivated land can significantly increase the grain yield only after 2012, so it can be considered that the policy can promote food security. After considering other contemporaneous water-related policies, the above conclusions still hold.

Although there is insufficient information to prove the potential mechanism of the impact of the policy on grain yield, we first clarify the impact of the policy on water productivity in agriculture and then examine the impact of the policy on the grain sown area, which is the most important input factor of grain production. The empirical results show that the policy also promotes agricultural water productivity and grain sown area, so we indicate that the policy may promote grain yield by increasing other input factors after improving the effective utilization of water resources and thus ensuring the water demand for grain production. This conclusion is consistent with Emerick et al. (2016), who believe that the improvement of agricultural technology will reduce agricultural risks, thereby increasing the input of production factors and thus increasing agricultural production. The agricultural water-saving policy can promote the construction of water conservancy facilities and the adoption of water-saving irrigation technologies, thereby reducing the risk of drought, thereby increasing the factors of grain production and thus ensuring food security. The potential mechanism can be further studied from the perspective of farmers’ behaviour. Therefore, future research can focus on the impact of agricultural water-saving policies on farmers’ behaviour in agricultural production.

From the perspective of heterogeneity analysis, in the upper, middle and lower reaches of the Yellow River Basin, there are regional differences in the impact of the agricultural water-saving policy on grain production. In the upper reach of the basin, the impact of the policy on grain yield is relatively small; in the middle reach of the basin, the impact is almost equivalent to the overall average impact in the basin; the impact is greatest in the lower basin. The reason may be due to the differences in geographical characteristics and land potential in the upper, middle and lower reaches of the basin. The slope and elevation of the upper basin are larger than those in the middle and lower reaches of the basin, while the land potential in the middle and lower reaches of the basin is greater than that in the upper basin. Therefore, strengthening the agricultural water-saving policy in the lower Yellow River Basin is more conducive to food security in China.

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

The authors declare there is no conflict.