Research on food security issues considering changes in rainfall

Abstract

Ensuring food security is not only vital to the adequate supply of food in the world, but also the key to the economic development and social stability of various countries. Based on the panel data of 29 provinces in China from 2016 to 2020, this paper selects the number of foodborne diseases patients and iodine deficiency disease patients as reference objects, uses stunting rate of children under 5 years old, malnutrition rate of children under 5 years old, obesity rate of children under 5 years old, and newborn visit rate to measure improving nutrition, proposes Meta Entropy Two-Stage Dynamic Direction Distance Function (DDF) Under an exogenous Data Envelopment Analysis (DEA) model to measure the efficiency of hunger eradication, food security, and improving nutrition under the influence of exogenous variable rainfall. The research results indicate that the sustainability of China’s agricultural economy is insufficient, and the focus of attention should be different in different stages. In addition, the average efficiency of the three regions generally shows a decreasing level in the eastern, western, and central regions. In order to improve China's ability to guarantee food security, we must continue to strengthen the construction of agricultural infrastructure, increase policy support for green agricultural production, promote the diversification of agricultural production, and enrich people’s agricultural product consumption varieties.

Introduction

The issue of food security is not only related to the livelihoods of countries but also to global development. In 2015, the United Nations General Assembly adopted the 2030 Agenda for Sustainable Development, putting forward 17 Sustainable Development Goals (SDGs), of which the second goal (SDG2) focuses on food security and commits to eradicating hunger, achieving food security, improving nutrition, and promoting sustainable agriculture by 2030, also known as the “Zero Hunger” goal. Food security is an important cornerstone and key issue for global sustainable development. Currently, Food production has made significant progress globally in eradicating hunger, food insecurity, and malnutrition. However, many people are still facing hunger and malnutrition due to the impact of various factors such as extreme weather, global COVID-19 pandemic, and geopolitical conflicts in recent years. In addition, the loss of arable land and urban expansion have adversely affected agricultural land and put enormous pressure on preventing the degradation of ecosystem service functions and adapting to climate change, which brings new uncertainties to global food security and new challenges to food security in China. To cope with the uncertainty of global food security, food security in China has become even more important.

Having entered a new stage of development, China has made significant achievements in food security. In the face of the global food crisis, China’s food production has achieved a good harvest for 19 consecutive years, the total food output has remained above 650 million tons for 8 consecutive years, the self-sufficiency rate of food rations has exceeded 100 percent and that of cereal foods has exceeded 95 percent, with the per capita food possession at approximately 480 kg, which is higher than the internationally acknowledged food security line of 400 kg, and China has achieved basic self-sufficiency in cereals and absolute security in rations. Using 9 percent of the world's arable land and 6 percent of its freshwater resources, China has been able to feed nearly 20 percent of its population, making a historic transition from hunger to subsistence to well-being. However, after a long period of sustained improvement, China's food security situation was reversed in 2015 due to multiple challenges, including agricultural environmental pollution and intensifying climate change. Climate change will have a negative impact on food production, which will increase the price of agricultural products and subsequently increase China’s food imports, which in turn will affect China’s level of food self-sufficiency. Currently, for every 0.1 °C increase in temperature, China’s yield per unit area of the three major food crops will decrease by about 2.6 per cent, and just a 1 per cent increase in precipitation will increase the yield per unit area by 0.4 per cent. In recent years, climate change has led to significant changes in China’s agroclimatic resources: From 1951 to 2021, the annual average surface temperature in China increased at a rate of 0.26 °C per decade; annual rainfall in China increased by an average of 4.9 mm per decade, showing a trend of “northern expansion of the rainfall belt”. The “double increase in water and heat” of climate change has led to significant changes in China’s agroclimatic resources, with the crop growing season lengthening by 1.8 days per decade. The impact of climate change on agricultural production is both negative and positive, but the negative impact of uneven rainfall and extreme weather on agriculture is significant and requires increased attention. The problem of uneven rainfall is reflected in the redistribution of global rainfall, with increased rainfall in some areas causing flooding and damage to crop roots and soil structure, thus reducing food production; reduced rainfall in some other areas leads to drought, which affects crop growth and development, and likewise reduces food production. Droughts used to exist in the northern regions of China, but seasonal droughts are now occurring in many southern regions, especially at critical times of crop growth, leading to significant reductions in crop yields. At present, China's food security still faces many risks and challenges, with new problems in both production and consumption, such as the contradiction between the basic balance of food supply and demand and structural scarcity, the contradiction between food production methods and the upgrading of food demand, and the contradiction between the international food market linkage and the volatility of domestic food prices, which has resulted in a potentially further deterioration of the food security situation. These food insecurity trends will ultimately increase the risk of malnutrition and further affect the quality of diets, affecting people’s health in different ways. Currently, with less than a decade to go before the achievement of the 2030 SDGs, the global food security situation is still spiraling downwards. Therefore, food security should always be a matter of crisis awareness.

Food security is affected by several factors, and rainfall is one of the major influences on food production. The regional impact of rainfall on production is complex and can have an impact on the total food production in China. Although the national rainfall has not shown a significant trend in the last 50 years, there are significant regional differences. In the scientific study of global change, there will be a long way to go to study the impact of rainfall changes on food production and food security in different regions of China and to propose effective countermeasures.

Literature review

The concept of food security was first officially introduced by the Food and Agriculture Organization of the United Nations (FAO) in 1974. It is defined as ensuring sufficient global availability of basic food supplies at all times, particularly in the case of natural disasters or other emergencies to prevent the exacerbation of food shortages, while steadily increasing food consumption in countries with low per capita intake to reduce production and price fluctuations. Make food security one of the basic rights of human life. This concept reflects people’s concerns about the occurrence of global food crisis at that time, recognizing that the decline of food supply plays a major role in promoting the expansion of hunger, while the instability of food prices caused by supply–demand imbalances exacerbates the severity of the hunger situation¹. Although the early definition of food security primarily emphasized the quantity of food supply, namely the accessibility of food, and measures to address hunger mainly focused on expanding food production, there has been a growing recognition of the importance of food stability as a crucial aspect of food security². As the world's economic situation evolves, people have gained a better understanding of food security, leading to an expanded conceptual framework. In 1982, the FAO revised the definition of food security to ensure sufficient food supply, stable food flows, and stable food sources for individuals or households. This new interpretation incorporates some micro considerations into the existing macro perspective, emphasizing the significance of balancing food supply and demand³. During the World Food Summit in 1996, the FAO updated the definition of food security to ensure that all individuals have physical and economic access to sufficient, nutritious and safe food at all times, and the effective utilization of these food nutrients, and defined four pillars of food security: availability, accessibility, utilization, and stability. In 2001, the FAO added the term “social” to the original definition of food security, which has become the most widely cited definition in current international food policies, that is, to ensure that all individuals at all times have physical, social and economic access to sufficient, nutritious and safe food to meet people's needs and preferences regarding food and promote people to lead positive and healthy lives.

Food security is closely related to people's lives, and it has always been the focus of academic attention. The existing research mainly analyzes the impact of resource endowment, climate change and government policy on food security, and then explores practical paths for various countries and regions to ensure food security in the future.

The literature primarily focuses on water resources, land resources, and human resources and other aspects to study the impact of resource endowment on food security. From the perspective of water resources, Kang et al.⁴ summarized the evolution of irrigation water productivity in China over the past 60 years, studied the differences in food productivity under different planting patterns, fertilization levels, and irrigation water consumption, analyzed the current situation of water resources’ impact on food security and explored comprehensive measures to improve agricultural water use efficiency in the future; Chloe et al.⁵ combining interviews and surveys from British farmers with the resilience theory to analyze the influencing factors of water scarcity risk and management strategies, found that farmers need to establish resilience by maintaining the buffer of water resources or increasing the availability of backup resources to minimize the negative impacts of water scarcity on food production and farmer’s economic income. From the perspective of land resources, Charoenratana and Shinohara⁶ pointed out that land and its legal rights are crucial factors for farmer income and agricultural production, and sustainable food security can only be achieved if land is kept safe. Li et al.⁷ indicated that while there has been a strong transition of cultivated land from non-staple food production to food production in the suburbs of Changchun after rapid urbanization, overall, the utilization diversity of suburban cultivated land in the black soil region of Northeast China has decreased, leading to a reduction in local supply of non-staple food. From the perspective of human resources, Yang et al.⁸ found that the relationship between non-agricultural employment and food production presents an inverted U-shaped pattern, which means that in the case of a small supply of non-agricultural labor force, increasing non-agricultural employment will have a positive impact on food output, while in the case of a large supply of non-agricultural labor force, increasing non-agricultural employment is not conducive to food output increase. Abebaw et al.⁹ investigated the impact of rural outmigration on food security in Ethiopia, and the results showed that rural outmigration significantly increased the daily calorie intake per adult by approximately 22%, reducing the gap and severity of food poverty by 7% and 4%, respectively.

Climate change. There is no consensus on the impact of climate change on food security. The majority of scholars assert that climate change will have significant negative effects on the availability, accessibility, and stability of food. Bijay et al.¹⁰ argued that the ongoing global climate change has caused a range of issues, including increased carbon dioxide, frequent droughts, and temperature fluctuations, which pose significant obstacles to pest management, consequently impeding increased food production. Muhammad et al.¹¹ concluded through empirical analysis that climate change has a substantial adverse impact on irrigation water, agriculture, and rural livelihoods, and the latter three have a significant positive correlation with food security, suggesting that climate change is detrimental to food security. Atuoye et al.¹² examined the influence of gender, migration, and climate change on food security, and their findings revealed that as global climate changes, the impact of controlling carbon emissions on non-migrant food insecurity in Tanzania is reduced, while it exacerbates the impact on migrant food insecurity. However, some scholars contend that climate change can improve agricultural production conditions in certain regions, thereby facilitating increased food production and positively impacting food security^13,14.

Government policy. Bizikova et al.¹⁵ evaluated 73 intervention policies in a sample of 66 publications, of which 49 intervention policies had a positive impact on food security, 7 intervention policies had a negative impact, and 17 intervention policies had no impact. Chengyou et al.¹⁶ used data such as mutual aid funds of impoverished villages in China to evaluate the effect of agricultural subsidies, and the empirical conclusion pointed out that agricultural subsidies can improve farmers' willingness to plant food, promote farmers in impoverished areas to increase the planting area, and help farmers improve their own food production capacity and economic income. Na et al.¹⁷ proposed that food subsidies can increase the working time of part-time farmers in agricultural work, especially in food planting, and promote farmers to better switch between non-agricultural work and agricultural work. This subsidy effect is conducive to maintaining sufficient supply and sustainable development of food production.

The existing literature studies food security from different perspectives and draws reasonable conclusions and policy recommendations, but it fails to analyze the issue of food security under the comprehensive effect of resource endowment, climate change and government policy. Based on this, this paper proposes the Entropy Window two-stage DDF to measure the efficiency of hunger eradication, food security and improving nutrition in 29 provinces of China under the influence of exogenous variable rainfall. From the perspective of food security, the impact of resource endowment, climate change and government policy on food security is comprehensively considered. In addition, in terms of climate change, different from the existing research focusing on the negative effects of high temperature, low temperature and drought on food production, this paper focuses on the impact of extreme changes in rainfall on food security, providing a certain complement to the existing literature on food security research.

Research methods

The evolution of DEA methods has seen many discussions of the dynamic DEA model. Färe and Grosskopf¹⁸ first established the concept of dynamic DEA, devised a form of dynamic analysis, and then proposed a delayed lag (carryover) variable for the dynamic model. Tone and Tsutsui¹⁹ then extended it to a dynamic DEA approach based on weighted relaxation, including four types of connected activities: (1) desired (good); (2) undesired (bad); (3) discretionary (free); and (4) non-discretionary (fixed). Battese and Rao²⁰ and Battese et al.²¹ next demonstrated that it is possible to compare the technical efficiencies of different groups using the Meta-frontier model. Portela and Thanassoulis²² proposed a convex Meta-frontier concept that can take into account the technology of all groups, the state-of-the-art level of technological production, as well as the communication between groups and can be further extended to improve business performance. O’Donnell et al.²³ proposed a Meta-frontier model for defining technical efficiency using an output distance function, which accurately calculates group and Meta-frontier technical efficiencies and finds that the level of technology of all groups is superior to the level of technology of any one group.

In this paper, the evaluation performance based on DDF is better, which can provide more accurate estimation results. Therefore, this paper modifies the traditional DDF model, combines Dynamic DEA with Network Structure^19,24 and Entropy method²⁵, and considers exogenous issues to construct Meta Entropy Two-Stage Dynamic DDF Under an Exogenous DEA Model in order to measure the efficiency of hunger eradication, food security, and improving nutrition in 29 provinces of China under the influence of rainfall.

The entropy method

In this model, the stage 2 (Hunger eradication and improving nutrition of sustainable stage) output item “Improving nutrition” covers four detailed indicators: (1) stunting rate of children under 5 years old; (2) malnutrition rate of children under 5 years old; (3) obesity rate of children under 5 years old; and (4) newborn visit rate. If these detailed indicators are put into DEA, then there will be problems that cannot be solved. Therefore, this model first uses the Entropy method and then finds the weights and output values of four detailed indicators of improving nutrition in stage 2. The Entropy method mainly includes the following four steps.

Step 1: Standardize the data of the four detailed indicators of improving nutrition in stage 2 in 29 provinces of China.

$$ \begin{array}{*{20}c} { r_{mn} = \frac{{\mathop {\max }\limits_{m} x_{mn} - x_{mn} }}{{\mathop {\max }\limits_{m} x_{mn} - \mathop {\min }\limits_{m} x_{mn} }} \left( {m = 1, \ldots , 29; n = 1, \ldots ,N} \right)} \\ \end{array} $$

(1)

Here, $r_{mn}$ is the standardized value of the $n$th indicator of the $m$th province; $\mathop {\min }\limits_{m} x_{mn}$ is the minimum value of the $n$th indicator of the $m$th province; and $\mathop {\max }\limits_{m} x_{mn}$ is the maximum value of the $n$th indicator of the $m$th province.

Step 2: Add up the standardized values of the four detailed indicators of improving nutrition in stage 2.

$$ \begin{array}{*{20}c} { P_{mn} = \frac{{R_{mn} }}{{\mathop \sum \nolimits_{m = 1}^{29} R_{mn} }} \left( {m = 1, \ldots , 29; n = 1, \ldots ,N} \right)} \\ \end{array} $$

(2)

Here, $P_{mn}$ represents the ratio of the standardized value of the $n$th indicator to the sum of the standardized values for the $m$th province.

Step 3: Calculate the entropy value (${\text{e}}_{{\text{n}}}$) for the ${\text{n}}$th indicator.

$$ \begin{array}{*{20}c} { e_{n} = - \left( {\ln 29} \right)^{ - 1} \mathop \sum \limits_{m = 1}^{29} \left[ {P_{mn} \ln \left( {P_{mn} } \right)} \right] \left( {m = 1, \ldots , 29; n = 1, \ldots ,N} \right)} \\ \end{array} $$

(3)

Step 4: Calculate the weight of the ${\text{n}}$th indicator $\left( {{\text{w}}_{{\text{n}}} } \right)$.

$$ \begin{array}{*{20}c} { w_{n} = \frac{{1 - e_{n} }}{{\mathop \sum \nolimits_{n = 1}^{N} \left( {1 - e_{n} } \right)}} \left( {n = 1, \ldots ,N} \right)} \\ \end{array} $$

(4)

Using the above steps, we are able to find the weights and output values of the four detailed indicators of improving nutrition in stage 2.

Meta entropy two-stage dynamic DDF under an exogenous DEA model

Suppose there are two stages in each $t \left( {t = 1, \ldots ,T} \right)$ time periods. In each time period, there are two stages, including agricultural production stage (stage 1), hunger eradication and improving nutrition of sustainable stage (stage 2).

In stage 1, there are $b \left( {b = 1, \ldots ,B} \right)$ inputs $x1_{bj}^{t}$, producing $a \left( {a = 1, \ldots , A} \right)$ desirable outputs $y1_{aj}^{t}$ and $o \left( {o = 1, \ldots , O} \right)$ undesirable outputs $U1_{oj}^{t}$. Stage 2 takes $d \left( {d = 1, \ldots , D} \right)$ inputs $x2_{dj}^{t}$, creating $s \left( {s = 1, \ldots ., S} \right)$ desirable outputs $y2_{sj}^{t}$ and $c \left( {c = 1, \ldots ., C} \right)$ undesirable outputs $U2_{cj}^{t}$; the intermediate outputs connecting stages 1 and 2 are $z_{hj}^{t} \left( {h = 1, \ldots ,H} \right)$; the carry-over variable is $c_{lj}^{t} \left( {l = 1, \ldots ,L} \right)$; the exogenous variable is $E_{vj}^{t} \left( {v = 1, \ldots ,V} \right)$.

Figure 1 illustrates the framework diagram of the model. In stage 1, the input variables are agricultural employment, effective irrigation area and total agricultural water use, and the output variables are total agricultural output value and agricultural wastewater discharge. In stage 2, the input variable is local financial medical and health expenditure, and the output variables are the number of foodborne disease patients, the number of iodine deficiency disease patients, and improving nutrition. The link between stage 1 and stage 2 is the intermediate output: total agricultural output value. And the exogenous variable is rainfall.

Fig. 1

Model framework.

Under the frontier, the DMU can choose the final output that is most favorable to its maximum value, so the efficiency of the decision unit under the common boundary can be solved by the following linear programming procedure.

• (a) Objective function

Efficiency of ${\text{DMUi}}$ is:

$$ \begin{array}{*{20}c} { \max GFE = \mathop \sum \limits_{t = 1}^{T} w_{1}^{t} \theta_{1}^{t} + w_{2}^{t} \theta_{2}^{t} } \\ \end{array} $$

(5)

Here, ${\text{w}}_{1}^{{\text{t}}}$ and ${\text{w}}_{2}^{{\text{t}}}$ are the weights for stages 1 and stage 2, and ${ }\theta_{1}^{{\text{t}}}$ and $\theta_{2}^{{\text{t}}}$ are the efficiency values for stages 1 and stage 2.

Subject to:

Stage 1: Agricultural production stage

$$ \begin{array}{*{20}c} \begin{gathered} \mathop \sum \limits_{j}^{n} \lambda_{j1}^{t} x1_{bj1}^{t} \le x1_{bi1}^{t} - \theta_{1}^{t} q_{bi1}^{t} \;\;\;\forall b, t = 1, \ldots ,T \hfill \\ \mathop \sum \limits_{j}^{n} \lambda_{j1}^{t} y1_{aj1}^{t} \ge y1_{ai1}^{t} + \theta_{1}^{t} q_{ai1}^{t} \;\;\; \forall a, t = 1, \ldots ,T \hfill \\ \mathop \sum \limits_{j}^{n} \lambda_{j1}^{t} U1_{oj1}^{t} \le U1_{oi1}^{t} - \theta_{1}^{t} q_{oi1}^{t} \;\;\;\forall o, t = 1, \ldots ,T \hfill \\ \mathop \sum \limits_{j}^{n} \lambda_{j1}^{t} = 1,\lambda_{j1}^{t} \ge 0\;\;\;\forall j, t = 1, \ldots ,T \hfill \\ \end{gathered} \\ \end{array} $$

(6)

Here, ${\text{q}}_{{{\text{bi}}1}}^{{\text{t}}}$, ${\text{q}}_{{{\text{ai}}1}}^{{\text{t}}}$, and ${\text{q}}_{{{\text{oi}}1}}^{{\text{t}}}$ denote the direction vectors associated with stage 1 inputs, desirable outputs, and undesirable outputs.

Stage 2: Hunger eradication and improving nutrition of sustainable stage

$$ \begin{gathered} \mathop \sum \limits_{j}^{n} \lambda_{j2}^{t} x2_{dj2}^{t} \le x2_{di2}^{t} - \theta_{2}^{t} q_{di2}^{t} \;\;\; \forall d, t = 1, \ldots ,T \hfill \\ \mathop \sum \limits_{j}^{n} \lambda_{j2}^{t} U2_{cj2}^{t} \le U2_{ci2}^{t} - \theta_{2}^{t} q_{ci2}^{t} \;\;\; \forall c,t = 1, \ldots ,T \hfill \\ \mathop \sum \limits_{j}^{n} \lambda_{j2}^{t} z_{{hj\left( {1, 2} \right)}}^{t} \ge z_{{hi\left( {1, 2} \right)}}^{t} + \theta_{{\left( {1,2} \right)}}^{t} q_{{hi\left( {1,2} \right)}}^{t} \;\;\; \forall h,t = 1, \ldots ,T \hfill \\ \mathop \sum \limits_{j}^{n} \lambda_{j1}^{t} = 1,\lambda_{j1}^{t} \ge 0 \;\;\; \forall j, t = 1, \ldots ,T \hfill \\ \end{gathered} $$

(7)

Here, ${\text{q}}_{{{\text{di}}2}}^{{\text{t}}}$, ${\text{q}}_{{{\text{ci}}2}}^{{\text{t}}}$, and ${\text{q}}_{{{\text{hi}}\left( {1,2} \right)}}^{{\text{t}}}$ denote the direction vectors associated with stage 2 inputs, undesirable outputs, and the intermediate outputs connecting stages 1 and 2.

The link of two periods

$$ \begin{array}{*{20}c} { \mathop \sum \limits_{j = 1}^{n} \lambda_{j}^{t - 1} c_{lj}^{t} = \mathop \sum \limits_{j = 1}^{n} \lambda_{j}^{t} c_{lj}^{t} \;\;\; \forall l, \forall t, t = 1, \ldots , T} \\ \end{array} $$

(8)

The exogenous variables

$$ \begin{array}{*{20}c} { \mathop \sum \limits_{j = 1}^{n} \lambda_{j}^{t} E_{vj}^{t} = E_{vi}^{t} \;\;\; \forall v,t = 1, \ldots ,T} \\ \end{array} $$

(9)

From the above results, the overall efficiency, the efficiency in each period, the efficiency in each stage, the efficiency in each stage in each period are obtained.

Input, desirable output, and undesirable output efficiencies

The disparity between the actual input–output indicators and the ideal input–output indicators under optimal efficiency represents the potential for efficiency improvement in terms of input and output orientation. This paper chooses the ratio of actual input–output values to the computed optimal input–output values as the efficiency measure for the input–output indicators. The relationship between the optimal value, actual value, and indicator efficiency is as follows:

$$ \begin{array}{*{20}c} {{\text{Input }}\;{\text{efficiency }} = \frac{{{\text{Optimal }}\;{\text{input}}}}{{{\text{Actual}}\;{\text{ input}}}}} \\ \end{array} $$

(10)

$$ \begin{array}{*{20}c} {{\text{Desirable}}\;{\text{ output }}\;{\text{efficiency }} = \frac{{{\text{Actual}}\;{\text{ desirable }}\;{\text{output}}}}{{{\text{Optimal }}\;{\text{desirable }}\;{\text{output}}}}} \\ \end{array} $$

(11)

$$ \begin{array}{*{20}c} {{\text{Undesirable}}\;{\text{ output}}\;{\text{ efficiency}} = \frac{{{\text{Optimal }}\;{\text{undesirable}}\;{\text{ output}}}}{{{\text{Actual }}\;{\text{undesirable}}\;{\text{ output}}}}} \\ \end{array} $$

(12)

If the actual input and undesirable output equals the optimal input and undesirable output, then the efficiencies of that input and undesirable output are equal to 1 and known as efficient. However, if the actual input exceeds the optimal input, then the efficiency of that input indicator is less than 1, which denotes being inefficient.

If the actual desirable output equals the optimal desirable output, then the efficiency of that desirable output is equal to 1 and is referred to as efficient. However, if the actual desirable output is less than the optimal desirable output, then the efficiency of that desirable output indicator is less than 1 and is considered inefficient. ME (Mean Efficiency) reflects the average efficiency of a certain region throughout the study period, with higher values indicating higher efficiency in that region.

Empirical study

Comparative analysis of total efficiency values considering and not considering exogenous variables

As shown in Fig. 2, in terms of the average total efficiency value for each region, without considering the exogenous variable rainfall, from 2016 to 2020, the average total efficiency values of the eastern, central, and western regions show a pattern of “eastern > western > central” in descending order. With the exogenous variable rainfall taken into account, the average total efficiency values for each region for each year were greater than the corresponding average total efficiency values without taking into account the exogenous variable rainfall, which may be attributed to the fact that rainfall plays a key role in irrigating the farmland and replenishing the soil moisture, which is an important factor in the process of agricultural production, and that the addition of rainfall has a more pronounced marginal effect on the increase in the total efficiency values. With the exogenous variable rainfall taken into account, the average total efficiency values for each region in each year are larger than the corresponding average total efficiency values without taking into account the exogenous variable rainfall, indicating that there is more room for improvement in the average total efficiency values without taking rainfall into account than in the efficiency values with rainfall taken into account. Except for 2016, when the average total efficiency value of the western region was greater than that of the eastern region and the central region, the average total efficiency values of the eastern, central, and western regions from 2017 to 2020 also showed a pattern of “eastern > western > central” from largest to smallest. It can be concluded that whether or not the exogenous variable rainfall is taken into account, the eastern region has a better overall efficiency in agricultural production and achieving food security than the western and central regions due to its better agricultural infrastructure, good economic base, and better educated labor force.

Fig. 2

Average efficiency by region from 2016 to 2020.

The three regions of the East, Central and West maintain a similar fluctuating upward trend. The average efficiency in the eastern and western regions is relatively high, and the five-year fluctuation interval is small, ranging from 0.75 to 0.85. After considering the exogenous variable rainfall, the average total efficiency value in the central region increased from 0.62 to 0.66. However, compared with the eastern and western regions, the total efficiency in the central region is still at a lower level and the five-year fluctuation interval is larger, between 0.55 and 0.70, with the largest fluctuation interval in the average efficiency in 2017–2018, at − 0.11. This may be due to the downsizing of grain sowing area under the structural reform of the agricultural supply side, leading to a small decline in the total national grain output in 2018, which in turn affects the level of efficiency in eradicating hunger, guaranteeing food security and improving nutrition. From this, it can be concluded that the eastern and western regions should give full play to their original advantages and promote the modernization and sustainable development of agricultural production in order to accelerate the achievement of the three major goals of eradicating hunger, guaranteeing food security and improving nutrition, while the central region still has more room for improvement and needs to further play the role of agricultural policies to alleviate the people’s worries about food.

Table 1 Average efficiency by province and city from 2016 to 2020 demonstrates the average efficiency values for each province and city from 2016 to 2020 when rainfall is considered and not considered. From the point of view of the annual average total efficiency by province, after considering the exogenous variable rainfall, the efficiency value of most provinces has been improved. The average efficiency has also been improved from 0.6134 to 0.6189. Among them, the efficiency value of Qinghai increases from 0.8167 to 1, and the ranking also rises from 11th to 1st place. Qinghai is deep inland, with less rainfall throughout the year, and its agricultural and animal husbandry production is more sensitive to the changes of rainfall, and the addition of exogenous variable rainfall makes the average total efficiency more accurately portrayed, and achieves the DEA validity. Shandong’s ranking drops from 9 to 11th after considering the exogenous variable rainfall. As a major agricultural province, Shandong’s food production will be seriously affected by persistent heavy precipitation and other extreme weather events, which indicates that Shandong needs to take measures to strengthen the ability of its agricultural production to cope with extreme precipitation.

Table 1 Average efficiency by province and city from 2016 to 2020.

Two-stage average efficiency analysis

The average efficiency values of the two stages in both cases of considering exogenous variable rainfall and not considering exogenous variable rainfall are very similar, indicating that exogenous variable rainfall does not have much effect on the efficiency of stage 1 and stage 2, and therefore only the specific case with exogenous variable rainfall is discussed. Figures 3 and 4 show the efficiency values for Stage 1 and Stage 2 for each province and city for the years 2016–2020 when rainfall is considered. As shown in Fig. 3, the difference between the efficiency values for Stage 1 and Stage 2 is still relatively significant. The efficiency of agricultural production in Stage 1 is significantly higher than that of hunger elimination, food security and nutritional improvement in Stage 2, and the fluctuation is relatively smooth, which indicates that there is still much room for improvement in China’s food production in terms of hunger elimination, food security and nutritional improvement, and that how to develop high-quality and high-efficiency agriculture and increase the output of food units is an urgent problem to be solved by each province.

Fig. 3

Comparison of the average efficiency of the two phases by province from 2016 to 2020.

Fig. 4

Average efficiency values for the two phases in each province from 2016 to 2020.

Specifically, there are large gaps in the efficiency of agricultural production in China's provinces, which can be roughly categorized into three types: the first type has an efficiency value of 1, realizing the DEA is effective, and is filled in green in Fig. 4; the second type has an efficiency value between 1 and the average, and is filled in yellow in Fig. 4; and the third type has an efficiency value below the average, and is filled in red in Fig. 4.

In the first stage, the first category is Shanghai, Shandong, Tianjin, Beijing and other 15 provinces, whose agricultural production efficiency values are all 1, at the meta-frontier, and these provinces rely on a solid economic foundation and sound agricultural infrastructure to realize the optimal efficiency of effective inputs and outputs; the second category is Guangxi, Hubei, Sichuan, and Liaoning, whose agricultural production efficiencies are higher than the national average and close to the meta-frontier; the third category consists of 10 provinces such as Gansu, Inner Mongolia, Jilin, Heilongjiang, etc., whose economic development is relatively slow, meteorological conditions are poor, agricultural production is susceptible to meteorological disasters, and the efficiency of agricultural production is below the average level, among which the value of Gansu’s agricultural production efficiency is the lowest, 0.496.

In the second stage, the first category includes seven provinces, including Yunnan, Tianjin, Beijing, and Ningxia, which either have higher economic levels or better climatic conditions, and have the highest efficiency in eradicating hunger, achieving food security, and improving nutrition, with an efficiency value of 1; the second category includes eight provinces, including Shanghai, Chongqing, Jilin, and Shaanxi, which have an efficiency in eradicating hunger, achieving food security, and improving nutrition higher than the national average, and are close to the meta-frontier; the third category includes 14 provinces, including Fujian, Shanxi, Inner Mongolia, and Guangxi, which are below the national average, among which Sichuan has the lowest efficiency value of 0.1, which is evident that Sichuan, as a “Heavenly Grain Silo,” is more likely to speed up the realization of mechanization and digital development to improve comprehensive grain production capacity.

In summary, provinces with high efficiency values in agricultural production and in eradicating hunger, achieving food security and improving nutrition can be categorized into two groups, one of which is the developed and coastal provinces with good economic and climatic conditions, such as Beijing, Shanghai, Tianjin, and Hainan, can enhance agricultural sustainable efficiency and actively promote the sustainable development of the agricultural economy; the other category is the provinces with relatively backward economic development, including Yunnan, Ningxia, Qinghai, Heilongjiang and other central and western regions, although their development is relatively late and low, they have unique climatic conditions, geographic conditions, ecological conditions, and other resource advantages, which bring opportunities for sustainable agricultural development in the central and western regions. As for the provinces with lower efficiency values for agricultural production and hunger eradication, reaching food security and improving nutrition, they are not only affected by the level of economic development and ecological conditions such as climate and environment, but also by the level of urbanization, such as Fujian, Jiangsu, Zhejiang, Guangdong and other eastern coastal provinces with a high level of urbanization will also face pressure on the supply of agricultural products as the sown area of crops continues to decrease due to a combination of factors such as the occupation of arable land by construction sites as well as abandonment of land.

Comparative analysis of output indicator efficiency in the regions

Taking rainfall as an exogenous variable into account, the efficiency of the number of foodborne diseases patients and improving nutrition showed a higher pattern in the eastern and western regions and a lower pattern in the central region. Table 2 shows the efficiency values of each output indicator for 2016–2020. From 2016 to 2020, the efficiency of these two output indicators in the eastern and western regions showed an upward trend, while that in the central region showed a downward trend. It shows that the contribution of agricultural production to food security in the eastern and western regions is small, and more perfect institutional measures should be formulated to ensure food security; the contribution of agricultural production to improving nutrition in the central region is relatively small, and corresponding health expenditures need to be increased to improve people's own nutritional supplements. In terms of the efficiency of the number of iodine deficiency disease patients, the efficiency in the eastern and central regions was low and showed a downward trend from 2016 to 2020, while the efficiency in the western region was high and the fluctuation was relatively small. As people in the eastern and central regions can easily buy kelp, laver and other iodine-rich foods, local residents eat iodine-rich food at high frequency and in large amounts, while in the western region, which is far from the sea, daily eating may not meet the human body's daily demand for iodine. Therefore, in order to reduce the incidence of iodine deficiency diseases caused by geographical location and dietary habits, governments in the western region need to speed up the opening of transportation channels and purchasing channels for iodized salt and iodine-rich foods.

Table 2 Efficiency of output indicators by region from 2016 to 2020.

Conclusions and policy recommendation

The key to sustainable agricultural development lies in the organic integration of ecological sustainability, economic sustainability, and social sustainability, emphasizing the coordination between agroecological production capacity and human development. The conclusions of this paper are as follows.

First, in the total factor efficiency analysis, the average total efficiency values of the eastern, central, and western regions in each year when the exogenous variable rainfall is taken into account are higher than the corresponding average total efficiency values without considering exogenous variable rainfall. This may be due to the fact that rainfall is an important factor in the agricultural production process and the inclusion of rainfall has a more pronounced marginal effect on the increase in the total efficiency value. In addition, there is a certain difference between the average total efficiency values of the eastern and western regions regardless of whether exogenous variable rainfall is considered. Still, the difference is not very large, and all three regions maintain a similar trend of fluctuating upward. However, the average total efficiency value of the central region is still at a lower level compared to the eastern and western regions, and the fluctuations of the eastern and western regions over the 5 years are small, fluctuating between 0.75 and 0.85, while the average efficiency of the central region over the 5 years is low and fluctuates greatly, fluctuating between 0.55 and 0.70, and the fluctuations of the average efficiency in 2017–2018 are the largest, at − 0.11. Besides the average efficiency of the eastern region was slightly lower than that of the western region in 2016, the average efficiency of the three regions generally showed a decreasing hierarchy of eastern, western, and central regions one by one. In terms of the annual average total efficiency of each province, after considering the exogenous variable rainfall, the efficiency values of most provinces have improved, with Qinghai's average total efficiency rising to 1, achieving optimal input–output efficiency.. In contrast, Shandong's average efficiency ranking has declined.

Second, under the condition of considering the exogenous variable rainfall, the efficiency value in stage 1 (agricultural production stage) is significantly higher than the efficiency value in stage 2 (eliminating hunger, achieving food security and improving nutrition), and the fluctuation is relatively smooth, which suggests that China's food production still has a large room for improvement, and that the focus of attention should be different in different stages. Specifically, in stage 1, the provinces with lower agricultural production efficiency values belong to the central and western inland provinces with slower economic development and poorer meteorological conditions, while in stage 2, the provinces with lower efficiency value also include the more economically developed eastern coastal provinces, such as Fujian, Jiangsu, Zhejiang, Guangdong, etc. The rapid population growth in the developed eastern coastal areas, coupled with the impact of the construction of arable land and the impact of a combination of factors such as the abandonment of land, crop sowing area has been decreasing, resulting in per capita arable land area is lower than the national average level. This shows that although the developed eastern coastal provinces have a better foundation for agricultural development, they are also facing enormous pressure on the supply of agricultural products and increasingly fierce competition in the future industrial development.

Third, a comparative analysis of the efficiency of output indicators by region, taking into account the exogenous variable of rainfall, reveals that the efficiency of the number of foodborne diseases patients and improving nutrition are both high in the eastern and western regions and low in the central region and that the efficiency of these two output indicators shows a rising trend in the eastern and western regions and a declining trend in the central region in the period from 2016 to 2020. In terms of the efficiency of the number of iodine deficiency disease patients, the efficiency of the eastern and central regions is low and shows a similar downward trend over the five-year period, while the efficiency of the western region is high and fluctuates relatively little, with no significant trend of change.

Through the above empirical analysis, it can be seen that rainfall, an exogenous variable, has a significant impact on the average efficiency in the eastern, central and western regions. Therefore, this paper puts forward corresponding policy recommendations on hunger eradication, food security and improving nutrition. The specific recommendations are as follows:

First, continue to strengthen the construction of agricultural infrastructure and increase the per capita arable land. All regions, especially the central and western regions, need to continue to increase investment in agriculture, build agricultural infrastructure such as water conservancy facilities, transportation facilities, and electric power facilities, promote the transformation and upgrading of old agricultural infrastructure, help the rapid development of agricultural mechanization in China, further enhance the ability to resist natural disasters, and improve agricultural output and production efficiency. In this way, the contradiction between food production and the growing rigid demand for food can be alleviated.

Second, increase policy support for green agricultural production to ensure China's food security. Due to the developed industry and serious pollution, the eastern region should pay more attention to green agricultural production. Each province shall formulate corresponding subsidy plans for green agricultural production according to the specific conditions of the province, strengthen green technology to lead the green development of agriculture, increase the enthusiasm of farmers to carry out green agricultural production, promote the promotion of green agricultural production, decrease the use of harmful fertilizers, pesticides, agricultural film, etc., to reduce agricultural pollution, so as to increase the supply of green agricultural products on the market to decrease the prevalence of foodborne diseases.

Third, promote the diversification of agricultural production and enrich people's agricultural product consumption varieties. On the one hand, each province extends local agricultural production varieties according to climate conditions and resources, rationally layout the supply structure of agricultural products, and increase policies to encourage farmers to carry out diversified agricultural production. On the other hand, some regions are limited by resource endowments and cannot expand the types of agricultural production, so it is necessary to speed up the construction of infrastructure such as logistics and preservation, improve the system of connecting production and marketing of agricultural products, enrich the "vegetable basket" of people in these regions with poor agricultural resources, and meet people's diversified consumption demand for agricultural products. In addition, nutrition guidance, publicity and education should be strengthened to raise people's awareness of rational diet and nutritious diet.