Agricultural Efficiency in Different Regions of China: An Empirical Analysis Based on Dynamic SBM-DEA Model

Abstract

This study applies the dynamic slacks-based measure (DSBM) and the total-factor agricultural efficiency (TFAE) to explore the overall agricultural production efficiency of 30 administrative regions and the eastern, central, and western regions of China from 2012 to 2016. The previous literature has mainly focused on China’s economic development and experience, but as the economy continues to grow, more food is needed and agricultural labor is shifting to urban areas. Little attention has been paid to the impact of limited agricultural land on agricultural production efficiency. Therefore, this paper uses the agricultural land area as the carry-over variable and uses agricultural labor, total agricultural machinery power, rural electricity consumption, agricultural fertilizer use, and agricultural GDP as variables to discuss the efficiency of agricultural production in different regions. The empirical results show that from 2012 to 2016, the best administrative region in terms of overall agricultural production efficiency in China was the east. In terms of the overall analysis of the region, the east had the highest overall agricultural production efficiency, while the central region had the lowest. The input variable that needed the most improvement was rural electricity consumption, with the largest adjustment in rural electricity consumption being observed in Hebei and Liaoning provinces of the eastern region. Furthermore, from 2012 to 2016, both overall agricultural production efficiency and agricultural GDP showed upward trends. However, adjustments are still needed for other relevant agricultural input variables to effectively allocate resources and improve the overall agricultural production efficiency.

1. Introduction

Due to climate change, the world is facing unprecedented challenges such as droughts, floods, water quality changes, and pest infestations, which affect global food quality, harvests, and distribution. In 1996, the Food and Agriculture Organization of the United Nations (FAO) released the “World Food Summit Declaration” and the “World Food Summit Plan of Action”, both of which highlighted the global food, agriculture, forestry, and fisheries sectors facing problems such as pest infestations, droughts, and human resources. In 2014, the Intergovernmental Panel on Climate Change (IPCC) held a meeting in Japan and pointed out that “climate change is impacting food and human security”. In 2019, the IPCC released the Climate Change and Land Report, which highlighted that the global average yield of corn, wheat, and soybeans had decreased from 1981 to 2010 due to the impact of climate change. It was predicted that climate change would cause a 5% to 30% decline in global grain production capacity by 2050. In 2020, the OECD and FAO jointly published the “2020–2029 Agricultural Outlook Report”, which stated that although the global grain output would continue to increase in the next decade, production technology and planting technology improvements would be responsible for 85% of the increase in crop output, with 10% due to multi-period crop harvests and only 5% due to an increase in cultivated land area. With the outbreak of COVID-19 in 2020, the global food supply chain is facing challenges, and solutions must be proposed for agricultural production and the labor market. As the second-largest economy in the world, China urgently needs to address the issue of insufficient per capita grain production due to changes in industrial structure, the gap between the urban and rural agricultural labor forces, climate change, extreme weather conditions, and differences in geographical environment and land resource advantages.

Studying the agricultural policies during the 12th Five-Year Plan period can provide us with valuable experience and lessons, so as to better understand today’s agricultural policies and provide guidance for future agricultural development. From 2012 to 2016, China’s agricultural development was in the implementation period of the 12th Five-Year Plan. At that time, the Chinese government mainly focused on improving agricultural productivity and farmers’ income, as well as strengthening rural infrastructure construction. These policies laid the foundation for agricultural development in the 14th Five-Year Plan of China. However, in the past decade, China’s agricultural development direction has undergone significant changes. From the basis of simply improving productivity and income in the 12th Five-Year Plan to the 14th Five-Year Plan, China’s agricultural development has shifted towards a greater emphasis on modernization, quality and efficiency, and competitiveness. This shift in focus highlights the need for a dynamic analysis of China’s agricultural production efficiency, rather than just using static analyses that cannot compare annual data or understand China’s agricultural development trends. To address this, the paper employs a dynamic analysis, using DSBM and TFAE to analyze China’s food production issues, and provides some recommendations for China’s agricultural policies. To gather the literature related to agricultural production efficiency, productivity changes, and other relevant studies, the following steps were taken: Previous studies have found that important factors affecting agricultural productivity in Asian countries include scientific and technological progress [1,2,3,4,5], taxation [6], and capital investment [7]. The main factors affecting agricultural efficiency include machinery and fertilizer inputs [6,8,9,10,11,12], energy use [8,9,10,11,12,13,14,15,16], irrigation infrastructure [16,17], agricultural machinery [18], and pesticides [19]. The main factors related to agriculture and the environment include the burning of crops [20,21], greenhouse gas emissions [22], national agricultural policies [23], and differences between greenhouse and open field cultivation [24]. The existing literature has extensively discussed energy and environmental efficiency, as well as agricultural productivity, across countries, often relying on DEA or Malmquist productivity index analysis. However, there has been comparatively less attention given to analyzing the agricultural production efficiency within a single country. Typically, the input variables in these studies include land, labor, capital, diesel, machinery, and fertilizer, while the output variables mainly include crop outputs or agricultural gross output values. It is worth noting that the total power of agricultural machinery, rural electricity consumption, and cultivated land area have been underutilized as variables in these analyses. Therefore, this study utilized the DSBM and TFAE models and expanded the list of input variables to include the agricultural labor force, the total power of agricultural machinery, rural electricity consumption, pesticides, and fertilizers. The selected output variable was the agricultural GDP, while the cultivated land area was set as an intertemporal variable and was kept as a carry-over (fixed) to evaluate the total-factor efficiency of the agricultural production in different regions of China. This decision was made due to the challenge of altering the cultivated land area in the short term.

2. Research Method

This study is based on DEA for empirical analysis. DEA can handle multiple inputs and outputs at the same time, has a wide range of applications, and can evaluate the efficiency results of decision makers or decision-making units (DMUs).

2.1. DEA and DSBM Models

DEA is a theory based on a boundary production function proposed by Farrell in 1957. It is used to analyze the energy efficiency of DMUs, connect the most efficient production point to a theoretical production boundary, and prove that the gap between any real production point and the theoretical production boundary represents the efficiency of the production point. After the development of early CCR [25] and BCC [26] models, the authors of [27] proposed the DSBM model, using a carry-over as the cross-period link (good, bad, free, and fixed). There are n DMUs. Each DMU has input and output variables in phase t, which are linked to the next phase (phase t + 1) through carry-over, as shown in Figure 1.

Set n DMUs (j = 1, …, n) through T terms (t = 1, …, T). DMUs in each phase have f discrete input items (i = 1,…, f, q non-discrete (fixed) input items (i = 1, …, q), S output items (i = 1,…, s) and ν Non-discretionary (fixed) output items (i = 1, ν), $x_{ijt}\left(i=1,\dots ,f\right),x_{ijt}^{fix}\left(i=1,\dots ,q\right),y_{ijt}\left(i=1,\dots ,s\right),y_{ijt}^{fix}$(i = 1,…, ν) To measure the value of DMU in term T. Carry-over has four forms $z^{good}, z^{bad}, z^{free}, z^{fix}$.

The following is the linear programming formula of the DSBM basic model. The production of the basic model may be set as follows: $\left\{x_{it}\right\},\left\{x_{it}^{fix}\right\},\left\{y_{it}\right\},\left\{y_{it}^{fix}\right\},\left\{z_{it}^{good}\right\},\left\{z_{it}^{bad}\right\},\left\{z_{it}^{free}\right\},\left\{z_{it}^{fix}\right\}$ The definition is as follows:

$$x_{it}\ge {\sum }_{j=1}^n{x}_{ijt}{\lambda }_j^t,\left(i=1,\dots ,\mathrm{f};t=1,\dots ,T\right)$$

$$x_{it}^{fix}={\sum }_{j=1}^n{x}_{ijt}^{fix}{\lambda }_j^t,\left(i=1,\dots ,q;t=1,\dots ,T\right)$$

$$y_{it}\le {\sum }_{j=1}^n{y}_{ijt}{\lambda }_j^t,\left(i=1,\dots ,\mathrm{s};t=1,\dots ,T\right)$$

$$y_{it}^{fix}={\sum }_{j=1}^n{y}_{ijt}^{fix}{\lambda }_j^t,\left(i=1,\dots ,\nu ;t=1,\dots ,T\right)$$

$$z_{it}^{good}\le {\sum }_{j=1}^n{z}_{ijt}^{good}{\lambda }_j^t,\left(i=1,\dots ,\mathrm{ngood};t=1,\dots ,T\right)$$

$$z_{it}^{bad}\ge {\sum }_{j=1}^n{z}_{ijt}^{bad}{\lambda }_j^t,\left(i=1,\dots ,\mathrm{nbad};t=1,\dots ,T\right)$$

$$z_{it}^{free}:\mathrm{free},\left(i=1,\dots ,\mathrm{nfree};t=1,\dots ,T\right)$$

$$z_{it}^{fix}={\sum }_{j=1}^n{z}_{ijt}^{fix}{\lambda }_j^t,\left(i=1,\dots ,\mathrm{nfix};t=1,\dots ,T\right)$$

$${\lambda }_j^t\ge 0,\left(j=1,\dots ,n;t=1,\dots ,T\right)$$

$${\sum }_{j=1}^n{\lambda }_j^t=1,\left(t=1,\dots ,T\right)$$

(1)

${\lambda }_j\in R^n\left(t=1,\dots ,T\right)$ represents the intensity vector of the t period. Equation (2) is a mathematical formula that satisfies the inter-temporal variability conditions from period t to period t + 1, and is an important restriction for DSBM to link activities from period t to period t + 1. Equation (2) β can be expressed as good, bad, free and fix, which, respectively, represent the number of good links, bad links, changeable links, and immutable links. In order to ensure the connectivity of the period t to t + 1, the following assumptions must be met:

$${\sum }_{j=1}^n{z}_{ijt}^{\beta }{\lambda }_j^t={\sum }_{j=1}^n{z}_{ijt}^{\beta }{\lambda }_j^{t+1} , \left(\forall i;t=1,\dots ,T-1\right)$$

(2)

Using (2) for production mode, we can set the $DM{U}_o$ (o = 1,…,n) as follows:

$$x_{iot}={\displaystyle \sum }_{j=1}^n{x}_{ijt}{\lambda }_j^t+s_{it}^{-} ,\left(i=1,\dots ,f;t=1,\dots ,T\right)$$

$$x_{iot}^{fix}={\displaystyle \sum }_{j=1}^n{x}_{ijt}^{fix}{\lambda }_j^t ,\left(i=1,\dots ,q;t=1,\dots ,T\right)$$

$$y_{iot}={\displaystyle \sum }_{j=1}^n{y}_{ijt}{\lambda }_j^t-s_{it}^{+} ,\left(i=1,\dots ,s;t=1,\dots ,T\right)$$

$$y_{iot}^{fix}={\displaystyle \sum }_{j=1}^n{y}_{ijt}^{fix}{\lambda }_j^t ,\left(i=1,\dots ,\nu ;t=1,\dots ,T\right)$$

$$z_{iot}^{good}={\displaystyle \sum }_{j=1}^n{z}_{ijt}^{good}{\lambda }_j^t-s_{iot}^{good} ,\left(i=1,\dots ,ngood;t=1,\dots ,T\right)$$

$$z_{iot}^{bad}={\displaystyle \sum }_{j=1}^n{z}_{ijt}^{bad}{\lambda }_j^t+s_{it}^{bad} ,\left(i=1,\dots ,nbad;t=1,\dots ,T\right)$$

$$z_{iot}^{free}={\displaystyle \sum }_{j=1}^n{z}_{ijt}^{free}{\lambda }_j^t+s_{it}^{free} ,\left(i=1,\dots ,free;t=1,\dots ,T\right)$$

$$z_{iot}^{fix}={\displaystyle \sum }_{j=1}^n{z}_{ijt}^{fix}{\lambda }_j^t ,\left(i=1,\dots ,nfix;t=1,\dots ,T\right)$$

$${\sum }_{j=1}^n{\lambda }_j^t=1\left(t=1,\dots ,T\right)$$

(3)

$${\lambda }_j^t\ge 0,s_{it}^{-}\ge 0,s_{it}^{+}\ge 0,s_{it}^{good}\ge 0,s_{it}^{bad}\ge 0 and s_{it}^{free}:free\left(\forall i,t\right)$$

$$s_{it}^{-}, s_{it}^{+}, s_{it}^{good}, s_{it}^{bad}, s_{it}^{free}$$

They represent excessive input, insufficient output, insufficient linkage, excessive linkage, and linkage gap. In three directions: input-oriented, output-oriented, and non-oriented, $\left(\left\{{\lambda }^t\right\},\left\{s_t^{-}\right\},\left\{s_t^{+}\right\},\left\{s_t^{good}\right\},\left\{s_t^{bad}\right\},\left\{s_t^{free}\right\},\left\{s_t^{fit}\right\}\right)$.

Assessment DMUo (o = 1,…,n), and this paper uses the unguided model, which is explained as follows: Non-oriented combination of input-oriented and output-oriented, overall efficiency value ${\theta }_o^{*}$:

$${\theta }_o^{*}=min\frac{\frac{1}{T}{\sum }_{t=1}^T{w}^t\left[1-\frac{1}{f+nbad}\left({\sum }_{i=1}^f\frac{w_i^{-}{s}_{it}^{-}}{x_{iot}}+{\sum }_{t=1}^{nbad}\frac{s_{it}^{bad}}{z_{iot}^{bad}}\right)\right]}{\frac{1}{T}{\sum }_{t=1}^T{w}^t\left[1+\frac{1}{s+ngood}\left({\sum }_{i=1}^s\frac{w_i^{+}{s}_{iot}^{+}}{y_{iot}}+{\sum }_{i=1}^{ngood}\frac{s_{iot}^{good}}{z_{iot}^{good}}\right)\right]}$$

(4)

When the difference is 0, the overall efficiency value is 1. $\left\{{\lambda }_o^{t^{*}}\right\}, \left\{s_{ot}^{{-}^{*}}\right\}, \left\{s_{ot}^{{+}^{*}}\right\}, \left\{s_{ot}^{goo{d}^{*}}\right\}, \left\{s_{ot}^{ba{d}^{*}}\right\},\left\{s_{ot}^{fre{e}^{*}}\right\},\left\{s_{ot}^{fi{x}^{*}}\right\}bring in \left(2-4\right)\theta {\rho }_{ot}$:

$${\theta }_{ot}=\frac{1-\frac{1}{f+nbad}\left({\sum }_{i=1}^m\frac{w_i^{-}{s}_{it}^{-*}}{x_{iot}}+{\sum }_{t=1}^{nbad}\frac{s_{it}^{bad*}}{z_{iot}^{bad}}\right)}{1+\frac{1}{s+ngood}\left({\sum }_{i=1}^s\frac{w_i^{+}{s}_{iot}^{+*}}{y_{iot}}+{\sum }_{i=1}^{ngood}\frac{s_{iot}^{good*}}{z_{iot}^{good}}\right)}$$

(5)

This study used DSBM’s non-oriented variable return to scale model. The input variables were the agricultural labor force, the total power of agricultural machinery, rural electricity consumption, pesticides, and fertilizers. The agricultural GDP was used as an output variable, and the cultivated land area was used as a carry-over (fixed) variable. Due to the inflexibility of the short-term adjustment of the cultivated land area, the cultivated land area was set as a fixed inter-period variable in this study. This model can measure the annual and overall efficiency of each DMU at the same time and can provide differential variable adjustments for DMUs with invalid rates. At the same time, it adjusts input and output variables and measures efficiency more objectively than a traditional DEA. It also provides important information such as the excessive input or insufficient output of an ineffective DMU, with suggestions for an adjustment range. This study used 30 DMUs (j = 1,…,30) through T terms (t = 1,… 5). Each phase of the DMUs had the same 4 inputs (i = 1,…4) and 1 output item (i = 1) The values in phase T were

$$x_{ijt}(i=1,\dots ,4), y_{ijt}(i=1),\mathit{Carry}-\mathit{over}\mathit{was}{z}^{fix}$$

The structure of this study is shown in Figure 2. The undirected mathematical formula is shown in Formula (5) and the restricted formula is shown in Formula (7).

$${\theta }_0^{*}=min\frac{\frac{1}{T}{\sum }_{t=1}^T{W}^t\left[1-\frac{1}{f}\left({\sum }_{i=1}^f\frac{W_i^{-}{S}_{it}^{-}}{X_{iot}}\right)\right]}{\frac{1}{T}{\sum }_{t=1}^T{W}^t\left[1+\frac{1}{S}\left({\sum }_{i=1}^s\frac{W_i^{+}{S}_{it}^{+}}{Y_{iot}}\right)\right]}$$

(6)

In order to ensure the connectivity of the links from period t to t + 1, the following assumptions must be met:

$${\sum }_{j=1}^n{z}_{ijt}^{\beta }{\lambda }_j^t={\sum }_{j=1}^n{z}_{ijt}^{\beta }{\lambda }_j^{t+1},\left(\forall i; t=1,\dots ,T-1\right)$$

(7)

We can express DMUo (o = 1,…,n) as Formula (8)

$$x_{iot}={\displaystyle \sum }_{j=1}^n{x}_{ijt}{\lambda }_j^t+s_{it}^{-},\left(i=1,\dots ,f;t=1,\dots ,T\right)$$

$$y_{iot}={\sum }_{j=1}^n{y}_{ijt}{\lambda }_j^t-s_{it}^{+},\left(i=1,\dots ,s;t=1,\dots ,T\right)$$

(8)

$$z_{iot}^{fix}={\displaystyle \sum }_{j=1}^n{z}_{ijt}^{fix}{\lambda }_j^t,\left(i=1,\dots ,nfix;t=1,\dots ,T\right)$$

$${\displaystyle \sum }_{j=1}^n{\lambda }_j^t=1,\left(t=1,\dots ..,T\right)$$

$${\lambda }_j^t\ge 0, s_{it}^{-}\ge 0, s_{it}^{+}\ge 0 \mathrm{among}{s}_{it}^{-} and s_{it}^{+}$$

The difference variables indicate are excessive input and insufficient output, respectively.

Use the best solution $\left\{{\lambda }_o^{t^{*}}\right\}$, $\left\{s_{ot}^{{-}^{*}}\right\}$, $\left\{s_{ot}^{{+}^{*}}\right\}$, $\left\{s_{ot}^{fi{x}^{*}}\right\}$ are brought into Formula (6).

The most effective efficiency value without guidance can be defined as (8)

$${\rho }_{ot}=\frac{1-\frac{1}{m}\left({\sum }_{i=1}^f\frac{W_i^{-}{S}_{it}^{-}}{X_{iot}}\right)}{1+\frac{1}{S}\left({\sum }_{i=1}^s\frac{W_i^{+}{S}_{it}^{+}}{Y_{iot}}\right)}\left(t=1,\dots ,T\right)$$

(9)

The unguided overall efficiency is defined as follows:

If all the best solutions of Formula (6) are satisfied ${\rho }_{ot}^{*}=1,DM{U}_o$ That is, the overall efficiency without guidance, $s_{iot}^{{-}^{*}}=0\left(\forall i,t\right),s_{iot}^{{+}^{*}}=0\left(\forall i,t\right), then {\rho }_o^{*}=1\left(\forall t\right)$.

The definition of the current effective efficiency without guidance is as follows:

If (6) all the best solutions are satisfied ${\rho }_{ot}^{*}=1 ,DM{U}_o$ That is, the efficiency of the current period without guidance, This means that the best variance variable in (6) is 0 in the current period, $s_{iot}^{{-}^{*}}=0\left(\forall i,t\right),s_{iot}^{{+}^{*}}=0\left(\forall i,t\right)$ (6). The basic model architecture of DSBM in this study is shown in Figure 2:

2.2. TFEE

Total-factor energy efficiency (TFEE) is based on DEA and an input-oriented Charnes, Cooper, and Rhodes (CCR) model. TFEE stands for “target energy input”, which means the actual minimum energy input standard of a region’s economic production. In order to achieve the best efficiency in energy consumption, the smallest gap between the target energy input and the actual energy input, obtain the total adjustment of energy consumption. The greater the total adjustment, the lower the energy consumption efficiency of the region. If the total adjustment of the energy input is equal to zero, it means that the region has reached the target energy input level and has the best energy consumption efficiency. TFEE is based on the view of total-factor productivity. The energy efficiency of a region is defined by Formula (10); that is, at time t, the energy efficiency of region i is the total-factor energy efficiency of the region (TFEE):

$$\mathrm{TFEE}\left(i,t\right)=\frac{Target energy input\left(i,t\right)}{Actual energy input\left(i,t\right)}$$

(10)

As shown in Formula (10), the TFEE index represents the energy consumption efficiency level of a region. Therefore, the TFEE index is between 0 and 1. When the actual energy input level of a DMU is equal to the target energy input level, the TFEE value is 1. If the actual energy input level is far from the target energy input level, the TFEE value is close to 0, which indicates that the efficiency of energy consumption is low. The total-factor agricultural efficiency (TFAE) used in this study is based on the Hu and Wang (2006) [28] total-factor energy efficiency (TFEE) model. According to Formula (10), the input variables of this study (the agricultural labor force, the total power of agricultural machinery, rural electricity consumption, pesticides, and fertilizers) and the output variable (the agricultural GDP) were revised to the TFAE. Because the intertemporal variable (the cultivated land area) is not easy to change in the short term and is an uncontrollable variable, it was set as a carry-over (fixed) in this study, so its efficiency was not calculated. The definitions are as follows:

$$\mathrm{Agricultural} \mathrm{labor} \mathrm{force} \mathrm{TFAE}=\frac{Target value of agricultural labor force}{Actual value of agricultural labor force}$$

(11)

$$\mathrm{Total} \mathrm{power} \mathrm{of} \mathrm{agricultural} \mathrm{machinery} \mathrm{TFAE}=\frac{Target value of total power of agricultural machinery}{Actual value of total power of agricultural machinery}$$

(12)

$$\mathrm{Rural} \mathrm{electricity} \mathrm{consumption} \mathrm{TFAE}=\frac{Target value of rural electricity consumption}{Actual value of rural electricity consumption}$$

(13)

$$\mathrm{Pesticide} \mathrm{and} \mathrm{fertilizer} \mathrm{TFAE}=\frac{Target value of pesticides and fertilizers}{Actual value of pesticides and fertilizers}$$

(14)

$$\mathrm{Agricultural} \mathrm{GDP} \mathrm{TFAE}=\frac{Real value of agricultural GDP}{Target value of agricultural GDP}$$

(15)

In this study, the DSBM and TFAE models were applied. Agricultural labor, total agricultural machinery power, rural electricity consumption, and pesticide and fertilizer usage were used as input variables, while the agricultural GDP was used as the output variable. The cultivated area was set as a carry-over (fixed) variable. The variables and their definitions are explained in

Table 1. Definitions of input and output variables.

Variable		Unit	Definition
Input items	Agricultural labor force	10,000 people	Refers to the labor force (aged 16 years and over) invested in agricultural production.
	Total power of agricultural machinery	10,000 kW	The sum of the rated power of all agricultural machinery.
	Rural electricity consumption	100 million kWh	Rural electricity directly used for rural economic and social development.
	Pesticide and fertilizer	10,000 tons	The amount of pesticides and fertilizers used in agricultural production.
Output item	Agricultural GDP	CNY 100 million	The total value of all agricultural products and agricultural production activities.
Carry-over	Cultivated land area	1000 hectares	The farmland that can be used to plant crops and frequently hoe is an uncontrollable variable because it is difficult to change the cultivated area in a short period of time. Therefore, the cultivated area in this study is in the form of a carry-over (fixed).

Source: China Statistical Yearbook (2020) [29], China National Bureau of Statistics.

.

3. Results and Discussion

3.1. Research Data Description

In order to evaluate the total-factor efficiency of the agricultural production in various regions of China, 30 administrative regions of China were taken as the research subjects and were divided into three regions according to their economic development. Regarding the input and output variables used in this paper, please refer to Table 1 for details. Descriptive statistical information is provided in

Table 2. Descriptive statistics of input and output variables from 2012 to 2016.

	Variable	Average	Max.	Min.	Standard Deviation
Input	Agricultural labor force (10,000 people)	921	2652	45	645
	Total power of agricultural machinery (10,000 kW)	3471.25	13,353.02	112.73	3056.08
	Rural electricity consumption (100 million kWh)	288.01	1869.27	4.46	288.01
	Pesticide and fertilizer (10,000 tons)	204.13	728.96	8.95	204.13
Output	Agricultural GDP (CNY 100 million)	1701.41	4662.61	117.09	1701.41
Carry -over	Cultivated land area (1000 hectares)	5486.41	14,879.73	120.94	3894.37

Source: DEA Solver 13.

.

Agricultural labor force: The average value was 9.21 million people, with a maximum value of 26.52 million people in Henan Province in 2014 and a minimum value of 450,000 people in Shanghai in 2014. The standard deviation was 6.45 million people (Table 2). Total power of agricultural machinery: The average value was 34.7125 million kilowatts, with a maximum value of 133,530.02 million kilowatts in Shandong Province in 2015 and a minimum value of 112.73 million kilowatts in Shanghai in 2012. The standard deviation was 30.5608 million kilowatts, indicating significant differences in total agricultural machinery power across different administrative regions (Table 2). Rural electricity consumption: The average value was 28.801 billion kilowatt-hours, with a maximum value of 186.927 billion kilowatt-hours in Jiangsu Province in 2016 and a minimum value of 4.46 billion kilowatt-hours in Qinghai Province in 2012. The standard deviation was 28.801 billion kilowatt-hours (Table 2). Pesticide and fertilizer: The average value was 2.0413 million tons, with a maximum value of 7.2896 million tons in Henan Province in 2015 and a minimum value of 0.0895 million tons in Qinghai Province in 2016. The standard deviation was 2.0413 million tons (Table 2). Agricultural GDP: The average value was CNY 170.141 billion, with a maximum value of CNY 466.261 billion in Shandong Province in 2015 and a minimum value of CNY 11.709 billion in Qinghai Province in 2012. The standard deviation was CNY 170.141 billion (Table 2). Cultivated land area: The average value was 5486.41 thousand hectares, with a maximum value of 14879.73 thousand hectares in Henan Province and a minimum value of 120.94 thousand hectares in Beijing. The standard deviation was 3894.37 thousand hectares (Table 2).

3.2. Empirical Results

This section first discusses the agricultural production efficiency and overall efficiency of 30 administrative regions in China. Then, the 30 administrative areas are divided into three regions (east, middle, and west), and the overall efficiency of each region is analyzed. An adjustment and range analysis of the difference variables are carried out for the administrative regions with invalid rates. Finally, the total-factor agricultural production efficiency is discussed and analyzed.

3.2.1. Overall Efficiency of Agricultural Production in 30 Administrative Regions of China (Table 3 and Figure 3)

(1) Efficiency of 30 administrative regions in China: The overall efficiency of agricultural production in China’s 30 administrative regions was 0.8978 from 2012 to 2016. The overall efficiency of 14 administrative regions, including Beijing, Heilongjiang, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, Guangxi, Hainan, Chongqing, Sichuan, Qinghai, and Xinjiang, was 1. These regions belonged to an efficient decision-making unit. The administrative regions that were more efficient than the average overall efficiency were Henan, Hubei, Shaanxi, Ningxia, Guizhou, and Tianjin. There were 10 administrative regions, such as Liaoning, Jiangxi, Hunan, Inner Mongolia, Hebei, Yunnan, Anhui, Jilin, Gansu, and Shanxi, that were less efficient than the average overall efficiency. The overall efficiency value of 0.6311 in Shanxi was the lowest. It is worth noting that the efficiency values of Jilin in 2012–2016 were 0.6263, 1, 1, 0.5166, and 0.4429, respectively, which fell sharply. Anhui’s efficiency values in 2012–2016 were 0.4809, 0.9999, 0.4489, 0.5842, and 1, respectively, with ups and downs.

(2) As previous studies have indicated, there are regional differences in agricultural production efficiency across different regions of China [14]. Specifically, based on data from 2012 to 2016, the average overall efficiency of agricultural production in all regions of China was 0.8978. The average overall efficiency of the eastern region was 0.9587, while the average overall efficiency of the western region was 0.9020, both higher than the average overall efficiency of the 30 administrative regions. The average overall efficiency of the central region was 0.8123, lower than the average overall efficiency of the 30 administrative regions. Eastern region: Beijing, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, Hainan, and Guangxi were all efficient administrative regions with efficiency values of 1. Tianjin, Hebei, Liaoning, and other administrative regions were lower than the average overall efficiency of the eastern region, of which Liaoning was the lowest. Central region: Only Heilongjiang, with an efficiency value of 1, was an efficient administrative region. The efficiency values of Henan and Hubei were higher than the average overall efficiency of the central region. The efficiency values of Jiangxi, Hunan, Inner Mongolia, Anhui, Jilin, Shanxi, and other administrative regions were lower than the average value of the overall efficiency of the central region, and Shanxi was the administrative region with the lowest efficiency value. Western region: Chongqing, Sichuan, Qinghai, and Xinjiang were efficient administrative regions with efficiency values of 1. The efficiency values of Shaanxi, Ningxia, and Guizhou were higher than the average overall efficiency of the western region. However, the efficiency values of Yunnan and Gansu were lower than the average overall efficiency of the western region, and Gansu was the administrative region with the lowest efficiency value.

Based on the above agricultural production efficiency values of all regions in China, the agricultural production efficiency values in the eastern region showed an upward trend, the agricultural production efficiency values in the central region showed an upward trend, and the agricultural production efficiency values in the western region showed an upward trend from 2012 to 2016. Among them, the agricultural production efficiency in the central region was unstable during the five-year period, while the agricultural production efficiency in the western region increased the most from 2012 to 2016. The magnitude reached up to 15.35%. For the total-factor productivity efficiency values of the agricultural production in the 30 administrative regions of China, please refer to Figure 3.

Table 3. Annual average efficiency and overall efficiency of agricultural production in various regions of China from 2012 to 2016.

	DMU	2012	2013	2014	2015	2016	Overall
Eastern Region	Beijing	1	1	1	1	1	1
	Tianjin	0.6273	1	1	1	1	0.9089
	Hebei	0.7256	0.9998	0.6558	0.9998	0.5451	0.7809
	Liaoning	0.6686	0.7129	0.6909	1	1	0.8145
	Shanghai	1	1	1	1	1	1
	Jiangsu	1	1	1	1	1	1
	Zhejiang	1	1	1	1	1	1
	Fujian	1	1	1	1	1	1
	Shandong	0.9999	1	1	1	1	1
	Guangdong	1	1	1	1	1	1
	Hainan	0.9999	1	1	1	1	1
	Guangxi	1	1	1	1	1	1
	Average	0.9184	0.9761	0.9456	1.0000	0.9621	0.9587
Central Region	Shanxi	0.4402	0.9999	0.4965	1	0.4763	0.6311
	Anhui	0.4809	0.9999	0.4489	0.5842	1	0.6841
	Jiangxi	0.4765	0.9999	0.7164	1	1	0.8115
	Henan	0.9999	0.9999	0.9999	0.9999	0.9999	0.9999
	Hubei	0.9999	0.9999	0.8669	0.9999	0.8576	0.9448
	Hunan	0.7916	0.8051	0.6851	1	0.6603	0.7886
	Jilin	0.6263	1	1	0.5166	0.4429	0.6678
	Heilongjiang	1	1	1	1	1	1
	Inner Mongolia	0.6761	0.6207	0.7150	1	1	0.7828
	Average	0.7213	0.9361	0.7699	0.9001	0.8263	0.8123
Western Region	Chongqing	1	1	1	1	1	1
	Sichuan	1	1	1	1	1	1
	Guizhou	0.6117	1	1	1	1	0.9165
	Yunnan	0.5953	0.7486	0.6967	1	0.6147	0.7221
	Shaanxi	0.6248	1	1	1	1	0.9240
	Gansu	0.5336	0.7089	0.6263	0.7637	0.5583	0.6384
	Qinghai	1	1	1	1	1	1
	Ningxia	0.6725	1	1	1	1	0.9169
	Xinjiang	1	1	1	1	1	1
	Average	0.7820	0.9397	0.9248	0.9737	0.9081	0.9020
Overall		0.8184	0.9532	0.8866	0.9621	0.9052	0.8978

Table 3. Annual average efficiency and overall efficiency of agricultural production in various regions of China from 2012 to 2016.

	DMU	2012	2013	2014	2015	2016	Overall
Eastern Region	Beijing	1	1	1	1	1	1
	Tianjin	0.6273	1	1	1	1	0.9089
	Hebei	0.7256	0.9998	0.6558	0.9998	0.5451	0.7809
	Liaoning	0.6686	0.7129	0.6909	1	1	0.8145
	Shanghai	1	1	1	1	1	1
	Jiangsu	1	1	1	1	1	1
	Zhejiang	1	1	1	1	1	1
	Fujian	1	1	1	1	1	1
	Shandong	0.9999	1	1	1	1	1
	Guangdong	1	1	1	1	1	1
	Hainan	0.9999	1	1	1	1	1
	Guangxi	1	1	1	1	1	1
	Average	0.9184	0.9761	0.9456	1.0000	0.9621	0.9587
Central Region	Shanxi	0.4402	0.9999	0.4965	1	0.4763	0.6311
	Anhui	0.4809	0.9999	0.4489	0.5842	1	0.6841
	Jiangxi	0.4765	0.9999	0.7164	1	1	0.8115
	Henan	0.9999	0.9999	0.9999	0.9999	0.9999	0.9999
	Hubei	0.9999	0.9999	0.8669	0.9999	0.8576	0.9448
	Hunan	0.7916	0.8051	0.6851	1	0.6603	0.7886
	Jilin	0.6263	1	1	0.5166	0.4429	0.6678
	Heilongjiang	1	1	1	1	1	1
	Inner Mongolia	0.6761	0.6207	0.7150	1	1	0.7828
	Average	0.7213	0.9361	0.7699	0.9001	0.8263	0.8123
Western Region	Chongqing	1	1	1	1	1	1
	Sichuan	1	1	1	1	1	1
	Guizhou	0.6117	1	1	1	1	0.9165
	Yunnan	0.5953	0.7486	0.6967	1	0.6147	0.7221
	Shaanxi	0.6248	1	1	1	1	0.9240
	Gansu	0.5336	0.7089	0.6263	0.7637	0.5583	0.6384
	Qinghai	1	1	1	1	1	1
	Ningxia	0.6725	1	1	1	1	0.9169
	Xinjiang	1	1	1	1	1	1
	Average	0.7820	0.9397	0.9248	0.9737	0.9081	0.9020
Overall		0.8184	0.9532	0.8866	0.9621	0.9052	0.8978

Figure 3. Overall efficiency map of various administrative regions in China from 2012 to 2016.

Figure 3. Overall efficiency map of various administrative regions in China from 2012 to 2016.

3.2.2. Slack Analysis

Using a slack analysis (

Table 4. Average adjustment ranges of agricultural production variables in various regions of China from 2012 to 2016 (unit: %).

	DMU	Agricultural Labor Force	Total Power of Agricultural Machinery	Rural Electricity Consumption	Pesticide and Fertilizer	Agricultural GDP
Eastern Region	Beijing	0	0	0	0	0
	Tianjin	0	−9.34	−1.38	−4.38	5.86
	Hebei	0	−28.48	−41.80	−11.61	1.84
	Liaoning	−15.46	−16.37	−42.39	0	0
	Guangxi	−0.01	0	0	−0.01	0
	Shanghai	0	0	0	0	0
	Jiangsu	0	0	0	0	0
	Zhejiang	0	0	0	0	0
	Fujian	0	0	0	0	0
	Shandong	0	0	0	0	0
	Guangdong	0	0	−0.01	0	0
	Hainan	0	0	0	0	0
	Average	−1.29	−4.52	−7.13	−1.33	0.64
Central Region	Shanxi	−12.37	−19.68	−33.11	−0.79	32.34
	Anhui	−21.11	−23.68	−25.80	−21.95	12.36
	Jiangxi	−13.05	−11.02	−13.62	−1.99	11.00
	Henan	0	−0.01	−0.01	−0.01	0
	Hubei	−1.22	−4.97	−2.60	−13.27	0
	Hunan	−29.16	−29.66	−13.48	−9.43	0.90
	Jilin	−9.19	−13.32	−8.04	−22.48	29.88
	Heilongjiang	0	0	0	0	0
	Inner Mongolia	−3.85	−9.21	−7.24	−7.39	18.90
	Average	−10.00	−12.39	−11.55	−8.59	11.71
Western Region	Chongqing	0	0	0	0	0
	Sichuan	0	0	0	0	0
	Guizhou	−13.52	−2.34	−10.50	0	1.92
	Yunnan	−47.23	−7.35	−21.14	−11.38	8.32
	Shaanxi	−8.47	−4.84	−10.55	−5.37	0.31
	Gansu	−41.64	−24.22	−28.04	−0.30	19.75
	Qinghai	0	0	0	0	0
	Ningxia	0	−3.41	0	−3.41	7.21
	Xinjiang	0	0	0	0	0
	average	−12.32	−4.69	−7.80	−2.27	4.17
	Average	−7.21	−6.93	−8.66	−3.79	5.02

), we objectively evaluated the efficiency value of each variable and identified effective suggestions to improve the performance of the overall efficiency. The analysis is as follows:

During the study period, the total agricultural labor input was reduced by 7.21% (Table 4). The adjustment range of the agricultural labor input in the central and western regions was greater than the overall average. The adjustment range of the total power input of agricultural machinery needed to be reduced by an average of 6.93%. Only the adjustment range in the central region was greater than the total average, while the adjustment range in the eastern and western regions was less. The overall adjustment range of the rural power consumption input needed to be reduced by an average of 8.66%. Only the adjustment range of the rural power consumption input in the central region was larger than the overall average, while the adjustment range of the eastern and western regions was smaller. The overall adjustment range of the pesticide and fertilizer input needed to be reduced by an average of 3.79%. The annual adjustment range of the central region was higher than that of the eastern and western regions. The adjustment range of the central region was greater than the overall average, and the adjustment range of the eastern and western regions is smaller. The overall adjustment range of the agricultural GDP output needed to increase by an average of 5.02%. The adjustment range of the central region was greater than the overall average. The annual adjustment range of the central region was higher than that of the eastern and western regions. The average adjustment ranges of various variables of agricultural production are shown in Table 4 and Figure 4. Eastern region: at 7.13%, the rural electricity consumption was the variable with the largest adjustment range, and at 0.64%, the agricultural GDP was the variable with the smallest adjustment range. Central region: at 12.39%, the total power of agricultural machinery was the variable with the largest adjustment range, and at 8.59%, pesticide and fertilizer use was the variable with the smallest adjustment range. Western region: at 12.32%, the agricultural labor force was the largest variable, and at 2.27%, pesticide and fertilizer use was the smallest variable. The central region had the largest adjustment ranges for the total power of agricultural machinery, rural electricity consumption, pesticides and fertilizers, agricultural GDP, and other variables. The adjustment ranges of the agricultural labor force, the total power of agricultural machinery, rural electricity consumption, pesticides and fertilizers, agricultural GDP, and other variables in the eastern region were the smallest.

3.2.3. Total-Factor Agricultural Efficiency (TFAE)

The total-factor agricultural efficiency values are shown in

Table 5. TFAE of various regions in China from 2012 to 2016.

	DMU	Agricultural Labor Force	Total Power of Agricultural Machinery	Rural Electricity Consumption	Pesticide and Fertilizer	Agricultural GDP	Average
Eastern Region	Beijing	1	1	1	1	1	1
	Tianjin	1	0.907	0.986	0.956	0.955	0.961
	Hebei	1	0.715	0.582	0.884	0.983	0.833
	Liaoning	0.845	0.836	0.576	1	1	0.852
	Guangxi	1	1	1	1	1	1
	Shanghai	1	1	1	1	1	1
	Jiangsu	1	1	1	1	1	1
	Zhejiang	1	1	1	1	1	1
	Fujian	1	1	1	1	1	1
	Shandong	1	1	1	1	1	1
	Guangdong	1	1	1	1	1	1
	Hainan	1	1	1	1	1	1
	Average	0.987	0.955	0.929	0.987	0.995	0.971
Central Region	Shanxi	0.876	0.803	0.669	0.992	0.797	0.828
	Anhui	0.789	0.763	0.742	0.781	0.901	0.795
	Jiangxi	0.869	0.890	0.864	0.980	0.915	0.904
	Henan	1	1	1	1	1	1
	Hubei	0.988	0.950	0.974	0.867	1.000	0.956
	Hunan	0.708	0.703	0.865	0.906	0.991	0.835
	Jilin	0.908	0.867	0.920	0.775	0.807	0.855
	Heilongjiang	1	1	1	1	1	1
	Inner Mongolia	0.962	0.908	0.928	0.926	0.858	0.916
	Average	0.900	0.876	0.885	0.914	0.919	0.899
Western Region	Chongqing	1	1	1	1	1	1
	Sichuan	1	1	1	1	1	1
	Guizhou	0.865	0.977	0.895	1	0.982	0.944
	Yunnan	0.528	0.926	0.789	0.886	0.935	0.813
	Shaanxi	0.915	0.952	0.894	0.946	0.997	0.941
	Gansu	0.584	0.758	0.720	0.997	0.836	0.779
	Qinghai	1	1	1	1	1	1
	Ningxia	1	0.966	1	0.966	0.947	0.976
	Xinjiang	1	1	1	1	1	1
	Average	0.877	0.953	0.922	0.977	0.966	0.939
Average		0.928	0.931	0.913	0.962	0.963	0.940

and Figure 5.

From 2012 to 2016, the average total-factor agricultural efficiency value of the 30 administrative regions in China was 0.940. The average total-factor agricultural efficiency of 14 administrative regions, including Beijing, Guangxi, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, Hainan, Heilongjiang, Chongqing, Sichuan, Qinghai, and Xinjiang, was 1, indicating efficient administrative districts. Inner Mongolia, Jiangxi, Jilin, Liaoning, Hunan, Hebei, Shanxi, Yunnan, Anhui, Gansu, and the other 10 administrative regions were all less efficient than the average total-factor agricultural efficiency. The average total-factor agricultural efficiency of Anhui was 0.7951, and the average total-factor agricultural efficiency value of Gansu was 0.7787. These were the two administrative regions with the lowest efficiency. The factor efficiency of all agricultural production inputs and outputs was analyzed in China’s 30 administrative regions from 2012 to 2016. The factor efficiency of the pesticide and fertilizer input and the agricultural GDP output was the best. The efficiency of input factors such as the agricultural labor force, the total power of agricultural machinery, and rural electricity consumption was lower than the average total-factor agricultural efficiency value. Among them, the rural power consumption efficiency was the worst and should be improved first. According to the analysis of the total-factor agricultural efficiency values in various regions of China from 2012 to 2016, only the average total-factor agricultural efficiency values in the eastern region were higher than the total average, while the average total-factor agricultural efficiency values in the central and western regions were lower than the total average. Among them, the agricultural labor force, the total power of agricultural machinery, rural electricity consumption, pesticides and fertilizers, and the agricultural GDP in the central region were lower than the total average. This was the most inefficient region.

3.3. China’s Agricultural Production Efficiency

Due to the food production crisis brought by climate change, all countries need to face the problem of food shortage in the future. This study summarizes the trends of the input and output variables of agricultural production (

Table 6. Trends of variables in China’s 30 administrative divisions from 2012 to 2016.

	2012	2013	2014	2015	2016	(%)	Ave.
Agricultural labor force (ten thousand people)	951.9	933.46	920.4	904.85	894.89	−5.99	921.1
Total power of agricultural machinery (ten thousand kilowatts)	3403.13	3446.32	3582.86	3703.61	3220.35	−5.37	3471.25
Rural electricity consumption (one hundred million kilowatt-hours)	250.25	284.95	296.11	300.86	307.9	23.03	288.01
Pesticides and fertilizers (ten thousand tons)	200.48	202.89	205.69	206.49	205.08	2.29	204.13
Agricultural GDP (one hundred million CNY)	1493.08	1629.53	1726.26	1804.58	1853.59	24.13	1701.41
Cultivated area (one thousand hectares)	5395.03	5449.22	5498.54	5553.32	5555.86	2.98	5490.39

) as follows:

During the study period, the agricultural production variables of rural electricity consumption, pesticides and fertilizers, cultivated land area, and the agricultural GDP showed upward trends, while the agricultural labor force and the total power of agricultural machinery showed downward trends. From 2012 to 2016, China’s cultivated land area increased by 2.98%, its agricultural GDP increased by 24.13%, and its agricultural production efficiency increased by 9.82%. Due to decreases in the agricultural labor input and the agricultural machinery and equipment input, agricultural production costs were reduced, the agricultural gross domestic product increased, and the agricultural production efficiency improved, which was in line with the expected trends of this study. It is worth noting that during the study period, the input of rural electricity consumption increased by 23.03% and the input of pesticides and fertilizers increased by 2.29%. This shows that the relevant departments invested considerable resources in rural economic development, which affected the performance of the overall efficiency of agricultural production. According to the TFAE model, the average value of agricultural labor efficiency was 0.928, the average value of the total power of agricultural machinery was 0.931, the average value of rural electricity consumption was 0.913, the average value of pesticides and fertilizers was 0.962, and the average agricultural GDP was 0.963. The average value of rural electricity consumption was the lowest, and its efficiency was the worst among all input variables. Rural electricity consumption had an important impact on China’s agricultural production efficiency and needed to be improved first. According to the analysis of the adjustment ranges of the input variables in various regions, the largest adjustment range of rural electricity consumption was in the eastern region. Among the 12 administrative regions in the eastern region, 8 administrative regions, including Beijing, Guangxi, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, and Hainan, did not need to be adjusted. Tianjin was adjusted by −1.38%; Guangdong was adjusted by −0.01%; and the Hebei and Liaoning administrative regions had the largest adjustment ranges at −41.80% and −42.39%, respectively.

The foundation of this study is based on the “World Food Security: Rome Declaration” and the “World Food Summit Plan of Action” proposed by the Food and Agriculture Organization of the United Nations and various governments in 1996. Building on this foundation, we conducted empirical research and analyzed relevant research results. The following is a summary of our findings.

(1) These empirical results are in line with the strategic goal of expanding planting areas and increasing food production, as mentioned in China’s “Outline of Medium and Long-term Plan for Food Security” issued in 2008. The cultivated land area in China should continue to grow. (2) From 2012 to 2016, the overall efficiency of the agricultural production in various regions of China showed an upward trend. The eastern region had the highest overall efficiency of agricultural production, while the central region had the lowest [14]. In order to promote agricultural development in China’s central and western regions, we need to take into account the differences in resources, environment, human resources, and other conditions across different regions, despite the economic development in the eastern region. (3) Our analysis of the DSBM difference variables reveals that the factors affecting agricultural production efficiency vary across the different regions of China, and appropriate resource allocation needs to occur based on the needs of each region [15,16]. For instance, the eastern region requires the most adjustment in terms of rural electricity consumption, with the Hebei and Liaoning administrative regions needing the most adjustment. Meanwhile, the central region needs to adjust the total power of agricultural machinery, with Anhui and Hunan requiring the most adjustment, and the western region needs to improve its agricultural labor force, particularly in Yunnan and Gansu. (4) The TFAE analysis shows that the efficiency of agricultural production factors varies [20], with the efficiency of the pesticide and fertilizer input and the agricultural GDP output being the highest. Input factors such as the agricultural labor force, the total power of agricultural machinery, and rural electricity consumption are less efficient than the average agricultural total-factor efficiency. Priority should be given to improving the efficiency of rural electricity consumption, which is the least efficient of all input factors. (5) Our empirical results demonstrate positive trends in China’s agricultural GDP growth and the overall efficiency of agricultural production, indicating that China’s agricultural production is heading in the right direction. By adjusting resources according to each region’s needs and effectively allocating them, we can improve the overall efficiency of agricultural production.

4. Conclusions

From 2012 to 2016, studies on agricultural production efficiency in China showed that the eastern regions performed the best while the central regions performed the worst. Rural electrification was identified as the input variable most in need of improvement, particularly in Hebei and Liaoning provinces. Although overall agricultural production efficiency and agricultural GDP showed an increasing trend, there were still other related agricultural input variables that needed to be adjusted to achieve maximum resource allocation and production efficiency.

Based on empirical analysis, this study emphasizes the interrelationship between agricultural production resource allocation and regional development, as well as the importance of agricultural economic development policies. These findings will provide substantial reference for the formulation and implementation of future agricultural policies, contributing to ensuring food security and sustainable development. These results offer valuable insights in related fields and contribute new perspectives to the study of agricultural production resource allocation and regional development.