Impact of Non-Agricultural Employment on Food Security in China’s Old Revolutionary Base Areas

Abstract

With the growing trend of arable land abandonment, the potential threat to the security of the food supply has sparked public concern. In order to examine the impact of non-agricultural employment on food security, this study builds linear regression models for research based on panel data from counties in China’s old revolutionary base areas. The empirical results show that, although the impact of non-agricultural employment on total grain production is not significant, it has a significant negative impact on both area and productivity, which indicates that non-agricultural employment poses a challenge to food security. In addition, the study examines the potential benefits of non-farm employment on two aspects of food security, including intensive management and the increase of new business entities. Non-farm employment can also significantly promote intensive management, thereby reducing the food-security challenges brought by non-farm employment, while the benefits of new management entities are insignificant. These findings contribute to the optimization of economic policies related to agricultural development, including exploring land property rights reform systems to promote land transfer, strengthening labor quality improvement in the agricultural sector, and formulating supporting policies to stabilize non-agricultural employment in accordance with local conditions.

1. Introduction

Food security has received global attention as a key strategy for reducing hunger, improving nutrition, and promoting long-term economic growth [1]. It is closely related to the aspect emphasized in the United Nations Sustainable Development Goal 2 (SDG2), namely, achieving food security for humanity [2]. The fulfillment of food security is one of the important matters for the realization of the SDGs. According to the Food and Agriculture Organization of the United Nations (FAO), food security exists when all people, at all times, have physical, social, and economic access to sufficient, safe, and nutritious food to meet their dietary needs and food preferences for an active and healthy life [3]. However, it is estimated that over 820 million people worldwide go hungry and around two billion people experience food insecurity each year [4]. At the global level, natural and socio-economic factors, such as natural disasters, climate change, and geopolitical conflicts, threaten food security [5]. One of the important factors, non-agricultural employment, has been emphasized by researchers, since it has a significant and negative impact on farmland output [6]. In addition, the transfer of rural labor to non-agricultural sectors is inevitable in the process of economic takeoff in developing countries or regions.

Non-agricultural employment, or what is known as off-farm work, is any activity undertaken by a farmer or farm household outside of agriculture as a source of income [7] and is a phenomenon that exists around the world [8]. Since the early 1990s, the importance of rural non-agricultural employment in developing countries has been increasingly recognized [9]. A large proportion of farmers in many countries are engaged in non-agricultural employment as an additional source of household income with a key function in rural and agricultural development [10]. The share of rural household income that stems from non-agricultural sources ranges from 35% in Asia to 40% in Latin America and 45% in Sub-Saharan Africa [11]. From the perspective of food security, the allocation of rural labor between the agricultural and non-agricultural sectors has far-reaching implications for rural development and agricultural production [12].

Especially in developing countries, non-farm employment plays an increasingly important role in sustainable rural and agricultural development [13]. The migration of a large number of highly qualified laborers has made women and the elderly the main force in rural farming production. In this process, how to fill the vacancies in the labor force caused by migration has become an urgent and critical issue to be resolved [6]. As China’s economy has accelerated in recent years [14], non-agricultural employment has assumed an increasingly important role in the country’s socio-economic structure. The share of non-agricultural employment in China has risen from about 50.00% in 2000 to 75.92% in 2022 [15]. During this historical process, China’s food production has been growing in the new century and the threat of food security has been diminished compared to the end of the last century [16], but structural contradictions in food production still exist. Urban sprawl covers fertile soil and makes food production more difficult [17]. The weakening of farmers and the diversion of arable land remain potential threats to food security [18]. It is worth noting that the shift of a large number of rural laborers to the non-farm sector has led to a shortage of human resources for food production, resulting in arable land abandonment [19] and a reduction in the utility of arable land and food-production capacity [20]. Against the backdrop of intensifying contradictions between global food supply and demand, the weight of China’s food security has become increasingly prominent [21]. Balancing non-agricultural employment development and food security while maintaining sustainable economic growth has emerged as an important issue. As a developing country, China suffers from uneven regional development [22], which means that there are still relatively underdeveloped areas in China. As China’s historical revolutionary strongholds, the old revolutionary base areas are almost underdeveloped areas, with relatively poor natural conditions, remote geographic locations, and significantly lagging behind in growth [23]. The problems of food production, population employment, and industrial development in these economically underdeveloped regions are intertwined and complicated. The old revolutionary base areas confront unique challenges and opportunities.

The relationship between non-agricultural employment and food security has been widely documented. Some studies have argued that the outflow of the agricultural population is intrinsically linked to the demographic transition and the land-use transition [24] and that the population-mobility transition is closely associated with the land-use transition [25]. Although the de-cultivation of agricultural land is the consequence of a combination of multiple drivers, the loss of labor in the agricultural sector is the direct reason for it [26]. Regarding utilization efficiency, the non-agricultural employment of China’s rural labor force has a U-shaped effect on the utilization efficiency of agricultural land, i.e., the utilization efficiency of agricultural land decreases and then increases as the proportion of non-agricultural employment increases [27]. Remarkably, the loss of labor due to non-agricultural employment produces two effects in completely opposite directions [28], namely, a negative lost-labor effect and a positive income or liquidity relaxing effect, or what are known as the complementarity and substitution effects [29]. The non-agricultural sector is widely recognized as negatively related to food security, e.g., mining, which may stimulate the local economy but jeopardizes food security [30]. Conversely, studies in specific regions have found that non-agricultural employment has an insignificant effect on total food-crop production and that the positive effects outweigh the negative effects [29]. Non-agricultural employment also has a positive effect [31]. For instance, those who stay in the countryside can use the remittances of the migrant workers to their families to invest in agricultural production operations, buy agricultural machinery or fertilizers to increase productivity [32], and perform the substitution role of the capital factor for the labor factor [33], thus reshaping the input–output relationship in food production. The micro-household perspective has also been taken into consideration, suggesting that non-agricultural employment is an essential way to provide improved food consumption for smallholder households [34], which can contribute to household nutrition [35] and play an important role in the reduction of malnutrition and the conservation of agro-biodiversity [36].

In summarizing the discussion of non-agricultural employment on food security in the extant research, it is obvious that most of the researchers spotlight exploring the implications of non-agricultural employment on farmland, including its use type and its efficiency. However, the impact of non-agricultural employment on food security is also characterized by its effect on the production of food crops per unit area, an indicator that largely reflects local food-production capacity. The existing literature relies on provincial or even national data and has not yet systematically examined the dual positive and negative impacts of non-agricultural employment on food security. Moreover, previous studies rarely focus on developing regions, especially the economically and socially backward ones, while the old revolutionary base areas are strongly representative of China’s underdeveloped economic regions and are typical of the developing regions. Taking the old revolutionary base areas as the main research object, this paper aims to analyze the impact of non-agricultural employment on food security in China’s old revolutionary base areas and the mechanism of related factors. This study is dedicated to bridging the research gap between employment structure and food security in the old revolutionary base areas, which helps to explore the special characteristics of this region and, thus, provides theoretical support and policy recommendations for local economic transformation and agricultural development. In addition, the typicality of China’s old revolutionary base areas as economically backward areas allows this research to offer lessons for other developing regions or even countries.

The rest of the paper is structured as follows: Section 2 provides a theoretical analysis that introduces the research hypothesis; Section 3 describes the design of the study, including a detailed description of the models, variables, and data sources used; Section 4 deals with the results and discussion and contains the results of the benchmark regression in Section 4.1, robustness tests in Section 4.2, a further discussion in Section 4.3, and the analysis of heterogeneity in Section 4.4; and Section 5 presents the final remarks, including conclusions and recommendations.

2. Background and Theoretical Analysis Hypotheses

2.1. Background of China’s Old Revolutionary Base Area

China’s old revolutionary base areas saw history as regions created under the leadership of the Communist Party of China (CPC) during the Agrarian Revolutionary War and the War of Resistance Against Japanese Aggression, spreading across 28 provinces in China. In China, the old revolutionary base areas are regarded as typical economically and socially backward areas. In response to the imbalance in regional development, China has implemented the Strategy for Revitalization and Development of Old Revolutionary Base Areas, which covers a series of government interventions and strategic planning, such as a series of supportive policies, including financial support and tax incentives, counterpart support, construction of industrial parks, science, technology and education, human resources, land, and ecological compensation, among others [22].

The old revolutionary base areas are rich in agricultural resources, and the industrialization process is relatively slow. Agriculture based on food production is the main economic component. Due to the influence of public policies to support the development of the old revolutionary base areas, rural residents in these areas have more job opportunities, and many rural residents have turned to non-agricultural employment. Agricultural endowment resources determine that the old revolutionary base areas are important areas for food production, and the change in employment structure provides a typical case for investigating how non-agricultural employment affects food security.

2.2. Theoretical Analysis and Hypothesis

This study proposes the theoretical mechanism and hypothesis of non-agricultural employment affecting food security, as Figure 1 shows. Non-agricultural employment may cause challenges or benefits to food security [14]. The negative challenges include the decline in the quantity and quality of labor, and the benefits include intensive management and the cultivation of agricultural entrepreneurship.

An increase in the share of non-agricultural employment implies an outflow of labor from the agricultural sector to the non-agricultural sector, resulting in a relative decrease in the supply of labor in the agricultural sector. The labor force in rural areas is the key to food production [37]. A decrease in the supply of labor can negatively affect the efficiency of food production and total grain production. Significantly, the negative impact of non-agricultural employment on labor resources is not only in quantity but also in quality. The outflow of labor is mainly young and middle-aged men, who tend to have a relatively higher level of education, are more easily combined with agricultural machinery and other means of production, and, therefore, have a stronger labor production capacity. Most of the labor force remaining in the agricultural sector has low labor quality and low productivity. The outflow of high-quality labor may also mean a reduction in factor inputs to the agricultural sector. After they are engaged in non-agricultural sector production, the investment in arable land and the investment in agricultural production capital may also be reduced accordingly [38]. All of the above factors may reduce total grain production, thus affecting food security. Based on the above findings, Hypothesis 1 is proposed in this study.

Hypothesis 1a:

The increase in the share of non-agricultural employment reduces food security by cutting down the total grain production.

Non-agricultural employment has a direct impact not only on the input of labor and capital factors in the agricultural sector but also on the amount of arable land being used. The shortage of labor resources and capital in agriculture, especially in the food-crop cultivation sector, has made it impossible to maintain the original scale of cultivated land area. Although land capability is also one of the causes, external drivers such as labor shortage play the most important role in arable land abandonment [39]. On the other hand, tens of millions of non-agricultural employees have provided sufficient human resources for the secondary and tertiary industries, which has led to the sustained economic development of the non-agricultural sector. This trend has been particularly significant during China’s industrial transformation since the Reform and opening up. In terms of land use, the expansion of the non-agricultural sector is mainly reflected in the expansion of industrial land use and the shrinking of agricultural land use, especially of arable land. This analysis is consistent with the actual history; the scale of agricultural land abandonment in China increased significantly after 2000. This trend became even more pronounced after 2010 and coincided with the process of labor outflow [40]. Nonetheless, maintaining a certain quantity and quality of arable land plays a crucial role in guaranteeing food supply [41]. The decrease in arable land can pose a hidden danger to national food security [42]. Accordingly, Hypothesis 1b can be proposed.

Hypothesis 1b:

The increase in the share of non-agricultural employment reduces food security by diminishing the cultivated land areas being used.

Non-agricultural employment may also lead to the loss of agricultural technical knowledge [43]. When laborers with more agricultural technology move to non-farming areas, there are fewer opportunities to pass on cultivation experience. It is difficult to realize the renewal of agricultural technology on the basis of accumulation. Farmers who are of a lower labor quality find it difficult to give full play to the potential of the land, which inhibits the increase of food-crop production per unit area [44]. Furthermore, non-agricultural employment will promote the development of industry in the local neighborhood, which may indirectly aggravate environmental problems, such as the pollution of arable land and water resources, which affect the capacity and quality of crops and reduce the yield of food crops per unit area [45]. This series of consequences means that the overall business capacity and production level in the field of food cultivation has decreased, which will constitute the food-security problem. Therefore, this study proposes Hypothesis 1c.

Hypothesis 1c:

The increase in the share of non-agricultural employment reduces food security by lowering food production per unit area.

However, non-agricultural employment does not have only negative effects on food security. Existing studies show that the stock and increase of scale agriculture in a county are positively correlated with local non-agricultural employment. The higher the share of non-agricultural employment, the more it increases the surrounding intensive operation of agriculture, especially large farms [46]. Although labor resources in agriculture are lost, the average area of arable land being used per laborer increases, which precisely creates conditions for intensive operation of the land. The centralized management of previously dispersed land through the transfer of land-management rights will help integrate the resources of the agricultural sector. Intensive operation makes the cost per unit of product decrease, transforms the traditional smallholder business model into an intensive business model, and deepens the division of labor to increase labor productivity. In short, the agricultural scale economy formed by intensive operation may promote the cost reduction and efficiency of the food-production chain, playing a positive role in local food security. As a result, it is possible to formulate Hypothesis 2a.

Hypothesis 2a:

The increase in the share of non-agricultural employment promotes intensive operation of the land, thereby contributing to food security.

Non-agricultural employment could push traditional agricultural units to subordinate themselves to corporatized business models and expose workers previously bound to the agricultural sector to business models and management practices in the non-agricultural sector [47]. The non-agricultural sector can provide them with broader information channels and platforms. Contemporary agricultural management techniques and marketing strategies can be transferred to rural areas, which is conducive to fostering entrepreneurship in the agricultural sector [47]. Physically, non-agricultural employment can provide farmers with more income, which can be channeled into entrepreneurial activities in the agricultural sector [48]. All of the above factors are likely to promote the formation of new types of agricultural operators represented by agricultural enterprises. As an important factor in promoting agricultural modernization, new types of agricultural operators can promote the integration of resources in the agricultural sector and the optimization of the supply chain through modern administration, thus enhancing the efficiency of food production and agricultural businesses. Therefore, this study proposes Hypothesis 2b.

Hypothesis 2b:

The increase in the share of non-agricultural employment will foster new types of agricultural operators, thereby contributing to food security.

3. Methodology: Model and Data

3.1. Model Design

In existing studies on the impact of non-farm employment, linear models are often used. In some of the literature, first-order linear models are used to study the impact of the non-agricultural economy on the regional development index [49]. There are also studies using first-order linear regression models to analyze the influencing factors of food-security indicators [50] or the employment index [51]. Based on the existing research, this article uses a multiple linear regression model to analyze the impact of non-agricultural employment on food security, which is reasonable and feasible. To test Hypotheses 1a, 1b, and 1c, this study examines the impact of non-agricultural employment on three indicators related to food production, respectively. The model is set up as follows.

$${YIELD}_{it}={\alpha }_0+{\alpha }_1{JOB}_{it}+{{\phi }_{1}X}_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(1)

$${AREA}_{it}={\beta }_0+{\beta }_1{JOB}_{it}+{{\phi }_{2}X}_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(2)

$${RATE}_{it}={\gamma }_0+{\gamma }_1{JOB}_{it}+{{\phi }_{3}X}_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(3)

where YIELD denotes total grain production, AREA is cultivated land area, and RATE means grain crop production per unit area. JOB is the share of non-agricultural employment. X is a control variable that may affect total grain production. α₀, β₀, and γ₀ are constant terms. α₁, β₁, and γ₁ stand for the effect of non-agricultural employment on total grain production. φ₁, φ₂, and φ₃ are the coefficients of the control variable. μ_i, v_t, and ε_it, respectively, represent individual fixed effect, time fixed effect, and error term.

3.2. Variable Selection

3.2.1. Core Explanatory Variable

The core explanatory variable of this paper is the share of non-agricultural employment (JOB). This variable refers to the proportion of non-agricultural employment to total employment, which is an index that captures the economic and demographic structure of a society. This study uses the ratio of year-end non-agricultural employment to total employment, quantifying the share of non-agricultural employment.

3.2.2. Dependent Variables

Three of the dependent variables are described below. Total food production is (YIELD). Total food output is the most direct indicator of food security and embodies the approximate level of food production [37]. The logarithm of the sum of grain yields is used to gauge total food production here.

Arable land being used is (AREA). Arable land is the dominant place for crop cultivation. The size of the arable land is related to the capacity for food production, while arable land abandonment threatens food security [41]. In this study, the logarithm of the amount of commonly used arable land is considered as the indicator corresponding to arable land being used.

Grain yield per unit area is (RATE). The output per unit area of food mirrors the potential productivity of a crop on a given piece of land [10]. Higher yields per unit area imply that a larger food supply is available, enhancing food security. In this article, the logarithm of crop yield per unit area, i.e., the logarithm of the ratio of total food production to the total area sown with crops, is utilized to estimate the magnitude of grain yield per unit area.

3.2.3. Control Variables

The following six control variables are selected for this paper. Urbanization is (URBAN). Urbanization poses challenges to land use and food production in many transition economies [52]. Studies have shown that urbanization has scale and neighborhood effects on the conversion of cropland [53]. Urban population growth and consumption habits brought about by urbanization may lead to changes in food market demand, which is closely related to cropping decisions and local cultivation structure. This study takes the ratio of the city household population to the total household population in a given year to quantify urbanization.

Industrialization is (INDUST). Population flows in industrialized countries have led to numerous problems, such as arable land abandonment [54]. Industrialization is often coupled with the modernization of agriculture. The adoption of new technologies and the spread of machinery or devices have contributed to productivity. Yet the pollutants and wastes generated during industrialization may have an adverse effect on soil and water sources, thus raising insecurity. This research uses the ratio of the value added of the secondary industry to the gross regional product to calculate this industrialization index.

Informatization is (INFORM). Intelligence and automation can be enhanced through the application of information technology. Informatization promotes knowledge and technological innovation and improves market distribution and sales channels for agricultural products. Informatization is strongly correlated with the digital economy, and studies have shown that an increase in the level of the digital economy significantly contributes to the development of food security in China [21]. In this study, the ratio of the number of local broadband access subscribers to the total number of households at the end of the year is used as an indicator of informatization.

Economic development is (CAPITA). Economic development has a vital bearing on local food production. GDP is closely related to agricultural land abandonment. Lower levels of mechanization and farmers’ income in areas with poor economic development are more likely to lead to agricultural land abandonment [43]. It influences agricultural technology inputs, farmers’ incomes, investment capacity, and agro-trade opportunities, shaping the environment of agricultural policy support. In this study, the logarithm of GDP per capita is employed to measure local economic development.

Government intervention is (GOV). On the one hand, through multifaceted interventions in agricultural policy, land policy, market regulation, and environmental protection, governments can guide and facilitate the promotion of local food production. On the other hand, the uncertainty of economic policies has a remarkable negative influence on food security [55]. This research adopts the share of the local general budget expenditures in the regional GDP as a measure of local government intervention.

The rural–urban income gap is (GAP). A striking urban–rural income gap tends to attract laborers to abandon farming [43]. The urban–rural income gap may imply that local rural areas lack sufficient funds and resources for technological innovation and agricultural modernization. Rural areas may not benefit from agricultural innovations due to resource constraints, thus limiting the release of the food-production potential. The urban–rural income ratio is used as an anchor for measuring the urban–rural income gap.

In addition to setting control variables, this paper further considers county and year fixed effects, which are crucial to controlling for the effects of unobservable variables. In this way, this study effectively mitigates the effects on the estimation results owing to omitted variables, thus strengthening the reliability and accuracy of the outcomes. Descriptive statistics for each variable are given in

Table 1. Descriptive statistics of variables.

Variables	Mean	SD	Min	Max	Kurtosis	Skewness	K-S Test
							D-Value	p-Value
YIELD	11.76	0.963	7.772	14.13	3.130	−0.411	0.0203	0.003
AREA	10.23	0.708	7.703	12.27	3.427	−0.359	0.0209	0.002
RATE	8.037	0.602	3.876	10.32	6.552	−0.828	0.0652	0.000
JOB	0.466	0.187	0.006	0.992	2.517	−0.177	0.0380	0.000
AVERAGE	0.449	0.676	0.030	27.40	707.4	21.46	0.2107	0.000
ENTER	4.003	1.671	0.000	9.076	2.398	−0.320	0.0420	0.000
URBAN	0.188	0.153	−1.572	0.831	37.66	−2.531	0.1174	0.000
INDUST	0.439	0.164	2.78 × 10−5	0.939	2.805	0.252	0.0315	0.000
INFORM	3.393	0.992	−3.851	7.993	7.175	−0.896	0.0212	0.002
CAPITA	9.696	1.085	4.443	16.69	3.715	0.047	0.0189	0.006
GOV	2.872	0.715	0.413	6.215	3.038	0.162	0.0137	0.069
GAP	1.074	0.451	−0.980	4.924	6.721	0.947	0.0643	0.000

. The data were routinely cleaned using multiple imputations to fill in missing values for some variables. Finally, this study acquired 7106 annual observations on 323 counties belonging to Class I old revolutionary base areas. The distribution histogram of the above 12 variables can be seen in Figure 2.

3.3. Data Sources

The sample of this study includes the panel data of county-level administrative units in China’s Class I old revolutionary base areas from 2000 to 2021, which are identified based on the official documents issued by a total of 18 provincial-level administrative units in Jiangxi Province, Hunan Province, and others. The old revolutionary base area is a typical underdeveloped area, and it is experiencing rapid non-agricultural employment under the influence of the national regional balanced development policy in the past decade [56]. The data for the research come from the 2000–2021 China County Statistical Yearbook. The information on the number of newly registered agribusinesses in the counties of the old revolutionary base areas in the past year is especially derived from the dataset of newly registered businesses in the sub-sectors of 2991 counties. Figure 3 shows China’s Class I old revolutionary base areas in the sample of this study.

4. Results and Discussion

4.1. Base Regression Analysis

The analysis of the data in this study was based on the software Stata (Stata/MP 16, StataCorp, College Station, TX, USA). The results of the benchmark regression of the non-agricultural employment (JOB) on total food production (YIELD) are presented in

Table 2. Benchmark results of the non-agricultural employment on total food production.

Variables	YIELD	YIELD	YIELD	YIELD
	(1)	(2)	(3)	(4)
JOB	−0.0518	−0.0652 *	−0.0288	−0.0290
	(−1.77)	(−2.26)	(−0.97)	(−1.01)
URBAN	−0.0820	−0.0990 *		−0.0682
	(−1.63)	(−1.98)		(−1.38)
INDUST	−0.238 ***	−0.205 ***		−0.514 ***
	(−6.30)	(−5.52)		(−12.12)
INFORM	−0.109 ***	−0.0943 ***		−0.0718 ***
	(−14.62)	(−12.87)		(−8.79)
CAPITA	0.179 ***	0.173 ***		0.163 ***
	(31.19)	(30.75)		(18.64)
GOV	−0.0463 ***	−0.0383 ***		−0.105 ***
	(−4.58)	(−3.83)		(−8.13)
GAP	−0.0449 ***	−0.0351 **		−0.0456 ***
	(−3.76)	(−2.98)		(−3.91)
County FE	NO	YES	YES	YES
Year FE	NO	NO	YES	YES
N	7106	7106	7106	7106

NOTE: ***, **, and *, respectively, signify significance levels of 1%, 5%, and 10%. The values in brackets are standard errors.

. From the results, it is evident that the coefficients of the effects are not statistically significant, except for the result in column (2), which shows that an increase in the share of non-agricultural employment does not necessarily have a negative effect on total food production. Therefore, Hypothesis 1a is not supported. Although this result does not support Hypothesis 1a, it is consistent with the findings of some established studies. In Nigeria, non-agricultural employment has no significant effect on total food-crop production [29].

Table 3. Benchmark results of the non-agricultural employment on arable land being used.

Variables	AREA	AREA	AREA	AREA
	(1)	(2)	(3)	(4)
JOB	−0.166 ***	−0.169 ***	−0.196 ***	−0.212 ***
	(−9.52)	(−9.83)	(−11.53)	(−12.44)
URBAN	0.419 ***	0.420 ***		0.355 ***
	(13.94)	(14.04)		(12.08)
INDUST	−0.112 ***	−0.109 ***		−0.0374
	(−5.01)	(−4.90)		(−1.48)
INFORM	0.0462 ***	0.0508 ***		0.0197 ***
	(10.45)	(11.61)		(4.07)
CAPITA	0.0586 ***	0.0567 ***		0.00869
	(17.22)	(16.85)		(1.68)
GOV	0.00376	0.00507		−0.0404 ***
	(0.62)	(0.85)		(−5.27)
GAP	0.0152 *	0.0172 *		0.0382 ***
	(2.14)	(2.44)		(5.52)
County FE	NO	YES	YES	YES
Year FE	NO	NO	YES	YES
N	7106	7106	7106	7106

NOTE: *** and *, respectively, signify significance levels of 1% and 10%. The values in brackets are standard errors.

provides the regression results of non-agricultural employment (JOB) on the area of arable land being used (AREA). The effect coefficients are significantly negative with or without fixed effects, suggesting that non-agricultural employment can have a detrimental effect on the stock of commonly used arable land. This is probably intimately associated with the decline in factor inputs in the agricultural sector resulting from an increase in the share of non-agricultural employment. The increase in non-agricultural employment has meant an outflow of labor previously engaged in food production, accompanied by a withdrawal of capital from the agricultural sector. As a result of these factors, the comprehensive food-production capacity has deteriorated, and it has become impossible to sustain operations on formerly utilized arable land. The abandonment of cultivated land has become an inevitable consequence, which is manifested in a decline in the total area of commonly cultivated land. Additionally, against the backdrop of labor migration, the economic takeoff of the non-agricultural sector has been accompanied by the expansion of its land use. This may lead to the conversion and shrinkage of the existing arable land. Based on the empirical results, Hypothesis 1b is confirmed.

Table 4. Benchmark results of the non-agricultural employment on grain production per unit area.

Variables	RATE	RATE	RATE	RATE
	(1)	(2)	(3)	(4)
JOB	−0.110 ***	−0.126 ***	−0.134 ***	−0.110 ***
	(−3.86)	(−4.41)	(−4.52)	(−3.86)
URBAN	−0.221 ***	−0.226 ***		−0.229 ***
	(−4.58)	(−4.56)		(−4.65)
INDUST	−0.305 ***	−0.293 ***		−0.499 ***
	(−8.31)	(−7.94)		(−11.81)
INFORM	−0.0816 ***	−0.0767 ***		−0.0768 ***
	(−11.36)	(−10.55)		(−9.43)
CAPITA	0.197 ***	0.196 ***		0.154 ***
	(35.35)	(35.08)		(17.75)
GOV	−0.0514 ***	−0.0446 ***		−0.124 ***
	(−5.26)	(−4.49)		(−9.60)
GAP	−0.00687	0.00136		−0.00168
	(−0.59)	(0.12)		(−0.14)
County FE	NO	YES	YES	YES
Year FE	NO	NO	YES	YES
N	7106	7106	7106	7106

NOTE: *** signifies the significance level of 1%. The values in brackets are standard errors.

presents the results of non-agricultural employment (JOB) on grain production per unit area (RATE). The coefficient is significantly negative, irrespective of whether individual or time-fixed effects are set or not. Non-agricultural employment significantly and adversely affects the output capacity of food. There are two possible explanations. From a technology-based perspective, it is evident that the exodus of the young and experienced workers that accompanies the flourishing of non-agricultural employment can drain away agricultural technology, including farm-machinery operation, breeding, and irrigation. Peasants with poor farming skills have limited utilization of arable land, which hampers the increase in food-crop yields per unit area. From an ecological perspective, the pollution that may be caused by the booming of non-farm employment equally endangers the quality of arable land, especially its output level. The regression results provide support for Hypothesis 1c.

4.2. Robustness Tests

The robustness test was conducted in this study by three methods: replacing the independent variable, trimming the tails, and adjusting the sample period to examine the reliability of the baseline regression results of non-agricultural employment on grain production per unit area.

The robustness test begins with changing the measure of the independent variable. The aggregate of the employees in the secondary and tertiary sectors at the end of the year is used as the index of the non-agricultural employment. The results are displayed in the first column of

Table 5. Robustness tests of the non-agricultural employment on grain production per unit area.

Variables	RATE	RATE	RATE
	(1)	(2)	(3)
JOB	−0.0142 **	−0.102 ***	−0.0975 ***
	(−2.69)	(−3.96)	(−3.87)
URBAN	−0.245 ***	−0.188 ***	−0.000337
	(−4.99)	(−4.26)	(−0.01)
INDUST	−0.504 ***	−0.415 ***	−0.147 **
	(−11.91)	(−10.98)	(−3.21)
INFORM	−0.0762 ***	−0.0725 ***	−0.0407 ***
	(−9.36)	(−9.97)	(−4.39)
CAPITA	0.153 ***	0.129 ***	0.108 ***
	(17.60)	(16.63)	(10.82)
GOV	−0.124 ***	−0.0914 ***	0.00249
	(−9.65)	(−7.95)	(0.19)
GAP	−0.000548	0.00624	0.0796 ***
	(−0.05)	(0.60)	(4.12)
County FE	YES	YES	YES
Year FE	YES	YES	YES
Constant	7.329 ***	7.425 ***	7.156 ***
	(79.60)	(90.14)	(62.33)
N	7106	7106	4199

NOTE: *** and **, respectively, signify significance levels of 1% and 5%. The values in brackets are standard errors.

, showing that the effect remains significantly negative. Considering the large sample size of old revolutionary base areas and the variety of natural and social factors involved, this study trimmed the data at the upper and lower 1% levels, i.e., values less than 1% were uniformly replaced with 1% values, and values greater than 99% were uniformly replaced with 99% values. The results in the second column of Table 5 suggest that the model estimates are generally consistent with the benchmark regression except for a slightly different regression coefficient. Adjusting the sample period is also employed in this study. On 1 January 2008, the Law of the People’s Republic of China on the Promotion of Employment was formally implemented, which explicitly stipulates that the state establishes and improves a system of equal employment for urban and rural workers, guides the orderly transfer of surplus agricultural labor for employment, guarantees the equal rights of laborers employed in different sectors, and promotes an increase in the proportion of non-agricultural employment. For this reason, this study shortens the time window and retains data only for 2008 and beyond, which helps to study the impact of non-agricultural employment on food production in a more targeted manner. Moreover, in order to avoid the possible disturbance caused by the new crown epidemic in 2020, this study excludes the data from 2020. The results after re-regression are listed in the third column of Table 5, except that the absolute value of the coefficient is a little varied, with it basically and roughly being the same as that of the benchmark regression.

4.3. Further Discussion

As the above results reveal, the de-agriculturalization of employment in the old revolutionary base areas will lead to a series of problems that threaten food production, including the abandonment of arable land and lower yield per unit area. This undoubtedly poses a serious challenge to the cause of food security. Nonetheless, the effects of non-agricultural employment on food security are complex and multifaceted, which are not limited to negative ones. This study will further explore the potential positive impacts of non-agricultural employment on food security in order to stimulate the positivity of non-agricultural employment, so as to offset or even eliminate the negativity brought about by the loss of labor in the agricultural sector. The following two dimensions will be considered: intensive operation of farmland and new types of agricultural entities.

4.3.1. Intensive Operation of Land

For the old revolutionary base areas, it is worthwhile to pay attention to the question of whether non-agricultural employment can contribute to the challenges of food security by positively influencing the intensive operation of land. This is analyzed in this study by constructing the following model.

$$Y_{it}={\alpha }_0+{\alpha }_1{JOB}_{it}+{{\phi }_{1}X}_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(4)

$${AVERAGE}_{it}={\beta }_0+{\beta }_1{JOB}_{it}+{{\phi }_{2}X}_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(5)

$$Y_{it}={\gamma }_0+{\gamma }_1{JOB}_{it}+{{\gamma }_{2}AVERAGE}_{it}+{{\phi }_{3}X}_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(6)

where Y_it represents the three dependent variables YIELD, AREA, and RATE in the benchmark regression analysis; AVERAGE_it equals the ratio of the common cultivated land area to the number of agricultural employees, which is an indicator of the intensive operation of land. The results of this model are displayed in

Table 6. Further discussion of intensive operation of land.

Variables	AVERAGE	YIELD	YIELD	AREA	AREA	RATE	RATE
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
JOB	0.660 ***	−0.0290	−0.0438	−0.212 ***	−0.239 ***	−0.110 ***	−0.124 ***
	(11.38)	(−1.01)	(−1.51)	(−12.44)	(−14.04)	(−3.86)	(−4.30)
AVERAGE			0.0224 ***		0.0412 ***		0.0207 ***
			(3.72)		(11.63)		(3.45)
Controls	YES	YES	YES	YES	YES	YES	YES
County FE	YES	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES	YES
N	7106	7106	7106	7106	7106	7106	7106

NOTE: *** signifies the significance level of 1%. The values in brackets are standard errors.

, from which it can be seen that the coefficient of the proportion for non-agricultural employment on the level of intensive operation of land is significantly positive. Moreover, the coefficients of the level of intensive operation of land on the level of total food production, area of arable land being used, and grain yield per unit area are all significantly positive. This can be attributed to the fact that producers formerly engaged in agriculture shifted to non-agricultural industries, and the average area of arable land being used per laborer in the agricultural sector increased, facilitating the intensification of operations. Non-agricultural employment may increase the intensive operation of the land being used, thereby raising the area of arable land being used and the yield of food crops per unit area. Cost reductions and economies of scale are realized in the food-production chain, which is beneficial for food security. Hence, Hypothesis 2a is valid. With the insertion of AVERAGE_it, however, the coefficient of the non-agricultural employment share on the level of acreage being used and grain yield per unit area remains significantly negative. To a greater extent, the rising share of non-agricultural employment reduces the overall level of grain production and operation. It is probably owing to incomplete supporting policies in the old revolutionary base areas and the influence of traditional land concepts that the transfer of land is still impeded in reality. Especially, some non-agricultural workers are unable or unwilling to transfer their use rights to the land. For these reasons, the average area of cultivated land per worker is inflated, and the positive impact of intensive operation on food production is greatly restricted. Arable land abandonment due to insufficient food-production capacity still exists, and grain yields per unit area are still falling.

4.3.2. New Types of Agricultural Operators

As an important factor in agricultural business management, new types of agricultural operators play a role in grain production. To study the association among non-agricultural employment, new types of agricultural operators, and the indicators related to food production, this study constructs the following model.

$$Y_{it}={\alpha }_0+{\alpha }_1{JOB}_{it}+{{\phi }_{1}X}_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(7)

$${ENTER}_{it}={\beta }_0+{\beta }_1{JOB}_{it}+{{\phi }_{2}X}_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(8)

$$Y_{it}={\gamma }_0+{\gamma }_1{JOB}_{it}+{{\gamma }_{2}ENTER}_{it}+{{\phi }_{3}X}_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(9)

where the meaning of Y_it is the same as the one in (4) and (6); ENTER_it, the log of the number of registered agricultural, forestry, animal husbandry, and fishery enterprises in the year, is used as an index of new types of agricultural operators. The regression results can be seen in

Table 7. Further discussion of new types of agricultural operators.

Variables	ENTER	YIELD	YIELD	AREA	AREA	RATE	RATE
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
JOB	−0.0591	−0.0290	−0.0273	−0.212 ***	−0.212 ***	−0.110 ***	−0.109 ***
	(−0.78)	(−1.01)	(−0.95)	(−12.44)	(−12.44)	(−3.86)	(−3.83)
ENTER			0.0290 ***		−0.000786		0.0171 ***
			(6.34)		(−0.29)		(3.74)
Controls	YES	YES	YES	YES	YES	YES	YES
County FE	YES	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES	YES
N	7106	7106	7106	7106	7106	7106	7106

NOTE: *** signifies the significance level of 1%. The values in brackets are standard errors.

. The coefficient of the share of non-agricultural employment on the growth of new types of agricultural operators is negative, but not statistically significant. Empirical results do not support Hypothesis 2b. Although new types of agricultural operators in old revolutionary base areas have a positive effect on increasing food-crop yields, non-agricultural employment does not promote the fostering of new agricultural operators. This may be a result not only of the relative shortage of labor and capital in agriculture but also due to the insufficient infrastructure and relevant services needed for the formation and expansion of agribusinesses in old revolutionary base areas. In reality, agribusinesses need a good industrial foundation and supportive policies, such as irrigation systems, agricultural processing facilities, logistics, and market channels. Yet the old revolutionary base areas are comparatively backward in these areas, which makes it difficult to meet the needs and greatly limits the growth of new types of agricultural operators. In addition, though non-agriculture brings workers into contact with new business models and management practices, the isolated surroundings and mindset of the old revolutionary base areas also hinder the cultivation of an entrepreneurial spirit among the laborers.

Through the above analysis in Section 4.1, Section 4.2 and Section 4.3, the test results of the five hypotheses in this study are presented in the following,

Table 8. Results of all the hypotheses.

Hypotheses	Contents	Accepted/Refuted
H1a	The increase in the share of non-agricultural employment reduces food security by cutting down total grain production.	Refuted
H1b	The increase in the share of non-agricultural employment reduces food security by diminishing the cultivated land areas being used.	Accepted
H1c	The increase in the share of non-agricultural employment reduces food security by lowering food production per unit area.	Accepted
H2a	The increase in the share of non-agricultural employment promotes intensive operation of the land, thereby contributing to food security.	Accepted
H2b	The increase in the share of non-agricultural employment will foster new types of agricultural operators, thereby contributing to food security.	Refuted

.

4.4. Heterogeneity Analysis

In the benchmark analysis, the effect of non-agricultural employment on unit yields of food crops is significantly negative. But, China is a large and sprawling country, with large variations in the level of economic development, the implementation of specific policies, and cultural attitudes among different districts. Existing research confirms that there are regional differences in grain production [57]. Whilst the districts and counties in this study are the same old revolutionary base areas, they are located in different economic districts with distinctive features and characteristics. So, further heterogeneity analysis is necessary. Initially, this study categorizes counties into two groups according to whether they are major grain-producing areas or not. At the provincial level, there are thirteen provinces or autonomous regions in China, including Heilongjiang, Jilin, Liaoning, Inner Mongolia Autonomous Region, Hebei, Henan, Shandong, Jiangsu, Anhui, Jiangxi, Hubei, Hunan, and Sichuan. The counties belonging to the above thirteen provincial administrative units are taken as one group and the rest as another group. Regression analyses are conducted separately for the two groups, and the results are shown in columns (1) and (2) of

Table 9. Heterogeneity analysis results of the non-agricultural employment on grain production per unit area.

Variables	RATE	RATE	RATE	RATE
	(1) Non-Major	(2) Major	(3) North	(4) South
JOB	−0.407 ***	0.171 ***	−0.190 **	0.300 ***
	(−4.85)	(8.72)	(−2.92)	(14.35)
Control Variables	YES	YES	YES	YES
County FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
N	3388	3718	3234	3872

NOTE: *** and **, respectively, signify significance levels of 1% and 5%. The values in brackets are standard errors.

. Next, this study classifies the counties into two groups: the northern region and the southern region. In all samples, the counties belonging to the provincial administrative units of Zhejiang, Anhui, Fujian, Jiangxi, Hubei, Hunan, Guangdong, Chongqing, Sichuan, and Guangxi Zhuang Autonomous Region are categorized as southern regions, and counties belonging to the provincial administrative units of Tianjin, Hebei, Shanxi, Shandong, Henan, Shaanxi, Gansu, and Ningxia Hui Autonomous Region are regarded as northern regions. The two groups were analyzed in separate regressions, and the results are shown in columns (3) and (4) of Table 9.

The results in columns (1) and (2) demonstrate that, for old revolutionary base areas that are not main grain-producing areas, non-agricultural employment has a significant negative impact on the yield per unit area, whereas, for old revolutionary base areas that are main grain-producing areas, the coefficient is significantly positive. Therefore, for the main grain-producing areas, increased non-agricultural employment may, instead, enable the local grain yield per unit area to increase. There are two possible explanations. To begin with, the main grain-producing areas are usually endowed with rich agricultural resources. Further, the increase in non-agricultural employment, on the contrary, alleviates the unfavorable situation of excess agricultural labor, and thus resources can be more optimally utilized. Secondly, in the main grain-producing areas, knowledge of agricultural mechanization and intensification can be disseminated and applied better to grain production, promoting modernization and efficiency, which will, in turn, increase the yield per unit area of grain. Notably, the results of (3) and (4) reflect north–south differences. For the old revolutionary base areas belonging to the northern region, the coefficient is significantly negative, while for those belonging to the southern region, the coefficient is significantly positive. This north–south disparity is ultimately due to the significant characteristics of the climate, soils, crop types, etc. in the northern and southern regions. The southern region usually has a warm and humid climate and fertile soils, coupled with more abundant water resources and temperatures suitable for multiple croppings in a year. Consequently, the increase in the share of non-agricultural employment is more beneficial than detrimental to the food-production capacity of the southern region, which is rendered as a positive coefficient in the regression results.

5. Final Remarks

5.1. Conclusions

In this study, it is discovered that, despite the insignificant effect of non-agricultural employment on total food production, it has a significant negative impact on both cultivated land area and the land’s food productivity, which indicates that non-agricultural employment does pose a challenge to food security of counties located in China’s old revolutionary base areas. The study adopted three robustness tests by replacing the independent variables, shrinking the tails, and adjusting the sample period. The results consistently supported the above findings. In the process of economic catch-up, food production, environment, and other sustainable problems in the old revolutionary base areas, which are unavoidably brought about by less developed areas, are also worthy of attention.

Furthermore, this study examines the potential positive benefits of non-agricultural employment on food security from three aspects: intensive operation of land, agricultural mechanization, and new types of agricultural operators. The outcomes show that non-agricultural employment can significantly promote the intensive operation of the land, thus offsetting the negative impact of non-agricultural employment on food security. Rather, the effects of non-agricultural employment on agricultural mechanization and new types of agricultural operators are not significant, and food security is still under certain threat. Heterogeneity analysis reveals significant regional heterogeneity in the role of non-agricultural employment on food security. For the main grain-producing regions and the southern counties, non-agricultural employment has a positive effect on increasing yields per unit area of food crops, which is stronger than the negative effect on food security. The same cannot be said for non-main grain-producing regions and the northern regions.

5.2. Recommendations

Apart from adhering to existing policy measures for food security, some important recommendations are proposed based on the research in this paper. Given that non-agricultural employment facilitates intensive operation of the land and, thus, supports food security, regulating land transfers has become the top priority of the policy recommendations [39]. Simplify land-transfer procedures and administrative approvals to attract more farmers to participate in land transfer [58]. Improve the system of land transfer and laws and regulations, and clarify the rights and responsibilities of relevant subjects, which can protect the legitimate rights and interests of all sides, activate idle rural land resources, and increase land-capitalization income [59]. In terms of financial policies, the establishment of special funds and the provision of loans or subsidies will encourage farmers to engage in land transfer and leasing. All of the above policies to promote the optimal allocation of land resources can promote intensive operation of agricultural land and exert its positive effects. It is also necessary to strengthen the education of farmers [60]. The government, especially the Ministry of Agriculture and Rural Affairs, should not only push for a change in the traditional concept of the land but also provide training to upgrade the quality of laborers, including skills in operating agricultural machinery and modern management. Agricultural departments at all levels of local government, such as agricultural bureaus, should also play an important role. Furthermore, local agricultural cooperatives can also be an important channel for farmer training. These entities are more attuned to the realities of local agricultural production and are, thus, in a better position to provide practical training.

The government should take measures to formulate supporting employment in accordance with local conditions [61] and release potential dividends by supporting the stable employment of rural workers in the non-agricultural sector. Measures should be taken to support and encourage surplus agricultural labor to adopt efficient agricultural production methods and modern agricultural industries. It is worth noting that the soil characteristics of China are diverse [62]. In addition, it is necessary to formulate corresponding soil-management measures and large-scale agricultural management policies according to the specific geographical environment, soil characteristics, and agricultural needs, especially in mountainous areas, hilly areas, and other special terrain [63]. Local governments should also strengthen the coordination of regional development plans and policies and promote mutually beneficial cooperation between major grain-producing areas and non-major grain-producing areas and between southern and northern regions. Through grain transfer and logistics, it can realize complementary resource advantages, mutual benefits, and a win–win situation and ensure national food security.