1. Introduction
Rice is the most important food crop in the world, and more than half of the global population consumes rice as their primary source of caloric intake. The world’s population is projected to reach 9 billion by 2050, and the demand for rice is expected to increase by 28% during that time span [
1,
2]. China is the world’s largest rice producer and consumer, and the continuous increase in rice production is important to ensure national food security in China. There is increasing interest in identifying effective options for sustainable rice production and food security.
Ratoon rice production provides an opportunity for productivity increases and food security in China [
3]. It refers to the practices of harvesting the first crop (main rice) and further obtaining a second crop that sprouts from the stem nodes on the previously harvested crop in a particular year [
4]. Ratoon rice achieves a yield increase with minimal agricultural inputs as the second crop does not require the activities of tillage, sowing, and transplanting [
5]. It increases the potential for additional revenue from a second crop on the same land base [
6]. Despite these benefits, the promotion of ratoon rice production remains in the preliminary stages. Yet, limited empirical evidence regarding farmers’ ratoon rice production behavior is available.
It has been estimated that the potential total paddy area suitable for ratoon rice in China is 18.69 million hectares [
7]. The Chinese government is expected to promote the cultivated areas of ratoon rice to 1 million hectares by 2025, especially in the middle and lower reaches of the Yangtze River. In recent decades, China’s rapid urbanization and continuous industrialization have created off-farm employment opportunities for agricultural households [
8,
9]. The number of rural off-farm laborers reached 171.7 million in 2021, accounting for 58.7% of the total number of migrant workers [
10]. Almost 85% of rural households in China have at least one family member who works in the non-agricultural sector [
11]. Some studies reported that off-farm employment insignificantly increases or decreases rice production [
12,
13], while others demonstrated that off-farm activities increase rice production [
14].
The empirical evidence of the relationship between off-farm employment and rice production is mixed. The intra-household division of labor and the socialized division of labor are considered to be popular strategies adopted by agricultural households working in off-farm activities [
15,
16,
17]. The intra-household division of labor is manifested by the gender-based division of labor within the household. This has led to the feminization of agricultural labor as men have migrated at a higher rate than women. Thus, women have become increasingly responsible for household farming [
18,
19]. The socialized division of labor includes agricultural socialization services represented by outsourced machinery services [
20]. The substitution of labor with agricultural machines may maintain or increase farm production [
21]. Nevertheless, the impact of the above strategies on farmers’ crop adjustment behavior has not reached a consistent conclusion. Some studies reported that off-farm employment reduces the labor available for agricultural production, thus decreasing rice production and increasing the production of other crops [
12,
13]. Farmers increase mechanical input to replace the shortage in the labor force, especially in grain production, which can alleviate the pressure from off-farm employment. Other studies found that rural labor transfer did not lead to a reduction in rice-planting areas from 1996 to 2016 in China [
14]. Furthermore, Zhong et al. [
22] found that the substitution of agricultural machines for labor is relatively difficult in non-plain regions, which inhibits the mitigating impacts of off-farm employment on food crop production.
However, it is noteworthy that existing studies mostly focus on how farmers have adjusted their behavior within traditional rice production. Off-farm employment has contributed to the reduction in multiple crop indexes and led to the shift from double-season to single-season rice [
12,
23]. As agricultural social services have developed rapidly in recent years, it is possible to mitigate the impacts of off-farm employment [
24]. Ratoon rice is increasingly attractive to farmers due to its relatively low labor intensity and high profit [
25]. Rice production is a complex process, especially for different production systems where there are differences in the intensity of labor and mechanical input. Thus, the conclusion obtained using traditional rice production behavior may not apply to ratoon rice production behavior decisions.
To our knowledge, this study is the first attempt to explore the impact of off-farm employment and outsourced machinery services on farmers’ ratoon rice production behavior in China. The main contributions of this study are reflected in the following three aspects: First, ratoon rice has been a promising alternative for single- and double-season rice to increase grain production and ensure food security [
26]. This study is the first attempt to explore farmers’ ratoon rice production behavior. Second, using a multinomial endogenous treatment effect model (METE) to control the impact of endogenous problems, we analyzed the impact of outsourced machinery services on farmers’ green production behavior. Third, we empirically examined the joint impact of off-farm employment and outsourced machinery services on rice production, in an attempt to provide improved policy suggestions for China to promote farmers’ ratoon rice production in the context of the rapid development of labor migration and outsourced machinery.
The rest of this study is organized as follows:
Section 2 presents the theoretical framework of this study.
Section 3 introduces the empirical model, data source, and variables used in the analysis.
Section 4 presents the main result, robustness check, and heterogeneity results, and
Section 5 draws conclusions and presents policy implications.
3. Empirical Methodology
Farmers consider potential combinations when making decisions regarding off-farm employment and outsourced machinery services. The combination of decisions made by farmers is not random, and they endogenously self-select which decisions regarding off-farm employment and outsourced machinery services they will make. These decisions are likely to be influenced by unobservable characteristics that are also related to outcome variables of interest (farmers’ rice production behavior); in this case, endogeneity problems may arise [
36]. For instance, farmers may make a decision about off-farm employment and outsourced machinery services based on unobservable factors such as their ability and motivation, and failure to account for this may overestimate or miscalculate the true impact of decisions concerning ratoon rice production.
An effective method to examine treatment effects without randomization in terms of multiple treatments is the multinomial endogenous treatment effect (METE) model [
37]. This model allows us to evaluate different combinations of decisions between off-farm employment and outsourced machinery services. Moreover, this model accounts for both the interdependence among the combination of decisions and selection bias as a result of observed and unobserved characteristics.
METE is modeled simultaneously in two stages. In the first stage, a farmer makes a combination of decisions between off-farm employment and outsourced machinery services. Following the work of Deb and Trivedi [
37], the first stage is estimated using a mixed multinomial logit (MMNL). Therefore, the utility of the
(
= 1,2,3,4) combination of decisions can be described as follows:
where
is a vector of observed exogenous covariates that represents household head characteristics, family characteristics, farmland characteristics, and socioeconomic characteristics;
is a vector of unobserved covariates such as a farmer’s risk preference and technical abilities;
and
are vectors of parameters to be estimated; and
is the random error term. While
is not observed, we observed the combination of decisions in the form of a set of binary variables
, and these are collected in a vector,
=
. Similarly, let
. Let j = 1 denote farmers with off-farm employment and outsourced machinery services and
. Then, the probability of treatment can be written as:
where
is an appropriate multinomial probability distribution, which can be estimated using the MMNL model as follows:
In the second stage of METE, we estimate the effect of the combination of decisions on outcome variables. The expected outcome equation is formulated as:
where
is ratoon rice adoption for farmer i, and
= 1 if farmer i adopts ratoon rice. Specifically, coefficient
indicates the effect of the combination of decisions on ratoon rice adoption. Since
is a function of the latent factors
, the outcome is affected by unobserved characteristics that also affect selection into treatment. When
is positive (negative), the treatment is positive (negative), but the outcomes are negatively (positively) correlated through unobserved characteristics. The resulting model was estimated using a maximum simulated likelihood (MSL) approach.
4. Data and Variables
4.1. Data Collection
The empirical data used in this study were from the representative household survey for rice farmers conducted between June and September 2018 in Hubei province, central China. The province is a major producer of rice in the country, and it is one of the hotspots of ratoon rice production. The 14th Five-Year Plan for National Planting Industry Development proposes the expansion of ratoon rice cultivated area in the middle and lower reaches of the Yangtze River. Hubei province is expected to experience progressive increases in the cultivated area for ratoon rice, with >220 thousand hectares of ratoon rice being harvested in 2018 in this province [
5].
To obtain a sample with a representative geographic coverage, 9 counties within this province, in which ratoon rice production was popularized, were purposively selected. These counties are distributed across three main paddy regions in Hubei, namely, Japonica rice in the northeast region (Xiaogan), single-season rice in the central region (Shayang and Yingcheng), and single- and double-crop rice fields in the Jianghan plain (Jianglin, Jianli, Honghu, Xianning, Wuxue, and Qichun). Within these regions, we used a multi-stage random sampling approach for data collection. Firstly, 2–3 towns were randomly selected from each county. In Xiaogan, Shayang, and Yicheng counties, we randomly selected three towns from each country. Among the remaining counties, we randomly selected two towns from each county. Then, two villages that were identified based on personal interviews with local agricultural agencies with knowledge regarding rice production were selected from each town. Finally, approximately 45 farmers were randomly selected from each village. Data were collected using field surveys and structured questionnaires at the farm level. Detailed information was provided regarding farmers’ ratoon rice production behavior, demographic characteristics, and agricultural production. The total number of household observations in the questionnaire survey was 1816. Out of these, 64 participating households were excluded from the study sample because of missing data and stark outliers for key outcome variables. Hence, 1752 household observations were used for the econometric analysis, for which, all the key variables were available.
4.2. Variable Selection
4.2.1. Dependent Variable
In this study, we used binary variables to measure farmers’ ratoon rice production behavior. Specifically, the dependent variable was assigned 1 if the respondents answered “yes” to the question “do you cultivate ratoon rice in 2018?” and 0 otherwise. Further, we used the proportion of ratoon rice cultivated area to the total rice cultivated area in a household as an indicator for the robustness test.
4.2.2. Independent Variable
In this paper, off-farm employment refers to all economic activities that take place outside the agricultural household; hence, this includes off-farm salaried and casual wage employment and off-farm self-employment. Outsourced machinery services refer to the process that machinery service organizations or mechanical operation services provide to producers without agricultural machinery equipment in the form of compensation, so as to realize mechanized production. The combination of decisions between off-farm employment and outsourced machinery services can be divided into four types: (1) farmers without off-farm employment and outsourced machinery services (NOEOMS); (2) farmers with off-farm employment (OE); (3) farmers with outsourced machinery services (OMS); (4) farmers with off-farm employment and outsourced machinery services (OEOMS). Finally, 43.4%, 11.4%, 39.9%, and 5.4% of farmers in our samples were classified into the NOEOMS, OE, OMS, and OEOMS groups, respectively.
We used four binary variables to measure the above types of farmers. Specifically, the NOEOMS variable was assigned 1 if a farmer neither participated in off-farm employment nor purchased outsourced machinery services for rice production, and 0 otherwise. Similarly, the OE variable was assigned 1 if a farmer only participated in off-farm employment, and 0 otherwise. The OMS variable was assigned 1 if a farmer purchased outsourced machinery services for rice production, and 0 otherwise. The OEOMS variable was assigned 1 if a farmer participated in off-farm employment and purchased outsourced machinery services for rice production, and 0 otherwise.
As is listed in
Table 2, there was a difference among the four groups with OEOMS as the base category. This shows that off-farm employment can inhibit farmers’ ratoon rice production behavior, but outsourced machinery services can effectively promote farmers’ ratoon rice production behavior and offset the negative impact of off-farm employment.
4.2.3. Control Variable
We included a set of control variables that are likely to affect farmers’ ratoon rice production behavior. Our control variables included household head characteristics (i.e., age, gender, years of education, and health status), household characteristics (i.e., income, agricultural labor force, number of children, and number of elderly people), plot characteristics (i.e., number of plots, plot area, soil fertility, distance to plots, soil texture, land property rights, rice system in adjacent plots, access to irrigation, and farmland road condition), and socioeconomic characteristics (i.e., cooperative services, rice selling price, commercialization rate, technical training, subsidies, and distance to the market).
4.2.4. Instrument Variable
In order to identify the equation for the METE model, an instrument variable is required. Specifically, the instrument needs to be both highly correlated with the combination of decisions and have no direct effect on the variables of interest (i.e., farmers’ ratoon rice production behavior).
In this study, we used the ratio of farmers with off-farm employment and farmers with outsourced machinery services to the total farmers in a village as the instrumental variable. Theoretically, villages with a higher ratio of off-farm employment will affect an individual farmer’s decision regarding off-farm employment through their social capital. Social capital can be regarded as an important informal institution [
38]. Under imperfect market conditions, social capital plays a critical role in reducing information asymmetry in the labor market and securing the off-farm work of farmers [
39]. In villages with higher rates of outsourced machinery services, the agricultural socialized service market is relatively mature within a certain region, which makes it convenient for farmers to purchase the service. These variables are likely to be correlated with the decision concerning off-farm employment and outsourced machinery services but are unlikely to have any direct effect on their adoption. Thus, the instrument is correlated and exogenous with the farmers’ ratoon rice production behavior.
4.3. Descriptive Statistics
Table 3 presents the definitions and descriptive statistics of the variables. The dataset contains 1752 farm households and, of these, approximately 50% adopted RR production. The average age of the sample household head was 58.25 and approximately 90% were male. The average years of education for the household head was approximately 6 years. Nearly half of the sample household heads were healthy. The average household income was CNY 66.99 thousand, and the average amount of household agricultural labor was 2.0. The average number of children and elderly was 0.469 and 1.123, respectively.
In terms of plot characteristics, for each farmer, the average number of plots was 11.78, and the average area of the largest plot was 0.479 ha. For most farmers, the soil fertility of their rice paddies was classified as medium fertility. A total of 26.72%, 21.2%, and 52.1% of the paddy soil texture was sandy, loam, and clay, respectively. Additionally, 81.7% and 11.5% of land property rights were from collectives and transferred in, respectively. Only 6.8% was allocated by both collectives and transferred in. For 93.6% of farmers, the rice system in adjacent plots was the same. A total of 64.8% of the farmers’ rice plots had access to irrigation, and 89.7% of farmers had a paddy field road that was passable for agricultural machinery.
With respect to other socio-economic characteristics, 17.9% of the total sample were cooperative members. On average, the rice selling price was CNY 2.426 per kilogram, and the commercialization rate of rice was 80.4%. A total of 45.3% of the sample households participated in technical training. The average government subsidy received from rice production was CNY 1469 for each farmer. The average distance to the main market was 22.37 min from the residence.
6. Discussion
Ratoon rice production is an effective way to improve productivity and ensure food security in China [
44]. As Chinese rural labor continues to migrate to urban areas for off-farm employment, agricultural machinery has been increasingly used to alleviate the adverse effects of labor shortages in recent years [
45]. In this context, understanding the relationship between factor allocation and ratoon rice production behavior has been of great importance for policymakers in the promotion of ratoon rice.
The results of this study show that outsourced machinery services to enhance rice production play an important role in alleviating the lack of manual labor and ultimately increasing the adoption rate of ratoon rice production. This is generally consistent with the findings of Ji et al. [
46] and Yang and Wei [
47]. In general, off-farm activities reduce household labor, which incentivizes farmers to switch to rice production with a lower labor intensity [
24,
48]. However, outsourced machinery services can act as an effective substitute for insufficient physical and production skills, thus encouraging farmers’ ratoon rice production behavior [
21]. Additionally, machinery services can reduce the technical barriers for farmers in ratoon rice production [
42]. Specifically, the professional skills required for ratoon rice production include the main crop harvest cutting height and the time and amount of fertilizer for the second crop [
4,
5]. Outsourced machinery services help to introduce this technology into ratoon rice production to stabilize grain yields and avoid risks to farmers during the process of learning this production method. This means that it is feasible to promote farmers’ ratoon rice behavior by inducing the agricultural factor flows of labor and mechanization.
There are group differences due to the effect of off-farm employment and outsourced machinery services. Due to physical strength and traditional gender roles, the young and male laborers will allocate more time to off-farm employment to achieve higher earnings [
15,
21]. The complex process of ratoon rice production requires a certain time input for farmers. The reduction in their time devoted to agricultural production results in a significantly lower probability of ratoon rice production relative to elderly female and female farmers. In other words, the elders and females are more likely to adopt ratoon rice production. Agricultural machinery services can be used as a modern production factor, which can effectively supplement the shortage of agricultural labor [
41]. Socialized services have provided a potential approach to replace traditional, under-qualified human capital with specialized services [
49,
50]. The elderly and females tend to purchase machinery services to cope with complex crop production, thus incentivizing them to adopt ratoon rice production. However, income from off-farm employment effectively relieves the liquidity constraints on the purchase of machinery services [
51]. Further, young people and males are more well-educated and skilled in agricultural production [
40,
42]. When they have access to both off-farm employment and socialized services, they are more likely to adopt ratoon rice production. In terms of plain and sloping areas, flat land is suitable for farming, and sloping areas provide a geographical problem for the application of agricultural machinery services [
52]. Based on this advantage, the probability of ratoon rice adoption by farmers in plain areas is significantly higher than that in non-plain areas. Overall, this suggests that farmers in different categories should be rationally guided to adopt ratoon rice production.
Our results can also be applied in a wider context, especially in tropical and sub-tropical areas where ratoon rice is suitable for cultivation. The production system has the potential to be replicated in developing countries, allowing food production to feed the entire population and achieve sustainable food security [
6]. As is the case in China, farmers in Southeast Asia are mostly smallholders, and off-farm employment has been an important way to reduce household livelihood vulnerability [
53]. Machinery services are needed to enhance the adopted technologies and ensure food security [
17]. For instance, in central Vietnam, ratoon rice production is more economically efficient, and this allows farmers to allocate more time to other off-farm activities for income generation [
54]. Meanwhile, increasing the use of socialized services resulted in an increase in the probability of ratoon rice production [
44]. Therefore, our research is helpful in providing a promotion strategy for ratoon rice production in these regions, which are mainly smallholder-based, with an adjustment to household labor and capital.
Although this study attempted to grasp the relationship between off-farm employment, outsourced machinery services, and farmers’ ratoon rice production, there are still some uncertainties. For instance, cross-sectional data may lose precision because ratoon rice production is an innovative agronomic practice in China and farmers’ growing experience will increase over time. Long-term investigation is still needed in future studies; in particular, studies should attempt to capture the dynamics of farmers’ ratoon rice adoption. In addition, the heterogeneity in off-farm employment arising from farmers’ self-selection was neglected. Further research should be conducted on the impact of gender and the intergenerational division of labor in household farmers. The research conclusions thus obtained may be more policy-oriented.
7. Conclusions and Implications
Using a representative household survey of 1752 rice farmers in Hubei province, this study empirically analyzed the impact of off-farm employment and outsourced machinery services on farmers’ ratoon rice production behavior. The main results can be summarized as follows:
(1) Off-farm employment has a significant and negative impact on farmers’ ratoon rice production. Our estimates suggest that the probability of ratoon rice production decreases by 17.2% for farmers with off-farm employment.
(2) Outsourced machinery services significantly encourage farmers’ behavior in ratoon rice production. Specifically, the probability of ratoon rice production increases by 78.9% for farmers with outsourced machinery services. This result suggests that outsourced machinery services can mitigate the negative impact of off-farm employment on farmers’ ratoon rice production behavior.
(3) Farmers with simultaneous off-farm employment and outsourced machinery services are more likely to adopt ratoon rice production. The probability of ratoon rice production increases by 57.0% for farmers with simultaneous off-farm employment and outsourced machinery services.
(4) We also found that outsourced machinery services have a regional heterogeneity effect. The substitution of mechanical machinery for labor is relatively easy in plain areas; therefore, the impact of off-farm employment on farmers’ ratoon rice production behavior is relatively weak. Outsourced machinery services can significantly promote ratoon rice production in non-plain areas.
In the dual context of off-farm employment and outsourced machinery services, ratoon rice production is important to increase grain productivity and ensure food security in China. Some important policy implications can be drawn from the results.
First, given the role of outsourced machinery services in promoting the multiple crop index or crop harvest frequency, the development of ratoon rice production should be promoted to ensure food security. On the one hand, farmers should be encouraged to participate in contiguous regional production. This can expand the market capacity of outsourced machinery services and promote the division of labor to achieve economies of scale. On the other hand, a larger market capacity can help farmers save on production costs and transaction costs. Thus, it is necessary to increase assistance and support to outsourced machinery service providers and actively promote the construction of their basic service conditions and capabilities.
Second, regional heterogeneity effects exist in terms of the impact of mobility factors such as labor and outsourced services on farmers’ ratoon rice production behavior. Specifically, efforts should be made to develop and apply small- and medium-sized agricultural machinery in non-plain areas since it is difficult to substitute outsourced services for labor these areas. Heavy-duty farm machinery should be promoted in plain areas. It is easy to promote agricultural machinery operation services across the region and share the economics of the division. Outsourced machinery services represent a feasible strategy to relieve the negative impact of off-farm employment on rice production.