Modern agriculture depends heavily on pesticide use, which has successfully increased productivity but led to increasing concerns regarding the environment and human health [1
]. For example, Richter [6
] estimated that about 26 million pesticide poisoning cases resulted in 220 thousand deaths per year worldwide. China is the world’s largest pesticide user and experienced a dramatic increase in pesticide use. In particular, the quantity of pesticide used was dramatically increased over time, from 76.53 thousand tons in 1990 to 176.30 thousand tons in 2014 (see Figure A1
in Appendix A
). The booming demand for pesticide in China expanded the share of total world pesticide use and maintained the proportion above 40% over time after 1995. Therefore, a zero-growth action plan for pesticide use has been proposed in China to mitigate agricultural environmental pollution and promote environmental sustainability [7
Chemical pesticide use contributes to increased agricultural production and improved product quality, but overuse or abuse of pesticides has been related to negative externalities such as human health effects and external ecological effects [3
]. For example, Lai [3
] found that a 10% increase in pesticide use in rice production can result in 2.13 million dollars in medical costs in China. Therefore, various policies and measures such as maximum pesticide residue limit (MRL) and integrated pest management (IPM) practices, which target food safety and conservation production, have been released in many developing and developed countries. For example, Midingoyi et al. [13
] found that the IPM-adopting farmers had used less insecticide in mango production in Kenya. In their analysis of Dutch farmers, Skevas et al. [14
] assessed the effectiveness of different economic policies including taxes, price penalties, and subsidies and quotas, and found that quotas of pesticides were a more effective method for reducing pesticide use.
In addition to the deleterious effects of pesticide use on the environment and health of consumers, the negative effects of chemical pesticides on sustainable development of agriculture have also been well documented [3
]. Despite these negative externalities of pesticide use, farmers continue to use a high quantity of pesticide, and policies in developing countries do not address the risk of pesticides adequately [16
Farm machines play an important role in sustainable and conservation agriculture development in emerging and developing countries [18
]. A considerable body of literature has analyzed the impact of the use of farm machines on agricultural production [21
]. For example, Benin [18
] revealed that agricultural mechanization services used by farming households significantly increase farm yields in Ghana. In their analysis of northern Bangladesh, Rahman et al. [25
] showed a positive association between farm machine use and wheat yields. However, there are no previous studies, to the best of our knowledge, that have examined the relationship between farm machine use and pesticide expenditure. It remains unclear whether farm machine use has increased or decreased pesticide expenditure, especially in the presence of self-selection bias associated with voluntary farm machine use.
The impact of farm machine use on pesticide expenditure may be negative or positive. On the one hand, farm machine use can help improve pesticide spraying efficiency, which induces smallholder farmers to spray less. In this case, farm machine use may reduce pesticide expenditure. On the other hand, farm machine use can increase the amount of pesticide use by eliminating the constraints and obligations of pesticide use (e.g., manually spraying skills and physical conditions), resulting in a positive relationship between farm machine use and pesticide expenditure. In addition, farmers themselves decide whether or not to use farm machines for pesticide application (i.e. a self-selection process). In particular, farmers’ farm machine use decisions are likely to be influenced by both observed factors (e.g., age, gender, household size) and unobserved characteristics (e.g., farmers’ innate abilities and managerial skills) that may be correlated to the outcome of interest (pesticide expenditure in our case) [19
]. This fact leads to a sample selection and endogeneity issue, which needs to be addressed in order to obtain an unbiased and consistent estimation of the treatment effect of farm machine use on pesticide expenditure.
The primary objective of this study is to analyze the impact of farm machine use on pesticide expenditure, utilizing data observed at the farm-level in rural China. This study aims to contribute to the growing literature on the role of farm machines in the development of sustainable and conservation agriculture from two aspects. First, we investigate the determinants of farm machine use for pesticide applications by using recently collected survey data from 493 smallholder maize farmers in rural China. The existing studies have so far paid more attention to the total machine power or machine use intensity in their efforts to analyze their determinants. However, studies are almost completely absent on farm machine use at a specific production stage such as pesticide application [20
]. Second, we take the possible endogeneity of farm machine use into account, in particular stemming from the self-selection process, in our estimated econometric models. To achieve this, we use an endogenous switching regression (ESR) model to address the potential selection-bias issue arising from both observed and unobserved factors [27
]. To date, no previous studies have accounted for the impact of farm machine use on pesticide expenditure.
The remainder of this paper unfolds as follows. The data and descriptive statistics are presented in the next section. Section 3
presents the empirical specification. This is followed by presentation of the estimated results and discussion in Section 4
, and Section 5
3. Data and Descriptive Statistics
The data used in this study were from a randomized questionnaire survey of Chinese rural households in January 2017, which refers to the production year 2016. The survey was carried out by a multistage random sampling technique. In the first stage, we randomly selected three provinces from western, central, and eastern China, respectively, including Gansu, Henan, and Shandong. The three regions were taken into account because they were different in terms of economic development levels as well as geographic characteristics. In 2016, the cultivated areas of maize in Gansu, Henan, and Shandong were 1.00, 3.31, and 3.21 million hectares, respectively, accounting for about 20.5% of China’s total maize area [36
]. In the second stage, one city within each selected province was randomly selected. In particular, Heze city in Shandong, Sanmenxia city in Henan, and Dingxi city in Gansu were randomly selected. In the third stage, in each city we randomly selected three villages. Finally, around 45–55 households were interviewed in each village, resulting in 493 samples in total.
Face-to-face interviews were conducted by well-trained enumerators who spoke both Mandarin and local dialects, using a detailed structured questionnaire. The enumerators were hired from local universities in each province. The survey gathered information covering household and farm-level characteristics (e.g., age, gender, education, household size, farm size), farm machine use status in production and postharvest management, maize yields, production inputs (e.g., pesticide, fertilizer and seed), off-farm work participation status, and access to credit institutions.
The definition and summary statistics of the variables used in this study are presented in Table 1
. The dependent variable used in the present study is a dichotomous variable that takes the value of 1 if the household used farm machines to spray pesticides, and the value is 0 otherwise. The outcome variable used in this study refers to the pesticide expenditure per mu. It can be observed from Table 1
that about 57.6% of households in the sample used farm machines to spray their pesticide. The average age of the farm household head is around 47 years, and with a mean household member number of about five. The average 3.5 mu of farm size suggests that the majority of households were small-scale farmers in our sample.
The mean difference in the characteristics of farm machine users and nonusers are presented in Table 2
. With respect to the expenditure of pesticides, the results show that the average pesticide expenditure for farm machine users was 14.07 Yuan/mu lower than that of nonusers, which is statistically significant for the mean difference. These descriptive comparison findings seem to suggest that farm machine use plays a critical role in pesticide use, and significantly reduce the pesticide expenditure of farm machine users, relative to nonusers.
In addition, farm machine users are more likely to participate in off-farm work relative to their non-user counterparts. Compared with non-users, farm machine users also have larger farm sizes to cultivate crops. On average, the farm machine users are less likely to take risks and have less probability of accessing credit than nonusers. The transportation conditions are more convenient to train/bus stations for farm machine users than for nonusers. Farm machine users are less likely to receive the agricultural subsidies and less satisfied with the extension service provided by the local government compared with nonusers. The farm machine nonusers’ villages have a higher probability of executing the environment improvement project.
Moreover, the number of farm machine users is significantly lower than the number of nonusers in both Gansu and Henan provinces, however, all the samples in Shandong province are farm machine users. Other household and farm level characteristics such as age and education of household head, household size, and extension contact hardly differ between farm machine users and nonusers. However, the findings in Table 2
provide insufficient basis to make inferences about the impact of farm machine use on pesticide expenditure, since the simple comparison of mean differences fails to account for confounding factors such as observed (e.g., age, gender, household size, farm size) and unobserved characteristics (e.g., farmer’s motivation for farm machine use, farmers’ managerial skills).
There is a growing body of literature that shows that farm machines boost farm production and agricultural efficiency, and promote sustainable and conservation agriculture. However, the impact of farm machine use on pesticide expenditure has not been previously analyzed. Whether and to what extent farm machine use influences the pesticide expenditure remains poorly understood. This paper investigated the factors that affect farmers’ decisions to use farm machines, and analyzed the impact of farm machine use on pesticide expenditure. The study used cross-sectional farm household level data of maize farmers collected from a randomly selected sample of 493 households from the Gansu, Henan, and Shandong provinces in 2017. Results from the farm machine use mean differences revealed statistical differences in pesticide expenditure between farm machine users and nonusers. However, the mean differences could not account for the effects of other confounding characteristics, which may provide a misleading conclusion. Given that farmers self-select themselves into farm machine users and nonusers, we employed an endogenous switching model to address the sample selection bias arising from both observed and unobserved factors.
The empirical findings generally showed that farm machine use has a significant impact on pesticide expenditure. In particular, farmers who used farm machines to spray pesticides tend to reduce the pesticide expenditure by 58.63%, while the farmers who did not use farm machines would decrease pesticide expenditure 33.42% if they chose to be farm machine users. These findings suggest that farm machines serve as an environment-friendly technology to reduce pesticide use and enhance sustainable agriculture. On the factors that influence a farmer’s decision to use farm machines, the results show that transportation condition and extension contact exert positive and statistically significant effects on the farm machine use decision.
The findings from this study do have policy implications for sustainable and conservation agriculture development by reducing pesticide expenditure through farm machine use. In particular, the positive and significant effects of transportation condition, extension contact, and off-farm work on farm machine use suggest that farm machine use in rural regions could be enhanced through strategies such as improving the transportation condition, establishing information channels such as extension services, and providing more off-farm work information. Furthermore, with the increasing feminization of agriculture due to the increased propensity for men to migrate to urban areas for better off-farm work opportunities than women, the genitive impact of household head gender on use of farm machines advocates for the incorporation of gender-specific interventions in developing farm mechanization programs.
A limitation of this study is that we have only considered the impacts of farm machine use on the total pesticide expenditure. Such effects may also exist due to the significant discrepancies of different types of pesticides such as herbicides and insecticide, which is a promising area for future studies examine with a disaggregated analysis. In addition, our analysis of the present study only focuses on the maize farmers and data collected from three provinces in China; studies focusing on other crops and other regions or other countries are necessary in order to obtain a better understanding of the heterogeneous impacts of farm machine use on pesticide expenditure in a broad context.