1.1 Theoretical analysis and research hypothesis
According to the theory of farmers' economic behavior, some seemingly "irrational" behavior of small farmers is a "rational" performance under the constraints of external conditions. Therefore, compared with the debate between "rationality" and "irrationality," it seems more appropriate to describe farmers' economic behavior as both "market rationality" and "survival rationality (Yang et al., 2015). On the one hand, the introduction of automated equipment by farmers follows the principle of market rationality. On the other hand, the farming entities introduce automation equipment according to the principle of utility maximization, i.e., to save costs and improve farming efficiency by adding automation facilities, thus achieving the goal of improving farming income. In short, public emergencies can affect the configuration of automated facilities through direct and indirect shocks.
(1) Direct impact
In the case of the ASF epidemic, for example, the ASF virus can be carried by the human body and infect pigs on the farm2, which in turn leads to the death of pigs in the farm stock. To reduce the risk of pig infection, farming entities will minimize the frequency of internal and external personnel contact with pigs to further ensure the stability of farming revenue. For this reason, farming entities have to alleviate the problem of front-line farming personnel not being able to participate in pig farming properly utilizing additional automation facilities. Improving the level of farming automation can reduce the risk of pig infection and mortality while ensuring the normal implementation of daily farming activities. Accordingly, hypothesis 1 is proposed.
H1: The direct shock generated by the unexpected event will force the farming body to improve the level of automation
(2) Indirect shocks
In the case of the COVID-19 epidemic, for example, the COVID-19 epidemic mainly affects front-line breeders. To minimize the negative impact of the spread of the epidemic, the local government has imposed stricter restrictions on personnel, material transfer ,transportation, etc. (Zhu et al.,2020; Luo et al.,2020).In the case of pig farming, the COVID-19 epidemic not only affects the feed distribution and outbound transportation of pig sales, resulting in repeated cases of delayed allocation of production materials; at the same time, front-line farm personnel in the farm are unable to participate in their daily work such as farming normally because of the infection. Therefore farmers may allocate additional automated facilities to minimize the uncertain impact of shocks such as COVID-19 epidemics on farming activities. Accordingly, hypothesis 2 is proposed.
H2: Indirect shocks generated by unexpected events will force the farming body to improve the level of automation
With the increasing trend of large-scale farming, the differences in economic strength and farming level of different scale farming subjects are becoming more and more obvious (Wang et al.,2019). Compared with non-scale farming entities, scale farming entities have more comparative advantages in coping with the shock of public emergencies and the configuration of automated facilities (Xiao et al.,2021). Hypothesis 3 is proposed accordingly.
H3: There is a difference in the impact of public emergencies on different scale farming subjects to enhance the level of automation
1.2 Data sources, variable selection and research methods
1.2.1 Data sources
The data used in this paper were obtained from field research conducted by the subject team from May to August 2022 on livestock farmers in seven cities in Hebei Province: Xingtai, Shijiazhuang, Handan, Baoding, Hengshui, Tangshan ,and Qinhuangdao. Firstly, the relevant local government departments organized training on pig breeding and resource utilization for farming subjects, followed by on-site research. Before the start of the survey, the person in charge of the research team explained the purpose of the survey, the basic content of the survey ,and the precautions for filling it out. A total of 309 questionnaires were obtained, and after excluding those with incomplete information and inconsistent answers, a total of 259 valid questionnaires were obtained, with an efficiency rate of 83.82%.
From the basic situation of the investigated farming subjects (as shown in Table 1), the proportion of male farming subjects is 89.19% and the proportion of females is 10.81%; the actual age is concentrated between 35–55 years old, accounting for 66.02% of the total sample; the education is mainly junior high school, accounting for 42.47%; 53.28% of the farming subjects belong to the risk conservative preference. The majority of farming subjects with an average annual slaughter of 101–500 head, accounting for 43.63% of the total sample.
Table 1
Basic information of breeding subjects
Indicators | Category | Frequency | Frequency % |
Gender | Male | 231 | 89.19 |
| Female | 28 | 10.81 |
Age | 35 years old and below | 52 | 20.08 |
| 35–45 years old (including 45 years old) | 85 | 32.82 |
| 45–55 years old (including 55 years old) | 86 | 33.20 |
| 55 years old and above | 36 | 13.90 |
Education level | Primary School | 27 | 10.42 |
| Junior High School | 110 | 42.47 |
| High School | 70 | 27.03 |
| College and above | 52 | 20.08 |
Non-farming experience (has been or is currently engaged in one or more occupations other than that of a farmer) | Never had | 105 | 40.54 |
There has been 1 kind of | 126 | 48.65 |
There have been 2 kinds of | 24 | 9.27 |
There have been 3 kinds of | 2 | 0.77 |
There have been 4 kinds of | 2 | 0.77 |
Risk Awareness | Risk Appetite | 87 | 33.59 |
| Risk Conservative | 138 | 53.28 |
| Risk Neutral | 34 | 13.13 |
Average annual slaughter | 100 heads and below | 42 | 16.22 |
| 101–500 heads | 113 | 43.63 |
| 501–2000 heads | 66 | 25.48 |
| 2001–3000 heads | 11 | 4.25 |
| More than 3000 heads | 27 | 10.42 |
1.2.2 Variable selection
This paper focuses on the impact of public emergencies on the level of automated facilities deployed by farming entities in the context of ASF and COVID-19. Therefore, the explanatory variables in this paper are "the number of automated facilities configured by farming subjects in the farming process after the occurrence of the double epidemic" and "the impact of public emergencies on farming subjects". The variables are set as follows.
(1) Explained variables
During the research, the group found that the automated facilities configured by the breeding subjects mainly include automatic feeding devices, automatic watering devices, automatic manure cleaning devices, intelligent environmental control devices,and other automated devices. Therefore, this paper sets the following question in the research: "After the occurrence of African swine fever and the COVID-19 epidemic, which automation facilities have you configured in the breeding process3 (multiple choice)", and the number of automation facilities configured by the breeding subjects in the breeding process is used as a measure of the level of automation facilities configured by the breeding subjects (hereinafter referred to as "automation level"), and the automation level will be recorded as 1 if any additional automation facility is allocated.
-
(2) Core explanatory variables
-
In this paper, the impact of public emergencies is carefully considered in terms of both direct and indirect shocks, taking into account the actual impact of the epidemic. In measuring the direct impact of the outbreak, six indicators are selected: "reduction in farming scale, decrease in pig production, increase in pig inventory, increase in farming cost, shortage of workers, and restriction in liquidity"; in measuring the indirect impact of the outbreak, six indicators are selected: "difficulty in capital turnover, high environmental protection pressure, land policy constraints, difficulty in selling pigs, unstable pork price, and difficulty in recruiting suitable workers. In measuring the indirect impact of emergencies, seven indicators were selected: "difficulty in capital turnover, environmental protection pressure, land policy constraints, difficulties in selling pigs, unstable pork prices, difficulty in recruiting suitable technicians and veterinarians". All the indicators were measured on a 5-point Richter scale, with values from 1 to 5 indicating that the degree of impact increased step by step.
Considering the reality that there are too many indicators and the values are too scattered, which affects the accuracy of the results, this paper combined with (Huang et al., 2020). We use the arithmetic average method to reduce the dimensionality of 13 indicators of direct and indirect shocks of public emergencies, and finally, obtain two indicators of the direct shock of emergencies (MEDA) and indirect shock of new emergencies (MEIA).
-
(3) Control variables
-
To alleviate the problem of omitted variable bias, the individual characteristics and farming characteristics variables of farming subjects are controlled in the model in this paper. Among them, individual characteristics mainly include age, education level, non-farming experience and risk awareness (Zhang et al., 2022) The farming characteristics mainly include the number of years of farming, average annual slaughter, number of farming training, number of farming participants, number of employees, number of farming information access channels, market risk expectation and difficulty in accessing external support (Kong et al., 2016).
-
The definition and descriptive statistics of each variable are shown in Table 2.
Table 2
Descriptive statistical analysis of variables
Variable Name | Variable abbreviations | Sample size | Average value | Standard deviation | Minimum value | Maximum value |
Automation level | ZDH | 259 | 1.838 | 1.173 | 0 | 5 |
Direct impact | MEDA | 259 | 3.118 | 0.849 | 1 | 5 |
Indirect shock | PXCS | 259 | 2.197 | 0.896 | 1 | 4 |
Age | NL | 259 | 44.838 | 9.527 | 20 | 74 |
Education level | EDU | 259 | 2.568 | 0.927 | 1 | 4 |
Non-farm experience | FNJL | 259 | 0.726 | 0.725 | 0 | 4 |
Risk Awareness | FXYS | 259 | 1.795 | 0.653 | 1 | 3 |
Years of breeding | YZNX | 259 | 12.247 | 7.650 | 1 | 43 |
Average annual slaughter volume | CL | 259 | 2.490 | 1.136 | 1 | 5 |
Number of farming training | MEIA | 259 | 2.984 | 0.423 | 2 | 4 |
Number of farming participants | YCRS | 259 | 2.710 | 4.974 | 0 | 80 |
Number of employees | GYRS | 259 | 3.077 | 8.280 | 0 | 80 |
Number of access channels to breeding information | WGM | 259 | 2.286 | 1.209 | 1 | 8 |
Market risk expectations | FXYQ | 259 | 3.610 | 0.727 | 1 | 5 |
Difficulty in obtaining external support | HQND | 259 | 3.500 | 1.055 | 1 | 5 |
1.2.3 Research Methodology
The number of automated facilities can characterize the level of automation of the farming body, which is an ordered multi-categorical variable. Combined with the existing research results (Chang et al.,2020; Ou et al.,2022) and the characteristics of the selected variables, this paper chooses to construct an Ordered Logit model with the following basic form.
$${y}_{i}^{\ast }={\beta }_{i}{X}_{i}+{C}_{i}+{\epsilon }_{i}$$
1
Where \({y}_{i}^{\ast }\) is the first\(\text{i}\) latent variable for the level of automation of the first farming subject.\({X}_{i}\) is the latent variable of ASF and COVID-19 on the first\(\text{i}\) actual effect of the first farming subject, and\({C}_{i}\) is the control variable affecting the level of automation of the first farming subject.\({\epsilon }_{i}\) is the random error term. Further, the selection criteria for the explained variable (level of automation) are.
$${y}_{i}=\left\{\begin{array}{c}0, {y}_{i}^{\ast }<{\delta }_{1}\\ 1, {\delta }_{1}\le {y}_{i}^{\ast }<{\delta }_{2}\\ ...\\ 4, {\delta }_{4}\le {y}_{i}^{\ast }<{\delta }_{5}\\ 5, {y}_{i}^{\ast }\ge {\delta }_{5}\end{array} \right.$$
2
Where \({y}_{i}\) is the\(\text{i}\) observable value of the automation level of the farming subject, assigned as 0–5, indicating that the farming subject has configured 0–5 automation facilities in the farming process after the outbreak, respectively.\({\delta }_{1}\) -\({\delta }_{5}\) is the cut point, which is obtained from the model estimation and expressed as a cut in the regression results.
[2] What is the transmission of African swine fever through? , Jingzhou City Bureau of Agriculture and Rural Affairs, http://nyj.jingzhou.gov.cn/ywbk/cmy/202008/t20200814_512371.shtml
[3] The options mainly include: automatic feeding, automatic watering, automatic manure cleaning, intelligent environmental control and other automated configurations.