2.3.1. Index Selection

Dependent variable: It is inevitable that the construction land expansion will occupy the paddy field. Preventing the conversion of paddy fields into dry land is of positive significance in protecting food security and the agricultural ecological environment. Therefore, proportion of paddy field area (PPF) was a suitable dependent variable.

Core independent variables: Socioeconomic development has led to farmland loss and labor transfer. The widespread use of nitrogen fertilizer and machinery in agricultural production replaces farmland and labor, respectively. Therefore, agricultural machinery per capita of agricultural employees (AMAE) and the input intensity of nitrogen fertilizer per hectare of farmland (INNF) were introduced as the core independent variables in the model. To explore the nonlinear characteristics of the substitution of agricultural production factors on paddy field change, we introduced the quadratic term of AMAE (AMAE²) and INNF (INNF²) to validate the nonlinear effect in the model.

Control variables: Apart from the core independent variables, the model should also add some control variables to reflect the factors affecting paddy field changes. These control variables should include climate change, agricultural production conditions, farmers’ risk awareness, and socioeconomic development. We selected the average annual temperature (AT) and average annual precipitation (AP) to represent climate change [48,49,50]. In terms of the influence factors of agricultural production conditions, we chose effectively irrigated farmland (PEIF) and land productivity (LP) to represent farmland water conservancy facilities and farmland quality [51,52]. Crop diversity (CD), which means the crop evenness [53,54], was chosen to represent farmers’ risk awareness [55]. In terms of the socioeconomic factors, the proportion of the number of non-agricultural employees (PNAE), proportion of cash crop sown area (PCC), and income inequality (II) were introduced into the model as control variables [56,57]. The independent variables are listed in Table 1 and descriptive statistics covering all variables are shown in Table 2.

Variables and definitions.

Statistical description of the variables.