**1. Introduction**

Promoting sustainable agriculture and meeting the global demand for food is a major challenge for humanity [1,2]. Increasing agricultural total factor productivity (ATFP) is crucial for promoting sustainable agricultural development [3]. Previous research has shown that a key factor in the sustained growth of Chinese agriculture is the increase in ATFP [4–6]. ATFP is the portion of agricultural output that is not explained by the inputs used for production [7]. As ATFP represents the ability of creation to grow under given input conditions, the level of ATFP is an important basis for assessing the sustainability of agriculture [8–10].

The Chinese government places great importance on improving ATFP. In recent years, China has been promoting a policy of building high-standard farmland. High-standard farmland construction is an agricultural land improvement project that aims to make farmland concentrated, flat, high-yielding, and ecologically improved by improving farmland infrastructure [11,12]. Theoretically, high-standard farmland construction can improve

**Citation:** Ye, F.; Wang, L.; Razzaq, A.; Tong, T.; Zhang, Q.; Abbas, A. Policy Impacts of High-Standard Farmland Construction on Agricultural Sustainability: Total Factor Productivity-Based Analysis. *Land* **2023**, *12*, 283. https://doi.org/ 10.3390/land12020283

Academic Editors: Yongsheng Wang, Qi Wen, Dazhuan Ge and Bangbang Zhang

Received: 24 December 2022 Revised: 13 January 2023 Accepted: 16 January 2023 Published: 18 January 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

farmland quality, promote agricultural scale operation, and thus increase ATFP [13,14]. However, in reality, the impact of high-standard farmland construction policy on ATFP is not yet known due to the geographical differences among provinces and the quality of policy implementation. The objective of this paper is to assess the impact of high-standard farmland construction on ATFP using an econometric approach, with the aim of providing a new perspective for sustainable agricultural development.

The method for measuring ATFP is complex. There are two main methods for calculating ATFP: stochastic frontier analysis (SFA) and data envelopment analysis (DEA). SFA is a parametric estimation method that involves setting specific functional forms and probability distributions for random error terms [15,16]. DEA, on the other hand, is a nonparametric estimation method that calculates efficiency by enveloping the production frontier [17]. Many studies use a combination of DEA and the Malmquist index to measure ATFP [18,19]. As ATFP measurement methods have improved, scholars have started to focus on the determinants of ATFP. Improving farmers' human capital enables them to adopt more advanced technology, which significantly improves ATFP [20]. Infrastructure development can improve agricultural production conditions and increase the land strength of cultivated land, contributing to the advancement of ATFP [21,22]. Agricultural subsidies can help farmers alleviate financial constraints in agricultural production and invest more or adopt more advanced production technologies, thereby increasing ATFP [23,24]. Agricultural and institutional reforms, such as the household responsibility system, agricultural taxation reform, and land system reform in China, have also played a significant role in the growth of ATFP in China [25–28].

Previous research focused on the factors that influence ATFP growth, such as human capital, infrastructure development, government investment, and agricultural policy changes [29,30]. Among these, agricultural policy changes and infrastructure development are particularly important factors in influencing ATFP. In China, a high-standard farmland construction policy is a government policy aimed at improving agricultural infrastructure. The Chinese government invested heavily in the construction of high-standard farmland. Previous research focused on the impact of this policy on farmers' income and eco-efficiency [31,32] but neglected its impact on agricultural sustainability. This paper aims to address this gap by using ATFP to assess the impact, heterogeneity, and mechanisms of high-standard farmland construction policy on agricultural sustainability in China.

This study makes four contributions to the literature. First, we examine the causal relationship between land reclamation and ATFP based on the construction of high-standard farmland in China, offering a new perspective on sustainable agricultural development. Second, this paper investigates the heterogeneous effects of policy implementation on ATFP from multiple angles, providing empirical evidence for the promotion and improvement of high-standard farmland construction policies. Third, this study's continuous difference-indifference-based research design effectively addresses the endogeneity problem of policy change, accurately identifying the causal relationship between high-standard farmland policy and ATFP. Fourth, this research provides empirical evidence for developing countries to promote the construction of high-standard farmland and sustainable agricultural development.

### **2. Policy Background**

The construction of high-standard farmland is an important component of land consolidation in China, which aims to promote sustainable agricultural development and ensure food security. According to the document "Standard for Construction of High-Standard Basic Farmland", high-standard farmland is defined as "Basic farmland formed through rural land remediation that is concentrated and contiguous, with supporting facilities, high and stable yields, good ecology, strong disaster resistance, and compatible with modern agricultural production and operation methods." The construction of high-standard farmland in China can be divided into two phases: the exploration phase (1988–2010) and the standardized implementation phase (2011–present).

Before 2011, there were no professional documents specifying the measurement standards and construction requirements for high-standard farmland. During this phase, the main focus of comprehensive land development was on increasing the area of arable land. In 2011, the Chinese government launched the National Land Improvement Plan (2011–2015), which established the construction standards and requirements for highstandard farmland. Local governments also formulated their own guidelines for implementing high-standard farmland based on the national document, marking the start of a standardized period for high-standard farmland in China. In this paper, data from 2013 and 2017 were selected for visual analysis because they are the years of advancement of high standards of construction, and the results are shown in Figure 1.

**Figure 1.** Change in the new area of high-standard farmland.

After 2013, high-standard farmland entered the stage of large-scale standardized construction, and the scope of construction gradually expanded to 31 provinces in China. In 2013, 99,000 hectares of high-standard farmland were constructed, with the top three provinces in terms of construction area being Jilin, Heilongjiang, and Henan, with 92,070 hectares, 71,470 hectares, and 69,470 hectares of high-standard farmland constructed, respectively, according to *China Financial Statistics Yearbook*. By 2017, China had accumulated a total of 46.67 million hectares of high-standard farmland construction, with Shandong, Henan, and Jiangsu accumulating over 3 million hectares of high-standard farmland construction each, according to *China Financial Statistics Yearbook*.

### **3. Model and Data**

#### *3.1. Model*

#### 3.1.1. Total Factor Productivity Measurement Model

In order to accurately measure ATFP in this paper, we adopt the DEA method and combine it with the Super Efficiency model and the global Malmquist index proposed by Pastor and Lovell [33] and Oh [34]. This hybrid approach, known as the EBM–SGM index and proposed by Tone and Tsutsui [35], is used to construct the production frontier. The EBM–SGM index considers both radial and non-radial slack variables and avoids

the defects of linear programming non-solution and non-transmissibility. To calculate the EBM–SGM index, we use the following formula:

$$\begin{aligned} \mathbf{r}^\* &= \min \theta - \boldsymbol{\varrho} \, \sum\_{i=1}^m \frac{\boldsymbol{\omega}\_i \mathbf{s}\_i}{\mathbf{m}\_0} \\ \text{s.t.} &\{\theta \mathbf{m}\_0 - \mathbf{M} \boldsymbol{\rho} - \mathbf{s} = 0; \boldsymbol{\rho} \mathbf{N} \ge \mathbf{n}\_0; \boldsymbol{\rho} \ge 0, \mathbf{s} \ge 0\} \end{aligned} \tag{1}$$

In Equation (1), r∗ represents the production efficiency value, θ represents the radial efficiency value, and ϕ represents the parameter considering both radial and non-radial slack variables. wi is the relative importance of the i factor of production and si is the slack variable of the i factor of production. ρ is the relative weight, and M and N represent the input and output vectors, respectively. m0 and n0 represent the input and output levels under the radial constraint, respectively.

In order to accurately measure ATFP in this study, the EBM Super-Global–Malmquist (EBM–SGM) index is used. This combination effectively avoids issues such as linear programming non-solution and non-transmissibility. The formula for the EBM–SGM index is shown below.

ATFPt,t<sup>+</sup>1 mt , nt ; mt<sup>+</sup>1, nt<sup>+</sup><sup>1</sup> = # 1+D<sup>t</sup> (mt ,nt ) 1+D<sup>t</sup> (mt+1,nt+1) <sup>×</sup> <sup>1</sup>+Dt+<sup>1</sup> (m<sup>t</sup> ,nt ) 1+Dt+<sup>1</sup> (mt+1,nt+1) \$ 1 2 <sup>=</sup> <sup>1</sup>+Dt (m<sup>t</sup> ,nt ) 1+Dt (mt+1,nt+1) × # 1+Dt<sup>+</sup><sup>1</sup> (m<sup>t</sup> ,nt ) 1+Dt (mt,nt) <sup>×</sup> <sup>1</sup>+Dt<sup>+</sup><sup>1</sup> (mt<sup>+</sup>1,nt<sup>+</sup>1) 1+Dt (mt+1,nt+1) \$ 1 2 = TE mt<sup>+</sup>1, nt<sup>+</sup>1; mt , nt × TC mt<sup>+</sup>1, nt<sup>+</sup>1; mt , nt (2)

In Equation (2), Dt and Dt+1 denote the set of production technologies in periods t and t + 1, respectively. Referring to Färe et al. [36], ATFP can be decomposed into agricultural technical change (TC) and agricultural technical efficiency (TE).

#### 3.1.2. Model of the Impact of High-Standard Farmland Construction Policy on ATFP

Referring to the existing literature [37], this paper uses a continuous DID model to estimate the effect of the high-standard farmland policy on ATFP. The following model is constructed in this paper based on this analysis.

$$\text{LNATFP}\_{\text{i,t}} = \alpha\_1 + \alpha\_2 \text{treated}\_{\text{i}} \times \text{time}\_{\text{t}} + \beta \text{X}\_{\text{i,t}} + \eta\_{\text{t}} + \gamma\_{\text{i}} + \mu\_{\text{i,t}} \tag{3}$$

In Equation (3), i stands for region and t stands for year. ATFP stands for agricultural total factor productivity, treated stands for proportion of high-standard farmland. timet stands for the dummy variable at the time of policy. When t ≥ 2011, time is 1; otherwise, it is 0. X stands for control variable, γ<sup>i</sup> and η<sup>t</sup> stand for year effect and region effect, respectively. μi,t stands for classical random disturbance term. α and β stand for parameters to be estimated. In particular, it is important to note that α<sup>2</sup> is the core estimated parameter in this paper, representing the net effect of high-standard farmland construction policy on ATFP.

#### 3.1.3. Parallel Trend Test

The parallel trend assumption is a crucial prerequisite for DID estimation. Based on previous research [38], the following model is constructed in this paper to test the parallel trend assumption.

$$\text{LNATFP}\_{\text{i,t}} = \sum\_{\mathbf{k}=2008}^{2017} \beta\_{\mathbf{k}} \text{treated}\_{\mathbf{i}} \times \mathbf{d}\_{\mathbf{t}} + \beta \mathbf{X}\_{\mathbf{i,t}} + \eta\_{\mathbf{t}} + \gamma\_{\mathbf{i}} + \mu\_{\mathbf{i,t}} \tag{4}$$

Equation (4) shows this model, where time represents the year dummy variable, treated represents the area of high-standard farmland construction, and other variables and coefficients are set consistently with Equation (3). We can determine whether the parallel trend assumption holds by examining the statistical significance of the estimated parameters of the interaction term. If the estimated parameters of the interaction term are statistically insignificant before 2011, we can assume that the parallel trend assumption is valid.

#### 3.1.4. Impact Mechanism Model

This paper will study the impact of high-standard farmland construction policies on total factor productivity (ATFP) by decomposing ATFP into technical change (TC) and technical efficiency (TE). This approach was used in previous studies [39,40]. The impacts on TC and TE will be analyzed separately to understand the mechanism behind the effects of these policies. Two models will be used for this analysis:

$$\text{LNTC}\_{\text{i,t}} = \alpha\_1 + \alpha\_2 \text{treated}\_{\text{i}} \times \text{time}\_{\text{t}} + \beta \text{X}\_{\text{i,t}} + \eta\_{\text{t}} + \gamma\_{\text{i}} + \mu\_{\text{i,t}} \tag{5}$$

$$\text{LNTE}\_{\text{i,t}} = \alpha\_1 + \alpha\_2 \text{treated}\_{\text{i}} \times \text{time}\_{\text{t}} + \beta \text{X}\_{\text{i,t}} + \eta\_{\text{t}} + \gamma\_{\text{i}} + \mu\_{\text{i,t}} \tag{6}$$

The estimated parameters of the treatedi × timet variable in Equations (5) and (6) represent the effects of high-standard farmland construction policy on TC and TE, respectively.

#### *3.2. Variables*

#### 3.2.1. Explained Variables

In this study, ATFP is the main explained variable. To calculate ATFP, we use the EBM–SGM method, which requires the selection of appropriate input and output variables. According to existing research [41–43], the input variables chosen in this study include: (1) the combined agricultural sown area and aquaculture area, (2) the number of people employed in the primary industry at the end of the year, (3) the total power of agricultural machinery, (4) the amount of fertilizer, and (5) the effective irrigation area in agriculture. These variables represent land, labor, machinery, fertilizer, and irrigation inputs in agricultural production. The total agricultural output value is used as the agricultural output indicator in this study.
