The Effect of High-Standard Farmland Construction Policy on Grain Harvest Losses in China

Abstract

The United Nations included reducing harvest losses as a Sustainable Development Goal in 2015, sparking heightened research and policymaker interest in reducing losses to ensure food security. High-standard farmland construction plays a crucial role in ensuring food security. Few studies have combined high-standard farmland construction with grain harvest losses. Drawing on the data from the 2022 Chinese Post-Harvest Loss Survey (CPHLS 2022), the study utilizes OLS (ordinary least square) and quantile regression models to explore the impact of high-standard farmland construction on grain harvest losses. Empirical results show that high-standard farmland construction can significantly reduce grain harvest losses. The research conclusions are still valid after passing a series of robustness tests. The heterogeneity analysis shows that high-standard farmland construction significantly impacts on grain harvest losses for farmers in major grain-producing areas, plain areas, and eastern regions. Mechanism analysis reveals that high-standard farmland construction mainly reduces grain harvest losses by expanding operational scale and enhancing mechanization application. Based on research findings, the Chinese government should formulate a targeted high-standard farmland construction policy, optimize the agricultural machinery operating environment, and promote appropriate operational scale to ensure national food security.

1. Introduction

Most scholars agree that reducing grain harvest losses is an important strategy to achieve food security in the future [1,2]. The Food and Agriculture Organization of the United Nations estimates that 13.8 percent of food produced yearly is lost before reaching the retail stage [3,4]. Harvest losses account for significant food losses, especially in developing countries with limited arable land and underdeveloped technological equipment. As the largest developing country, China experiences annual harvest losses as high as 121.7 million tons, with severe losses occurring during the harvesting stage, accounting for 19.67% of the total post-harvest losses [5]. These losses reflect the pointless consumption of invested resources. Reducing postharvest losses offers an important pathway for availing food, saving agricultural production resources, and improving the ecological environment [6,7].

The primary factors influencing grain harvest include operational scale and mechanization in China [8]. Family smallholder production represents the predominant livelihood for Chinese farmers. Given the abundance of small-scale farmers in China, each household possesses limited cultivated land, prohibiting large-scale operations. The fragmentation of farmland not only constrains operational scale and escalates production costs [9,10] but also diminishes machinery utilization due to extensive field distances [11,12], resulting in slow harvesting speeds and susceptibility to adverse weather conditions, leading to substantial harvest losses [13]. In response to the pressing concerns surrounding arable land management, China has strategically launched the “National High-Standard Farmland Construction Plan (2021–2030)”. This comprehensive initiative is designed to transform arable land through a series of targeted improvements, resulting in land that is not only flat, concentrated, and contiguous but also equipped with state-of-the-art irrigation and drainage systems, well-maintained field roads, and robust electrical infrastructure. The land development further encompasses the enhancement of soil fertility and the integration of ecologically sound practices, thereby endowing the land with a strong resilience to natural disasters and the flexibility to adapt to the demands of modern agricultural production and management techniques. High-Standard Farmland Construction is used to build public infrastructure on arable land that growers cannot afford, involving improvements in land, soil, water conservancy, roads, forestry networks, electricity, technology, and management. The implementation of the policy means that land can be improved from low-efficiency to high-efficiency, improving the allocation of land elements and production efficiency and safeguarding food security by reducing grain harvest losses.

Several studies have examined high-standard farmland construction policy and conducted extensive analyses on the impact on grain production, mainly focusing on grain yield and planting structure [14,15,16,17]. However, studies have yet to explicitly explore the impact of high-standard farmland construction policy on grain harvest losses and the underlying mechanisms associated with it. In this study, survey data collected from 3009 maize farming households in rural districts in China is used to analyze the impact of high-standard farmland construction on grain harvest losses. The study finds that high-standard farmland construction can significantly reduce grain harvest losses, primarily through increasing operational scale and mechanization application. Heterogeneous results indicate that the policy’s reduction in grain harvest losses is particularly pronounced in major grain-producing areas, plains, and eastern regions.

This study contributes to the literature in several ways. First, to the best of our knowledge, this is the first study to focus on grain harvest losses caused by high-standard farmland construction in China using nationally representative micro-level data. This analysis is important because substantial resources have been channeled toward increasing agricultural output to meet China’s growing food demand, limited resources restrict the expansion of grain production, and reducing post-harvest losses is critical to food security. Second, this study adds to the existing literature on policy responses to grain harvest losses by evaluating operational scale and mechanization application. Furthermore, considering the differences between agricultural production locations and terrains, the heterogeneous impact of high-standard farmland construction policy is studied. Third, this study utilizes quantile regression to examine the effects of high-standard farmland construction on grain harvest. Furthermore, we conducted a series of robustness tests to strengthen the reliability of our conclusions.

The rest of the paper is organized as follows. Section 2 summarizes the mechanism and research hypotheses. Section 3 introduces the research method. Section 4 describes variables and data sources. Section 5 presents the main results and includes a variety of robustness checks. Section 6 discusses the potential mechanisms and heterogeneity analysis. Section 7 concludes and makes recommendations.

2. Mechanism and Research Hypotheses

In economic theory, productivity is driven by two key elements: land and technological progress. For food production, arable land forms the foundation, while technological progress is the primary catalyst for reducing food losses. High-standard farmland construction integrates arable land with technology, optimizing the production factors of arable land and placing significant emphasis on technological advancements. This article argues that high-standard farmland construction enhances the efficiency of arable land and technology, thereby reducing food losses through increased operational scale and mechanization, as shown in Figure 1.

2.1. High-Standard Farmland Construction to Expand Operational Scale

High-standard farmland refers to basic farmland formed through rural land consolidation within a certain period that is concentrated and contiguous, having perfect irrigation and drainage facilities, smooth inter-field roads, fertile soil, ecological friendliness, strong resistance to disasters, and adaptability to modern agricultural production and management methods. High-standard farmland construction solves the problems of decentralization and cultivated land fragmentation by merging small fields with large fields, expanding farmland management to an appropriate scale, leveraging the comparative advantages of large-scale farmers [18,19,20], and selecting scientific guidance for harvest management based on the optimal timing and methods to tap the potential for reducing grain harvest losses. High-standard farmland policy improves farmland quality through soil improvement and improvement of supporting facilities, reduces factor input costs, improves land output efficiency, and promotes agricultural operators to transfer to farmland to develop moderate-scale operations. From scale economies theory, small scales often lead to inefficiency in a perfectly competitive market economy, and scale expansion leads to scale effects [21]. Large-scale farmers have more information acquisition capabilities and rich agricultural production experience than small-scale farmers. The Scale operations exert three impacts on grain harvest losses: First, large-scale farmers have a significant demand for agricultural social services, making them attractive to service providers. These farmers can obtain timely services such as grain harvesting during the optimal period, which shortens operation time, ensures complete grain collection, and minimizes harvest losses [22,23]. Second, the development of large-scale operations has promoted market specialization and the growth of local social services. This better meets local farmers’ needs for advanced equipment and high-quality agricultural machinery suited to local crop harvesting. Consequently, it improves the compatibility between harvesting equipment and the local crop’s planting density and growth status [24,25], enhances harvesting quality, and reduces issues like exposed grain and grain damage due to equipment incompatibility or machinery failure. Third, the income loss caused by harvesting losses is typically greater for large-scale farmers, prompting them to supervise and guide grain harvesting operations closely. They standardize the operational processes of harvester drivers, ensuring operators maintain a reasonable operation range, speed, optimal feeding rate, and appropriate header position [26,27]. This effectively improves the quality of harvesting operations and reduces problems such as unclean threshing or entrained grains resulting from poor operational quality.

2.2. High-Standard Farmland Construction by Promoting Mechanization Application

Mechanization is essential to reduce harvest losses and ensure food security, but fragmented land limits its effectiveness [28,29]. The main goals of high-standard farmland construction are to facilitate its transformation into mechanized farmland. This involves restructuring fragmented cultivated land patterns to create optimal working conditions for agricultural machinery and promote mechanization. Initially, high-standard farmland enhances road accessibility through the construction of mechanized farming roads [30,31]. Building upon this foundation, road widths are expanded appropriately, load capacities and road accessibility are enhanced, and barriers to machinery entering fields for operations are minimized. This ensures timely grain harvesting during maturity and reduces grain harvest losses caused by delays in farming operations. Moussa [32] showed that compared with manual harvesting, combine harvesters can save 94 h per hectare and shorten the harvest season by 4 days or more [33,34]. Mechanized harvesting can therefore reduce harvest losses from exposure to catastrophic weather risks.

Secondly, high-standard farmland involves construction measures like field merging to enable continuous operations, cut down on round trips and idle time during mechanical activities, enhance the working environment for machinery in the field, and reduce grain harvest losses. Mechanical operations need ample space for reciprocating and turning movements, but narrow farmland limits movement speed and mechanical efficiency. High-standard farmland construction achieves continuous operations through field merging and similar strategies, cutting round trips and idle time and allowing agricultural machinery to operate more smoothly. This approach prevents issues like frequent direction changes and misalignment with crop rows that occur when turning in small plots, thereby minimizing additional crop damage and grain losses [35].

Last, high-standard farmland construction involves increasing soil coverage thickness on field surfaces to adjust flatness, meeting machinery operation standards. This allows agricultural machinery to operate smoothly, enhancing efficiency and quality, ensuring accurate crop harvesting, and reducing grain losses due to operational deficiencies or incomplete harvesting.

Building upon the above analysis, the following hypothesis is proposed:

Hypothesis 1 (H1).

The implementation of high-standard farmland construction policy can reduce grain harvest losses.

Hypothesis 2 (H2).

The high-standard farmland construction policy may reduce grain harvest losses by increasing operational scale.

Hypothesis 3 (H3).

The high-standard farmland construction policy may reduce grain harvest losses by promoting mechanization application.

3. Research Method

3.1. Identification Strategy

3.1.1. Baseline Regression Models

In order to identify the impact of the high-standard farmland construction policy on grain harvest losses, the paper constructs the following baseline regression model:

$${Loss}_i=\alpha +\beta HL{rate}_i+\gamma X_i+{\epsilon }_i$$

(1)

where Loss_i represents grain harvest losses of maize farming household i. HLrate_i represents the core explanatory variable, which measures the impact of high-standard farmland policy for maize farming household i. X_i is a series of control variables that include gender, age, education, household income, labor number, crop yield, harvest weather, harvest maturity, rush to plant, harvest machinery, work attitude, harvest terrain, service center, hire harvesters, and region. The variable α is a constant term. The paper focuses on coefficient β of the core explanatory variable, namely the net effect of the high-standard farmland construction policy on the grain harvest losses. The variable γ is the coefficient to be estimated. ε_i is the random error term.

3.1.2. Quantile Regression Models

To more comprehensively test the causal relationship, this study further applies quantile regression models to estimate the effect of high-standard farmland construction on grain harvest losses. Compared with ordinary least squares regression, quantile regression provides comprehensive data analysis by estimating the conditional distribution of the target variable at different quantiles. The key to quantile regression is to express the solution as a minimization problem [36]. This model can be specified as follows:

$$Y_i=F(y)=Prob(Y_i<y)$$

(2)

where Y_i represents maize farming households’ grain harvest losses. F(y) is the probability distribution function of maize farming households’ grain harvest losses.

$$Q(\tau )=inf\{y:F(y)\ge \tau \}$$

(3)

where Q(τ) is the minimum y value of τ. The condition means of grain harvest losses is a linear function. The expected value of grain harvest losses is as follows:

$$E(Y_i|X_i=x)=E(L_i)=E(F(X_i,u_i))=X_i^{\prime }\beta$$

(4)

where X_i is the explanatory vector of grain harvest losses. β is the regression coefficient vector. When regressing the sample data of quantile point τ, the goal is to find the solution to the minimum problem of quantile point of τ, which is expressed as follows:

$$min\left\{{\displaystyle \sum Y_i\ge X_i^{\prime }\beta \tau |}{Y}_i-X_i^{\prime }\beta |+{\displaystyle \sum Y_i<X_i^{\prime }\beta (1-\tau )|}{Y}_i-X_i^{\prime }\beta |\right\}$$

(5)

Solving Formula (5) to obtain parameter estimates is expressed as follows:

$${\widehat{\beta }}_{\tau }=argmin\{{\displaystyle \sum Y_i\ge X_i^{\prime }\beta \tau |}{Y}_i-X_i^{\prime }\beta |+{\displaystyle \sum Y_i<X_i^{\prime }\beta (1-\tau )|}{Y}_i-X_i^{\prime }\beta \}$$

(6)

where ${\widehat{\beta }}_{\tau }$ is the estimated parameter, which indicates the degree of impact on grain harvest losses at the τ quantile point. Therefore, the distribution trajectory of Y_i under the corresponding conditional distribution can be described at different quantiles. In this study, grain harvest losses are used as explanatory variables, and τ is set to three representative quantile levels of 0.1, 0.5, and 0.9.

3.1.3. Mechanism Validation Model

To explore the internal mechanism of how the high-standard farmland construction policy affects grain harvest losses, this study constructs the following mechanism verification model based on that proposed by [37]. First, based on the mechanism analysis, the transmission mechanism variables of operational scale and mechanization application are introduced to test the impact of high-standard farmland construction on grain harvest losses. Secondly, the impact of operation scale and mechanization application on grain harvest losses is further examined. The specific regression model is as follows.

The first stage introduces operational scale and mechanization application to test the high-standard farmland construction policy on the verified mechanism variables.

$$M_i=\alpha +\beta HL{rate}_i+\gamma X_i+{\epsilon }_i$$

(7)

where M_i denotes the mechanism variables, which include operational scale and mechanization application. The rest of the variables are set as in Equation (1).

The second stage is to verify the impact of operational scale and mechanization application on the grain harvest losses.

$${Loss}_i=\alpha +\beta HL{rate}_i+\gamma X_i+\lambda M_i+{\epsilon }_i$$

(8)

In Equation (8), λ is the indirect effects of the high-standard farmland construction policy on the grain harvest losses. The other variables are consistent with Equation (1).

4. Variables and Data Sources

4.1. Selection of Variables

4.1.1. Explained Variable

Our explanatory variable is the grain harvest losses rate. The grain harvest losses may be quantitative or qualitative. This study measures grain harvest losses as actual physical losses because qualitative losses are challenging to measure [38], and there is limited awareness to differentiate corn grades before entering formal markets [39]. Referring to Qu Xue’s research [40], the harvest process is defined as the four links from grain harvesting, threshing, and field transportation to cleaning. Harvest loss (%) = total loss/(loss + output), the harvest losses (%) is the arithmetic average of the harvest loss rate of all households.

4.1.2. Core Explanatory Variable

In this paper, the core explanatory variable is the high-standard farmland construction policy. Considering the differences in cultivated land area in different regions, referring to existing literature, it is represented by the proportion of high-standard farmland construction area in the county’s total cultivated land area [17]. In addition, this study also uses comprehensive agricultural investment as a proxy variable for high-standard farmland construction as a robustness test.

4.1.3. Control Variables

Apart from the impact of the high-standard farmland construction policy on grain harvest losses, other factors that also contribute to these losses and therefore should be controlled for in the model. Referring to the research of existing research and considering the characteristics of data availability, this study selects the following control variables: gender, age, education, household income, labor number, harvest weather, harvest maturity, rush to plant, harvest machinery, work attitude, harvest terrain, service center, hire harvesters, and region.

4.2. Data Sources

Our analysis combines two major sources of data. The first source is the 2022 Chinese Post-Harvest Loss Survey (CPHLS), conducted by the Price and Cost Center of the China Development and Reform Commission. CPHLS is a cross-sectional data that mainly investigates the losses of grain crops such as corn, rice, and wheat during the harvest and storage stages, covering 4549 farming household in 26 provinces (municipalities and autonomous zones). The survey covers many aspects such as farmers’ household characteristics, grain production characteristics, and village development. To ensure the quality of collected information, investigators were given intensive training before the survey. Sample households were selected based on the following procedure. First, we allocated sample households to various provinces (municipalities and autonomous zones) according to the sample household’s grain production. Second, in each province, we randomly selected two counties from the top ten counties in terms of one crop production in the province. Third, we randomly selected two towns in each county and two villages in each town. Finally, 10 sample farmers were randomly selected in each village.

China is the world’s second largest crop producer, so this study focuses on the harvest losses of Chinese corn growers. This paper excluded non-maize-farming households and finally collected 3009 samples (

Table 1. Maize sample distribution.

Region	Province	Samples
Northern Plain	Liaoning, Jilin, Heilongjiang, Inner Mongolia	721
Huanghuaihai	Tianjin, Hebei, Shanxi, Shandong, Henan	1012
Southwest Mountain	Sichuan, Yunnan, Guizhou	515
Northwest	Shanxi, Gansu, Ningxia, Xinjiang, Qinghai	457
Southern	Guangdong, Fujian, Zhejiang, Jiangxi, Jiangsu, Anhui, Guangxi, Hunan, Hubei	304
Total		3009

Data source: The 2022 Chinese Post-Harvest Loss Survey (CPHLS).

). In China, corn planting area accounts for about 43% of the annual grain planting area and contributes about 40% to the total grain output. Corn harvest losses lead to significant wastage of scarce resources. Among China’s three major staple foods (corn, rice, and wheat), corn has the highest loss rate during the harvest process. Therefore, it is of great significance to study ways to reduce losses in corn harvests. Corn harvest data were collected from 26 provinces (municipalities and autonomous zones) (Liaoning, Jilin, Heilongjiang, Inner Mongolia, Tianjin, Hebei, Shanxi, Shandong, Henan Sichuan, Yunnan, Guizhou, Shanxi, Gansu, Ningxia, Xinjiang, Qinghai, Guangdong, Fujian, Zhejiang, Jiangxi, Jiangsu, Anhui, Guangxi, Hunan, and Hubei) in China (Table 1). The survey covered China’s five major corn-producing areas (the Northern Plain, the Huanghuaihai, the Southwestern Mountain, the Northwest, and the Southern).

The data on high-standard farmland construction originates from official reports that have been released. As the Ministry of Agriculture and Rural Affairs has not yet disclosed high-standard farmland construction areas in each county, this study collects and organizes data up to the end of 2021 from the 2022 government work reports of various counties. When specific area data is not available in the government work reports, it can be obtained through local statistical yearbooks, telephone consultations, and other methods. If these sources confirm that the high-standard farmland construction area in a county or city was zero by the end of 2021 or that construction only began after 2022, the area is classified as having no high-standard farmland.

Descriptive statistical analysis of the variables is shown in

Table 2. Variables definition and descriptive statistics.

Variables Name	Variables Definition	Mean	Std. Dev.
Harvest losses	Maize harvest average harvest losses (%) 1	3.036	0.128
HSFC	High-standard farmland construction area (mu 2)/cultivated area (mu 2)	0.327	0.179
Gender	Gender of family decision maker (male = 1)	0.609	0.488
Age	Age of family decision maker (years)	53.774	7.422
Education	Years of schooling of family decision maker (years)	9.459	1.547
Household income	Annual net income of the family in 2021 (CNY ten thousand 3)	7.526	4.847
Labor number	Number of family agricultural labor (person)	2.451	0.959
Harvest weather	Whether there was bad weather during harvest (yes = 1)	0.424	0.494
Harvest maturity	Was there too maturity during harvest (yes = 1)	0.092	0.289
Rush to plant	Whether to rush to plant for the next season (yes = 1)	0.073	0.261
Harvest machinery	Harvest using machinery (advanced = 1)	0.257	0.437
Work attitude	Whether the harvest driver is proficient in operation (proficient = 1)	0.100	0.300
Harvest terrain	Terrain characteristics during harvest (hills = 1)	0.195	0.396
	Terrain characteristics during harvest (mountains = 1)	0.234	0.423
Service center	Is there a grain post-harvest service center near the village (yes = 1)	0.292	0.454
Hire harvesters	Whether the village can hire harvesters (yes = 1)	0.719	0.449
Region	The region where the farmer belongs (Central = 1)	0.292	0.455
	The region where the farmer belongs (West = 1)	0.408	0.492
Mechanism variables
Operational scale	The household grain cultivation area in 2022 (mu 2)	17.978	7.084
Mechanization application	During harvesting, mechanization application: full mechanical combined harvesting = 1; semi-mechanized harvesting = 0 4	0.429	0.495

Data source: The 2022 Chinese Post-Harvest Loss Survey (CPHLS); Notes: ¹ average harvest loss (%) = total loss/(loss + output), the harvest loss (%) is the arithmetic average of the harvest loss rate of all households. ² mu is the unit of area in China; 1 ha = 15 mu. ³ The yuan (CNY) is the Chinese currency unit, USD 1 = CNY 7.25 (April 2024). ⁴ Semi-mechanized harvesting refers to a harvesting process where some tasks are carried out by machinery while others still require manual labor. The same applies below. The sample observations are 3009.

. The maize harvest average losses are 3.036%, with an average yield of 536.0804 kg per mu. This translated to an average loss of approximately 16.785 kg per mu. In 2021, China’s corn production totaled 272.5506 million tons, resulting in a loss of 827.46 million tons. The proportion of high-standard farmland construction is 32.7%, and the difference is related to the regional fiscal revenue gap, which is consistent with Peng’s research conclusion [41]. The sampled maize farmers are mainly male-dominated (60.9%), with an average age of 53 years old. Each maize farming household typically has two agricultural laborers. The average education level among household heads is junior high school. A total of 42.4% of farmers encountered terrible weather such as strong winds during harvest. Most of the surveyed are located in plains and hills regions. In total, 71.9% of villages find it easy to hire machinery for harvesting.

5. Empirical Result and Analysis

5.1. Baseline Regression Results

Based on Equation (1), this study uses the baseline regression method to empirically analyze the impact of high-standard farmland construction policy on grain harvest losses. The results are shown in

Table 3. Baseline estimation results.

Variables	(1)		(2)
	Coefficient	Standard Error	Coefficient	Standard Error
HSFC	−0.004 ***	0.001	−0.002 **	0.001
Gender			0.104	0.325
Age			−0.036	0.023
Education			−0.140	0.102
Household income			−0.001	0.005
Labor number			−0.411 ***	0.157
Harvest weather			4.269 ***	0.345
Harvest maturity			0.617	0.438
Rush to plant			1.080 **	0.549
Harvest machinery			−0.075	0.277
Work attitude			−0.074	0.363
Harvest terrain			−1.112 **	0.454
			−0.223	0.484
Service center			−0.163	0.314
Hire harvesters			−1.006 **	0.399
Region			−0.193	0.377
			−0.060	0.387
Constant	5.463 ***	0.160	8.844 ***	1.882
Sample size	3009		3009

Note: ***, ** denote significance at the 1%, 5% levels, respectively.

. Column (1) presents the estimation results without using the control variables. Column (2) shows the estimation results included in control variables. The estimation results show that the high-standard farmland construction policy significantly reduces grain harvest losses, regardless of whether or not control variables are added. The analysis shows that the policy significantly reduces grain harvest losses by an average of 0.2% when all policy factors are fully considered. The baseline regression results confirmed the research hypothesis H1.

With regard to the control variables, harvest weather significantly increases grain harvest losses, which may be attributed to abnormal weather conditions, such as strong winds and storms, which may can easily cause grain lodging. Similarly, rushing to plant the next corn crop increases harvest losses. The possible reason is that farmers’ rush to plant maize accelerates the harvesting speed, which results in incomplete threshing and subsequently large losses. Compared with villages where it is difficult to hire harvesters, villages where it is easy to hire harvesters have significantly reduced grain harvest losses because leaving the matured maize un-harvested results in high shattering losses, exposure to birds, and losses. In addition, household agriculture labor significantly negatively influenced grain harvest losses, indicating that farmers with more labor were less likely to encounter post-harvest losses. Grain harvesting activities are a labor-intensive operation that requires a substantial workforce. The various activities include prompt harvesting, timely drying of maize, winnowing, packaging, and transportation to storage of harvested maize.

5.2. Quantile Regression Results

This study applies the quantile regression method to estimate the coefficients separately at the 0.1, 0.5, and 0.9 quantiles.

Table 4. Quantile regression results.

Variables	(1)	(2)	(3)
	Quantile 0.1	Quantile 0.5	Quantile 0.9
HSFC	−0.0013 *** (0.0002)	−0.0006 (0.0009)	−0.0010 0.0093
Gender	0.0041 (0.0245)	0.0297 (0.1121)	0.3216 * (0.1677)
Age	−0.0001 (0.0016)	−0.0002 (0.0073)	−0.0418 (0.0762)
Education	−0.0003 (0.0078)	−0.0287 (0.0358)	−0.1965 (0.3734)
Household income	−0.0000 (0.0004)	−0.0050 *** (0.0018)	−0.0186 (0.0192)
Labor number	−0.0005 (0.0124)	−0.0014 (0.0569)	−0.9647 (0.5931)
Harvest weather	0.0056 (0.0254)	0.6282 *** (0.1162)	2.8370 *** (1.2104)
Harvest maturity	0.1022 ** (0.0411)	0.5066 *** (0.1884)	1.5943 * (0.9620)
Rush to plant	0.0995 ** (0.0449)	0.6499 *** (0.2059)	2.5795 (2.1441)
Harvest machinery	−0.1088 *** (0.0281)	−0.5188 *** (0.1288)	−0.2697 (0.3411)
Work attitude	−0.0965 ** (0.0402)	−0.1747 (0.1842)	−1.0558 (0.9183)
Harvest terrain	−0.0003 (0.0372)	−0.3204 (0.2928)	−2.5201 (1.5411)
	−0.0049 (0.0384)	−0.0288 (0.3025)	−1.1952 (1.5924)
Service center	−0.0070 (0.0262)	−0.0241 (0.1204)	−0.3043 (0.2537)
Hire harvesters	−0.0001 (0.0286)	−0.2968 ** (0.1311)	−2.6365 *** (0.3657)
Region	−0.0069 (0.0345)	−0.1606 (0.2715)	−0.3850 (1.4288)
	−0.0074 (0.0321)	−0.1110 (0.2532)	−0.8607 (1.3325)
Constant	−0.0022 (0.1319)	0.6768 (0.6043)	17.4843 *** (6.2925)
Sample size	3009	3009	3009

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. The figures in brackets are the standard errors.

presents the quantile regression results. The estimation results indicate that there is obvious heterogeneity in the impact of high-standard farmland construction on grain harvest losses. At the 0.1 percentile, high-standard farmland construction policy significantly reduces grain harvest losses. At the 0.5 and 0.9 percentile, high-standard farmland construction policy has no significant impact on grain harvest losses. The reason is that areas with low grain harvest losses have a higher level of agricultural modernization, local governments pay more attention to agricultural development, and the implementation of high-standard farmland construction is relatively greater, resulting in a more obvious loss reduction effect. On the other hand, areas with low grain harvest losses have a higher level of agricultural social services and a more adequate supply of machinery and equipment. High-standard farmland construction can achieve harvest loss reduction through ways such as improving the application of mechanical operations.

5.3. Robustness Test

The above results show that the high-standard farmland construction policy reduces grain harvest losses. However, omitted variables and sample self-selection may still have confounded the results. In order to improve the robustness of the model estimation, this section draws on existing research and conducts robustness tests by replacing the core explanatory variables, solving the self-selection problem (propensity score matching, PSM), and excluding other policy interference. The robustness test results are shown in

Table 5. Robustness test results.

Variables	(1) Replacement Variables	(2) PSM	(3) Excluding Other Policies
HSFC	−0.341 * (0.192)	−0.003 *** (0.001)	−0.002 ** (0.001)	−0.003 ** (0.015)	−0.003 *** (0.001)
Land policy			YES	YES	YES
Grain policy				YES	YES
Subsidy					YES
Control variables	YES	YES	YES	YES	YES
Region	YES	YES	YES	YES	YES
Constant	3.655 *** (0.574)	8.182 *** (1.799)	8.645 *** (1.859)	8.847 *** (1.879)	9.509 *** (1.898)
Sample size	3009	3009	3009	3009	3009
R-squared	0.004	0.071	0.051	0.073	0.075

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. The figures in brackets are the standard errors.

.

5.3.1. Replacement of Core Explanatory Variables

We consider that high-standard farmland construction can also be characterized by using comprehensive agricultural investment. Therefore, comprehensive agricultural investment is selected as a proxy for the core explanatory variables. The results are shown in column (1) of Table 5, where the regression coefficient is −0.341 and is significant at the 10% level, indicating that policy implementation still significantly affects grain harvest losses.

5.3.2. Solving the Self-Selection Problem

We consider the sample self-selection in the causal relationship between high-standard farmland construction policies and grain harvest losses. This study employs the propensity score matching method (PSM) for solving problems. First, the kernel matching method matches the control group to identify the net impact. After matching, there are no significant differences in the covariates between the experimental and control groups. The sign and significance of the regression coefficient are consistent with the baseline regression results, which verifies the robustness of the results.

5.3.3. Excluding Other Policies

China has also launched other grain policies, including the land rights confirmation policy (Land policy) while implementing the high-standard farmland construction policy, 100 billion grain projects (Grain policy), and agricultural machinery subsidy (Subsidy). These policies are likely to lead to a reduction in grain harvest losses. The agricultural machinery subsidy can effectively reduce the purchase cost of agricultural machinery, promote agricultural machinery application, quickly complete harvest operations, and reduce the risk of production reduction caused by weather. Among them, the land rights confirmation policy uses the proportion of land contract management rights to the total number of household contract operations, and the 100 billion grain projects are represented by the area where households are located. The agricultural machinery subsidy is based on the regional machinery subsidy amount in 2021. As shown in Table 5 Column (3), even if these policies are used as control variables, the high-standard farmland construction policy still significantly reduces grain harvest losses, confirming the robustness of the research conclusions.

6. Potential Mechanisms and Heterogeneity Analysis

6.1. Potential Mechanisms

Empirical results show that the high-standard farmland construction policy can significantly reduce grain harvest losses. However, the potential channels the policy affects still need to be clarified. Therefore, this study further explores its mechanisms. The National High-Standard Farmland Construction Plan (2021–2030) issued by the Ministry of Agriculture and Rural Affairs shows that the policy is conducive to increasing the agriculture scale and promoting the transformation of agricultural production methods. Therefore, this paper assumes that policy reduces harvest losses by expanding operational scale and improving mechanization application. Regression analysis is performed based on Equations (7) and (8) to verify the channel.

Table 6. Potential mechanisms.

Variables	(1)	(2)	(3)	(4)
	Operational Scale	Mechanization Application
HSFC	0.0008 *** (0.0001)	0.0003 *** (0.0001)
Operational scale			−0.0011 ** (0.0005)
Mechanization application				−0.3082 *** (0.1150)
Control variables	YES	YES	YES	YES
Region	YES	YES	YES	YES
Constant	0.417 *** (0.009)	1.668 *** (0.009)	3.0153 *** (0.0576)	3.2116 *** (0.0869)
Sample size	3009	3009	3009	3009
R-squared	0.0942	0.0015	0.0019	0.0024

Note: ***, ** denote significance at the 1% and 5% levels, respectively. The figures in brackets are the standard errors.

shows the estimation results.

6.1.1. Operational Scale Effects of The High-Standard Farmland Construction Policy

The grain planting area is selected to measure operational scale. Table 6, column (1) shows that the results of high-standard farmland construction policy on the operational scale are significant and the coefficient is positive, indicating that high-standard farmland construction policy has a significant promoting effect on operational scale. From the results in column (3), the operational scale significantly reduces grain harvest losses, which suggests that the operational scale plays a role in policy effects on grain harvest losses. The research hypothesis H2 is confirmed.

6.1.2. Mechanization Application Effect of The High-Standard Farmland Construction Policy

Mechanization application can improve agricultural production levels and efficiency. Table 6 column (2) shows that high-standard farmland construction policy significantly impacts mechanization application. The results in column (4) show that the coefficient of mechanization application on grain harvest losses is significantly negative, indicating that mechanical application can reduce grain harvest losses. The research hypothesis H3 is confirmed.

As a result, channels for reducing harvest losses through high-standard farmland policy include expanding operation scale and promoting mechanical application.

6.2. Heterogeneity Analysis

Up to now, the study has assumed that the high-standard farmland construction policy on grain harvest losses is identical across all regions. However, as China is a large country with diverse regions, the effect of high-standard farmland construction may vary depending on heterogeneous characteristics. This study tests variations in the high-standard farmland construction effect on the area under grain harvest losses by estimating it for subsamples in various agricultural production function areas, terrains, and regions. We use the same model as in Table 3.

6.2.1. Heterogeneity of Production Function Areas

This study divided the samples into main grain-producing and non-grain-producing regions based on the agricultural production function area in 2022. Results are reported in

Table 7. Heterogeneity analysis result.

Variables	Production Function Area		Terrain Features Area
	(1) Major Grain-Producing	(2) Non-Major Grain Producing	(3) Plains	(4) Hills	(5) Mountains
HSFC	−0.003 *** (0.001)	−0.003 (0.003)	−0.002 ** (0.001)	−0.008 (0.007)	−0.012 (0.014)
Control variables	YES	YES	YES	YES	YES
Region	YES	YES	YES	YES	YES
Constant	7.518 *** (2.362)	13.875 *** (3.739)	5.447 ** (2.602)	6.238 * (3.648)	18.713 *** (3.949)
Sample size	1736	1273	1720	586	703
R-squared	0.108	0.057	0.094	0.099	0.072

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. The figures in brackets are the standard errors.

and indicate that high-standard farmland policy significantly reduces grain harvest losses in major producing areas. By contrast, the effect in non-grain-producing areas is barely significant. One possible explanation is that the principle of “prioritizing the main grain-producing areas” was put forward in the “Opinions on effectively strengthening the construction of high-standard farmland to enhance the national food security guarantee capacity1”. The intensity of policy implementation in the main grain-producing areas may be greater than that in the non-main grain-producing regions, and its effect on reducing grain harvest losses is significant.

6.2.2. Heterogeneity of Terrain Features

This study conducts a terrain heterogeneity analysis on the effects of high-standard farmland construction on grain harvest losses. We divided the sample into three regions: plains, hills, and mountains. Table 7 shows that the impact of the policy on grain harvest is reflected in plain areas, but it cannot significantly affect hills and mountainous regions, probably because the terrain hinders the wide adoption of mechanization, making it challenging to utilize the policy’s impact on grain harvesting in hills and mountainous areas.

6.2.3. Heterogeneity of Different Geographic Locations

This study conducted a regional heterogeneity analysis on the impact of high-standard farmland construction on grain harvest losses. The sample is categorized into three areas: Eastern, Central, and Western regions and the effect of policy on grain harvest losses for these three regions was estimated separately.

Table 8. Results in different geographical locations.

Variables	(1) Eastern Regions	(2) Central Regions	(3) Western Regions
HSFC	−0.003 *** (0.001)	−0.003 (0.006)	−0.010 (0.009)
Control variables	YES	YES	YES
Region	YES	YES	YES
Constant	4.694 (2.892)	10.557 *** (3.503)	10.350 *** (3.294)
Sample size	880	900	1229
R-squared	0.095	0.125	0.059

Note: *** denotes significance at the 1% levels. The figures in brackets are the standard errors.

shows that high-standard farmland construction policy significantly reduces grain harvest losses in the Eastern region. By contrast, the effect impact on the Central and Western regions is barely significant. Compared with the Eastern region, the Central and Western regions are relatively economically backward and have a low level of agricultural supporting infrastructure construction. These factors make it difficult to improve the operational scale and mechanization application. Therefore, the impact of the policy on grain harvest losses is limited.

7. Conclusions and Recommendations

This study utilizes the quantile regression method to verify the impact of high-standard farmland construction on grain harvest losses. The 2022 CPHLS database is used to study its mechanism and heterogeneous effects. The empirical results show that high-standard farmland construction policy significantly reduces grain harvest losses. The results remain robust after controlling for potential interference from other policies and addressing self-selection issues. Mechanism analysis reveals that high-standard farmland construction reduces grain harvest losses by expanding operation scale and improving mechanization applications. Heterogeneity analysis shows that due to differences between regions, the impact of high-standard farmland construction on grain harvest losses is heterogeneous. This study finds that the effect of the policy is more pronounced in major grain-producing areas, Eastern regions, and plain areas.

We can draw three main policy implications from our analysis. First, it is necessary to continue to strengthen policy organization and implementation, increase capital investment, and make up for the shortcomings of agricultural infrastructure. China’s policy needs to improve construction standards effectively, enhance the effect of reducing food losses, and ensure global food supply.

Second, focusing on critical impacts such as large-scale planting and the application of agricultural mechanization is essential to reduce grain harvest losses, and it serves as an important mechanism through which high-standard farmland construction influences these losses. Therefore, during the policy implementation, attention should be paid to constructing mechanized farmland roads, widening existing roads, and improving overall accessibility. In addition, special attention needs to be paid to promoting land transfer and accelerating moderately large-scale operations to improve China’s comprehensive grain production capacity through land leveling and centralized and continuous operations.

Third, the Chinese government should pay attention to the differences in impact effects and the accuracy of policy measures. For major grain-producing areas, plain areas and more economic development areas, it is necessary to stabilize further to give full play to the economic benefits of high-standard farmland construction and promote sustained and high-quality agriculture development. Policy should be strengthened for areas with low grain production, poor terrain conditions, and underdeveloped economies. There is a need to actively innovate and promote practical high-standard farmland construction plans and models and expand the space for high-standard farmland construction policy.

We analyze the cost–benefit of high-standard farmland construction from two dimensions. First, from the time dimension, the short-term financial investment for such construction is substantial, with subsidies ranging from CNY 1300 to 2400 per mu, predominantly funded by the government. This initial outlay is anticipated to yield a long-term return, as the construction is projected to enhance the average yield per mu by 10% to 20%. The investment is expected to recoup within a five-year period, with the potential for even greater returns in the subsequent years. Secondly, examining the matter from various research perspectives reveals that the benefits of high-standard farmland construction extend beyond economic gains. It also encompasses social and ecological advantages. Socially, the increase in large-scale grain growers and cooperatives is likely to bolster the overall grain production capacity. Ecologically, the project’s implementation is designed to decrease the application of chemical fertilizers and pesticides, as well as water and energy loss. This is achieved by improving ancillary infrastructure such as irrigation systems and roads and by adopting water-saving irrigation and precise agricultural technologies for fertilization and pesticide application. These measures are instrumental in safeguarding the agricultural ecosystem. Consequently, when viewed from these dual dimensions, the long-term benefits of implementing high-standard farmland construction are deemed to significantly outweigh the initial costs.

This paper further compares the impact of high-standard farmland construction on food security with relevant literature from other countries. High-standard farmland construction is a form of land consolidation, a practice implemented globally. Land consolidation reduces the economic costs associated with land fragmentation, thereby creating favorable conditions for large-scale agricultural development [42,43]. It has been demonstrated to enhance food productivity in several European countries [44,45,46,47]. Similar conclusions have been found in some developing countries [48,49]. For instance, Boonchom argues that land consolidation improves harvester efficiency by reducing activities such as turning, waiting, and mechanical adjustments [50]. Asiama used cases from Rwanda and Ethiopia to illustrate that land consolidation in sub-Saharan Africa addresses land tenure fragmentation, increases farmers’ willingness to boost crop yields, enhances productivity through mechanization, ensures stable food supply, and ultimately strengthens national food security [51].

This study has two limitations and identifies areas for future research. First, the data limitations also restricted our research of grain harvest losses before and after policy reform. Further research on its long-term effect requires better panel data and other more advanced methods. Second, the loss quantity is an approximation rather than an objective measurement, which needs to be confirmed by field measurement data. Finally, this paper only considers the corn harvest stage, and other post-harvest stages such as storage and drying have not been analyzed. Therefore, future research will combine actual measurements and farmer declaration methods to produce more reliable quantitative data on maize post-harvest losses.