Does Construction of High-Standard Farmland Improve Total Factor Productivity of Grain? Evidence from China, 2000–2021

Abstract

This study investigates the impact of China’s construction of high-standard farmland (CHSF) initiatives on grain productivity, focusing on total factor productivity growth of grain (TFPG) from 2000 to 2021. Using a continuous Difference-in-Differences (DID) approach based on balanced panel data from 31 Chinese provinces, this paper identifies significant productivity improvements, with TFPG increasing by an average of 7% post-implementation of CHSF. However, the effects are not uniform across regions—productivity gains are more pronounced in non-major grain-producing and plain areas, emphasizing the role of region-specific infrastructure and adaptive strategies. These findings provide empirical evidence on how large-scale farmland improvement enhances productivity through mechanization and better land use. However, the reliance on provincial-level data may result in localized variations in CHSF implementation being overlooked, suggesting the need for further micro-level analysis. Overall, this study highlights the importance of tailored agricultural policies to enhance their effectiveness and promote agricultural sustainability in China and other developing economies.

1. Introduction

Ensuring food security remains one of the most pressing global challenges of the 21st century, as it is fundamental to human well-being, economic stability, and sustainable development. In 2023, nearly 282 million people across 59 countries faced acute food insecurity, an increase of 24 million from the previous year [1]. This escalation, driven by factors such as soil degradation, climate change, declining arable land, and water scarcity [2], underscores the urgency of enhancing agricultural productivity. Against this backdrop, the increase in food production must come from “smarter production”, i.e., productivity gains [3]. Therefore, improving the total factor productivity of grain and forming an intensive growth pattern are the best choices to ensure food security and sustainable development of production.

China, as the world’s largest agricultural producer and most populous nation, faces unique challenges in maintaining food security. Its food self-sufficiency rate currently stands at approximately 70%, reflecting a persistent supply gap [4]. These challenges are exacerbated by rapid population growth, limited per capita arable land, and environmental constraints. Recognizing these challenges, the Chinese government has actively implemented a series of reforms, with the construction of high-standard farmland (CHSF) being a key initiative. The Chinese government first formally proposed this initiative in 2004, and it was widely implemented after 2011. CHSF aims to enhance farmland infrastructure, improving its efficiency, productivity, and environmental sustainability [5]. CHSF has been shown to support agricultural mechanization and significantly increase both the scale and specialization of agricultural production [6].

Recent studies suggest that improvements in TFPG have played a pivotal role in bolstering China’s grain output and self-sufficiency [7]. Consequently, enhancing TFPG is one of China’s most critical strategies for strengthening food security [8]. Since 2004, numerous studies have explored the importance of TFPG, highlighting its role in improving agricultural efficiency and food security. Existing studies employ diverse methodologies. While the DEA–Malmquist index remains the standard method for measuring TFPG, other methods, such as the SBM–Malmquist index and stochastic frontier models, are also used [9,10,11]. Determinants of TFPG range from macro-level factors, such as agricultural policies and infrastructure [12,13], to micro-level inputs, including fertilizers, machinery, and labor [14].

International evidence also underscores the potential benefits of similar land consolidation and infrastructure improvement policies. For instance, Nguyen et al. [15] found that land consolidation in Vietnam’s highly fragmented Red River Delta significantly improved spatial organization, facilitated crop structure transformation, increased household income, and accelerated agricultural mechanization. Similarly, Hendricks et al. [16]’s study on Germany’s large-scale infrastructure projects emphasized the importance of land consolidation in reducing fragmentation, minimizing landowner losses, and supporting economic development. Asiama et al. [17] demonstrated that aligning land consolidation strategies with local economic and technological conditions improved food production and rural sustainability in the Netherlands, Lithuania, and Rwanda. These international cases suggest that similar policies, such as China’s CHSF, can promote agricultural productivity by facilitating mechanization and improving land use efficiency.

However, while land consolidation and farmland infrastructure improvement policies have been widely studied in countries such as Vietnam, Germany, and the Netherlands, as stated, their findings provide context-sensitive insights rather than universally transferable solutions. China, like these countries, grapples with persistent issues such as land fragmentation, declining rural labor availability, and the need to enhance agricultural productivity through mechanization and scale operations. These shared structural challenges justify a comparative examination of international experiences. However, it is equally important to acknowledge China’s unique institutional context. Unlike many Western countries that operate under systems of private land ownership, China’s collective land ownership model necessitates a different institutional approach to land consolidation and infrastructure investment. Therefore, this study draws on international evidence not as a direct template, but as a comparative framework for understanding the potential mechanisms and outcomes of CHSF policies in China.

Despite these global insights and the significance of CHSF, empirical studies specifically evaluating the contribution of China’s high-standard farmland (CHSF)’s infrastructure improvements to total factor productivity growth (TFPG) remain limited. While previous studies have examined the effects of land consolidation, a component of CHSF, on productivity and mechanization efficiency [18], few studies employ micro-econometric techniques or province-level panel data to rigorously assess the effectiveness of CHSF policies in the Chinese context. Moreover, while mechanization is widely recognized for its role in boosting grain output and productivity [19,20], empirical evidence on how CHSF facilitates this process is still scarce.

This study makes several novel contributions to the literature on CHSF and TFPG. First, it addresses a significant research gap by empirically examining the impact of China’s high-standard farmland (CHSF) on total factor productivity growth (TFPG), using a continuous Difference-in-Differences (DID) model and provincial panel data spanning 2000–2021. Unlike previous studies that primarily focus on land expansion or yield effects, this study emphasizes TFPG as a more comprehensive indicator of long-term agricultural productivity. Second, it investigates the mechanisms through which CHSF contributes to TFP and identifies mechanization as one of the channels. Third, the study conducts heterogeneity analysis by comparing functional grain production areas with non-functional grain production areas and plain areas with non-plain regions, thereby highlighting the influence of both natural conditions and policy environment on the effectiveness of CHSF interventions.

The remainder of this paper is structured as follows: Section 2 outlines the theoretical framework and research hypothesis. Section 3 describes the variable selection and model construction. Section 4 presents the empirical results and analysis. Section 5 discusses the findings. Finally, Section 6 concludes the study.

2. Theoretical Analysis and Research Hypothesis

2.1. Background on Construction of High-Standard Farmland (CHSF)

The construction of high-standard farmland (CHSF) was first formally put forward in 2004 and broadly implemented in 2011 as a strategic initiative to enhance China’s agricultural productivity and sustainability. The policy aims to create productive, flat, concentrated farmland areas that are environmentally friendly. CHSF involves a series of infrastructure improvements, including enhanced irrigation systems, improved soil quality, land leveling, and road construction to facilitate mechanization and access [21]. The construction focus varies across regions; for example, in plains, efforts are concentrated on mechanization, while in hilly areas, the emphasis is on soil erosion control and irrigation [22].

The implementation of CHSF has progressed in distinct phases. The initial phase (2005–2010) focused on pilot projects and laying the groundwork for broader application, targeting key grain-producing areas. The second phase (2011–2015) expanded coverage and introduced more advanced agricultural technologies and soil improvement methods. The third phase (2016–2020) aimed to standardize high-standard farmland construction across provinces, with increased investment and more stringent evaluation criteria. By 2021, approximately 70 million hectares of high-standard farmland had been constructed, contributing to increased grain output and enhanced rural resilience. The current phase (2021–present) emphasizes sustainable practices, precision agriculture, and improving farmer income through better land management and mechanization [23].

Research widely recognizes that this policy supports sustainable agricultural development, alleviates food scarcity, and contributes to the broader goals of agricultural modernization and rural revitalization [24,25].

2.2. TFP Theory and TFPG

Total factor productivity (TFP) is generally regarded as a measure of technological progress and efficiency improvement. It captures output growth that cannot be attributed to increases in conventional inputs such as labor, capital, and land [26]. In agriculture, TFP reflects improvements in production techniques, resource allocation, and input utilization efficiency [27]. TFP evolves through technological advancements, improved infrastructure, and better institutional arrangements, making it an essential driver of sustainable productivity growth [28]. Therefore, we conceptualize total factor productivity of grain (TFPG) as the overall efficiency of all inputs, including land, capital, and labor to produce grain.

2.3. Research Hypothesis

H1: 

The construction of high-standard farmland (CHSF) promotes total factor productivity growth of grain (TFPG) in agriculture.

High-standard farmland construction (CHSF) has the potential to significantly enhance total factor productivity growth (TFPG) through multiple, interrelated channels. At its core, TFPG reflects improvements in efficiency and technological progress that go beyond input accumulation. CHSF contributes to this process in several essential ways.

First, CHSF improves land quality and production stability by enhancing land leveling, irrigation, drainage, and soil fertility. These improvements help reduce natural risks and increase the marginal productivity of land. This aligns with findings that CHSF significantly improves grain yield and income by enhancing land productivity and stability of returns [29]. Furthermore, high-quality farmland reduces the overuse of inputs like fertilizers and water, thereby improving input efficiency and ecological sustainability [30].

Second, CHSF facilitates mechanization and technological adoption. By creating contiguous and standardized plots with road access and irrigation systems, CHSF removes the physical barriers to machinery operation and service delivery, supporting scale-efficient production [31].

Third, CHSF reduces production-related transaction costs (e.g., costs related to water allocation, land consolidation, and transportation), which enhances both technical and allocative efficiency. These cost savings enable farmers to allocate inputs more rationally and adopt innovative technologies more readily, both of which are crucial drivers for TFPG [32].

Finally, CHSF is often implemented alongside the modernization of agricultural services and institutional support. This includes digital monitoring systems, internet-based platforms, and artificial intelligence, all of which contribute to the development of smart agriculture and can indirectly boost TFPG [33].

H2: 

CHSF affects TFPG through the allocation of factors of production and mechanization level at the technological level, and agricultural land scale management and service scale management at the agricultural scale management level.

CHSF aligns with TFP theory by improving technological efficiency and optimizing input usage. One study suggests that CHSF enhances grain output capacity by optimizing both the distribution of production elements at the technological level and the scale of agricultural operations at the institutional level [34]. We propose that these processes can be extended to improve the total factor productivity of grain (TFPG), as shown in Figure 1.

At the technological level, CHSF can impact TFPG through more efficient factor allocation [35] and higher mechanization levels [36]. Specifically, CHSF promotes land leveling, irrigation infrastructure, and road improvements, which reduce the heterogeneity of land parcels and transaction costs among farmers. These changes facilitate the reallocation and concentration of land resources, enabling larger-scale operations and the reorganization of production factors such as labor, fertilizers, and water resources [35,37]. Additionally, by improving field accessibility and plot regularity, CHSF creates physical conditions conducive to the use of agricultural machinery—such as tractors, harvesters, and seeders—across larger, contiguous plots [22,36]. This mechanization reduces labor intensity and increases operational efficiency, ultimately raising TFPG [19]. These mechanisms are consistent with production theory, as they enhance the marginal productivity of both capital and labor inputs.

At the institutional level, CHSF facilitates the consolidation of fragmented land, promoting scale management in agriculture [38]. Agricultural scale management encompasses two forms: agricultural land scale management (ALSM) and service scale management (SSM) [39]. Through land consolidation, CHSF addresses issues of land fragmentation, expands agricultural scale management to reduce the costs of production conversion across plots, and enhances cultivation efficiency [40]. By freeing operational space for mechanized technologies, CHSF helps prevent soil erosion, enriches soil organic matter, and improves both grain quality and quantity [41], thereby boosting TFPG. In terms of service-scale management, CHSF fosters land contiguity and factor concentration, facilitating both horizontal and vertical labor specialization among farmers. This resolves the challenges of decentralized agricultural management [42], promoting industry aggregation, which raises productivity in different grain production segments. Such industry agglomeration further increases TFPG through specialization [43].

3. Variable Description and Model Construction

3.1. Variable Description

3.1.1. Outcome Variable

Explained variable (TFPG): This variable represents the overall efficiency of all inputs, including land, capital, and labor, within a specific period to produce grain. To measure TFPG, the DEA–Malmquist index, based on Färe et al. [44], was employed. This nonparametric approach has the advantage of not requiring a specific functional form or assumptions about data’s stochastic properties, allowing for the inclusion of inefficiency. The DEA–Malmquist index divides total factor productivity into two components: technical efficiency and technical progress. On one hand, technical efficiency is viewed from two angles—input and output. The input perspective assesses the ratio of actual inputs to the minimum possible inputs required for a given output level, while the output perspective examines the ratio of actual outputs to maximum potential outputs at a given input level, thus indicating resource allocation and utilization efficiency. This technical efficiency can be further divided into pure technical and scale efficiency. On the other hand, technological progress captures innovations and improvements in production processes, intermediate inputs, and product-specific production skills, forming the basis of the DEA–Malmquist index for TFPG assessment. Thus, we construct the following:

$$\begin{gathered} \begin{array}{c}{M}_i(X^{t+1},Y^{t+1},X^t,Y^t)=\\ {\left[{\displaystyle \frac{d_i^t(X^{t+1},Y^{t+1})}{d_i^t(X^t,Y^t)}}{\displaystyle \frac{d_i^{t+1}(X^{t+1},Y^{t+1})}{d_i^{t+1}(X^t,Y^t)}}\right]}^{{\displaystyle \frac{1}{2}}}=\\ {\displaystyle \frac{d_i^{t+1}\left(X^{t+1},Y^{t+1}\right)}{d_i^{t+1}\left(X^t,Y^t\right)}}\times \\ {\left[{\displaystyle \frac{d_i^t(X^{t+1},Y^{t+1})}{d_i^{t+1}(X^{t+1},Y^{t+1})}}{\displaystyle \frac{d_i^t(X^t,Y^t)}{d_i^{t+1}(X^t,Y^t)}}\right]}^{{\displaystyle \frac{1}{2}}}=\end{array} \\ \begin{array}{c}E\left(X^{t+1},Y^{t+1},X^t,Y^t\right)\times T\left(X^s,Y^s,X^{s+1},Y^{s+1}\right)\\ \end{array} \end{gathered}$$

(1)

In Formula (1), $X^t$ represents inputs at stage t, $Y^t$ represents outputs at stage t, ($X^t$,$Y^t$) represents the input–output variable at stage t, $d_i^t(X^t,Y^t)$ and $d_i^t(X^{t+1},Y^{t+1})$ represent the level of technical efficiency of a producer with stage t technology as a reference for periods t and t + 1, respectively, and $M_i(X^{t+1},Y^{t+1},X^t,Y^t)$ represents the TFPG at stage t. Since the TFPG measurement process requires specific input and output indicators, this study uses total grain output as the output variable and includes four input indicators—land, labor, fertilizer, and machinery—as input variables for the baseline model, as shown in

Table 1. TFPG input–output indicator system.

Category	Indicators	Explanation of Indicators	Units
Output	Grain	Total grain output	10,000 tons
Input	Land input	Grain planting area	Kilo-hectares
	Labor input	Number of laborers engaged in grain production	10,000 people
	Fertilizer input	Fertilizer use for grain production	10,000 tons
	Machinery input	Total power of machinery used for grain production	10,000 kw

.

3.1.2. Variable of Interest

Variable of interest (CHSF—$Hig{h}_i\times I_{it}^{post}$): The variable ‘$High$’ represents high-standard farmland—arable land that is leveled, contiguous, well-facilitated, equipped with necessary infrastructure, fertile, ecologically sound, and resilient to disasters. This type of land is designated permanent basic farmland, ensuring reliable harvests under both drought and flood conditions, and is well suited to modern agricultural production and management practices. Given that CHSF is a further transformation and upgrading of basic farmland and low- and medium-yield fields, there was no concept of CHSF before the implementation of the policy. Therefore, before the widespread implementation of the policy in 2011, the area of CHSF in this study was represented by the area of transformed low- and medium-yield fields. After the implementation of the policy, the CHSF area is formed by the sum of the areas of high-standard farmland demonstration projects and renovated medium- and low-yield fields. Hence, $Hig{h}_i$ refers to the proportion of high-standard farmland demonstration projects and medium–low-yield farmland renovation areas in the total cultivated land area in the province i. On this basis, the interaction term between $Hig{h}_i$ and the policy point-in-time dummy variable I_it^post is constructed as the core explanatory variable CHSF (High_i × I_it^post).

3.1.3. Controls

In this study, we referred to the approach of Liang et al. [37] and Cai and Zeng [45] to control for a series of variables that may affect TFPG. This study selected the following control variables: (1) cultivated land area per capita (CLAP), which controls for the effect of differences in cultivated land area in different provinces on TFPG; (2) planting structure (PS), which is the share of the area sown with grain in the total area sown with crops as a measure; (3) per capita income from the grain industry (PIGI), which is the ratio of grain output to agricultural output multiplied by farmers’ disposable income per capita as the measure, referring to the method used by Wang et al. [46]; (4) disaster rate (DR), which is the ratio of the disaster-affected area to the area sown with grain; (5) rainfall deviation degree (RDD), which is the absolute value of the deviation of rainfall data from the mean for each province; (6) fiscal expenditure on agriculture (FEA), which is the share of agriculture, forestry, and water expenditure in public budget expenditure; (7) urbanization rate (UR), which is the share of urban population out of the total population; (8) industrialization level (IL), which is the share of industrial value added in the gross regional product; and (9) transport infrastructure level (TIL), which is the share of the total road transport in the road mileage.

3.1.4. Data Sources and Descriptive Statistics

This study uses panel data from 31 provinces in China from 2000 to 2021 to assess the impact of CHSF on TFPG. Provincial-level data are used because CHSF policies and funding are implemented at the provincial level, making it more suitable for capturing policy effects. City- or county-level data may introduce inconsistencies due to differences in reporting standards, funding timelines, and implementation strategies across regions. Moreover, reliable county-level CHSF data for the entire study period are not consistently available, limiting its feasibility for long-term analysis.

The study period (2000–2021) is chosen as it captures both pre- and post-policy effects. Although CHSF was formally introduced in 2004 and expanded nationwide after 2011, many provinces had already begun farmland consolidation projects earlier. Including data from 2000 establishes a solid baseline for comparison, while extending to 2021 allows for assessment of long-term impacts, considering the lagged effects of infrastructure investments.

CHSF indicators are sourced from the China Financial Statistics Yearbook. Input and output indicators used to measure TFPG, as well as Herfindahl–Hirschman Index (HHI)-related raw indicators, cultivated land area (CLA), planting structure (PS), per capita income in the grain industry (PIGI), and disaster rate (DR), are drawn from the China Rural Statistical Yearbook. Rainfall deviation degree (RDD) is sourced from the China Meteorological Data Network. Fiscal expenditure for agriculture (FEA), urbanization rate (UR), and industrialization level (IL) are drawn from the China Statistical Yearbook. Data related to transport infrastructure level (TIL) are from the China Transportation Statistics Yearbook.

Technological change (TC) is divided into two components: the allocation of production factors (APF) and the mechanization level (ML). Raw data related to the allocation of production factors (APF) are obtained from the National Compendium of Cost and Benefit Information on Agricultural Products. Data related to the mechanization level (ML) are obtained from the China Agricultural Machinery Industry Yearbook.

Agricultural scale management (ASM) is categorized into agricultural land scale management (ALSM) and service scale management (SSM), and the data are derived from the China Rural Business Management Statistics Annual Report and the China Rural Statistics Yearbook. Abnormal data are processed with some missing data calculated by interpolation.

Table 2. Definition and descriptive statistics of variables.

Variables	Variable Names	Variable Definition	Mean	Sd	Min	Max
Total factor productivity of grain	TFPG	Measurement of total factor production index for grain based on DEA	1.014	0.042	0.846	1.341
Area of high-standard farmland	HIGH	Share of sum of areas of high-standard farmland demonstration projects and renovation of medium- and low-yield fields in total cultivated land area (unit: %)	0.236	0.193	0.001	0.984
Cultivated land area per capita	CLAP	Cultivated land area per capita in logarithms (unit: kilo-hectares)	3.435	0.485	2.046	4.250
Planting structure	PS	Share of area sown with grain in total area sown with crops (unit: %)	0.656	0.131	0.328	0.968
Per capita income from the grain industry	PIGI	Ratio of grain output to agricultural output multiplied by per capita disposable income of farmers (unit: CNY 10,000)	0.200	0.163	0.016	1.048
Disaster rate	DR	Share of disaster-affected area in area sown with grain (unit: %)	0.328	0.221	0.130	0.998
Rainfall deviation degree	RDD	Precipitation deviation in logarithms (unit: millimeters)	2.716	0.498	0.094	3.594
Fiscal expenditure for agriculture	FEA	Share of agriculture, forestry, and water expenditure in public budget expenditure (unit: %)	0.115	0.059	0.049	0.468
Urbanization rate	UR	Share of urban population out of the total population (unit: %)	50.852	15.720	13.885	94.152
Industrialization level	IL	Share of industrial value added in the gross regional product (unit: %)	0.335	0.096	0.071	0.559
Transport infrastructure level	TIL	Share of total road transport in the road mileage (unit: %)	1.146	1.415	0.005	10.749
Technological change: the allocation of production factors	TC: APF	Average value of the share of input factors such as pesticides, fertilizers, seeds, etc., in the allocation of labor factors in the grain production process (unit: %)	0.310	0.199	0.006	1.735
Technological change: mechanization level	TC: ML	Number of machines per square hectometer (hm2) of a household (unit: set/hm2)	26.664	34.210	0.269	276.5127
Agricultural scale management: agricultural land scale management	ASM: ALSM	Number of farm households operating more than 10 acres of arable land (unit: 10,000 households)	132.135	108.488	0.199	977.461
Agricultural scale management: service scale management	ASM: SSM	HHI: A composite index to measure production concentration	0.361	0.188	0.0302	0.858

presents the descriptions and descriptive statistics of the main variables.

Figure 2 illustrates the national average CHSF and the individual provincial CHSF for all 31 provinces from 2011 to 2021. The level of CHSF varied over time within the same province and also differed significantly across provinces in the same year.

3.2. Empirical Strategy

3.2.1. Baseline Model

To examine the effect of CHSF on TFPG, this study employs a continuous DID model, identifying 2011 as the policy intervention year. This choice is grounded in both theoretical rationale and formal policy developments.

The year 2011 marks a critical turning point in the evolution of China’s CHSF policy, transitioning from an exploratory phase (1988–2010) to a standardized implementation phase. Prior to 2011, land consolidation efforts primarily aimed to compensate for arable land losses due to urbanization and industrialization, without a unified framework, measurable quality standards, or specific construction goals for CHSF. However, with growing concerns over ecological sustainability and agricultural efficiency, the policy direction shifted toward improving land quality and infrastructure, culminating in the formal institutionalization of CHSF.

Two landmark policy document releases in 2011 underscore this transition. First, the Ministry of Land and Resources released the Technical Specifications for High-standard Farmland Construction (Trial), which for the first time systematically defined the standards, content, and objectives for CHSF at the national level. Second, the National Land Consolidation Plan (2011–2015) [47], approved by the State Council and jointly formulated by several ministries, officially marked the institutionalization and scaling-up of CHSF construction efforts across China. According to the Construction Standards for High-standard Basic Farmland (TD/T1033–2012) [48], CHSF was defined as “basic farmland developed through rural land consolidation within a certain period, characterized by contiguous layout, supporting infrastructure, high and stable yields, ecological soundness, strong disaster resistance, and suitability for modern agricultural production and operations”. Together, these developments signaled a fundamental policy shift and provided a clear, exogenous shock, justifying the use of 2011 as the intervention point in our analysis.

Since 2011, China has broadly implemented the CHSF policy, affecting all provinces simultaneously. However, the extent of CHSF implementation varies across provinces at different stages of policy implementation—both across provinces at the same point in time and within the same province (cities and regions) before and after implementation.

As CHSF progresses, its coverage expands continuously within each province. In contrast to the standard DID model, this study uses the continuous variable $Hig{h}_i$ to represent CHSF instead of the traditional DID dummy variables. Continuous DID does not change the attribute characteristics of traditional DID and mitigates potential bias arising from subjective human judgment of the experimental and control groups. The baseline model setup for this study is as follows:

$$Y_{i,t}=\alpha +\beta Hig{h}_i\times I_{it}^{post}+\delta X_{it}+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(2)

In Formula (2), the explanatory variable $Y_{i,t}$ represents TFPG in period t in province i. The core explanatory variable is $Hig{h}_i\times { I}_{it}^{post}$, where $\mathrm{H}\mathrm{i}\mathrm{g}{\mathrm{h}}_{\mathrm{i}}$ represents the area of the CHSF in province i and $I_{it}^{post}$ represents a dummy variable at the point of policy, with t being the year of policy implementation. The dummy variable $I_{it}^{post}$ takes the value 1 when t ≥ post; otherwise, it takes the value 0. $X_{it}$ represents the control variable; ${\mu }_i$ and ${\gamma }_t$ represent province and year fixed effects, respectively; ${\epsilon }_{it}$ represents the random error term; $\alpha$ is a constant term; and $\beta$ and $\delta$ are the estimated coefficients. Standard errors are clustered at the provincial level.

3.2.2. Robustness Test

Parallel Trend Test

A key assumption for the validity of the DID framework is that treated and untreated units follow parallel trends in the absence of the policy intervention. Since CHSF serves as an exogenous shock to TFPG, we conduct a parallel trend test to verify whether TFPG trends were similar across provinces before CHSF implementation. In addition, we expanded the DID method to a continuous DID method. The empirical model is as follows:

$$Y_{i,t}=\alpha +\underset{\mathrm{t}=2000}{\sum ^{2021}}{\beta }_t(Hig{h}_i\times yea{r}_t)+\delta X_{it}+{\mu }_i+{\gamma }_i+{\epsilon }_{it}$$

(3)

In Formula (3), $yea{r}_t$ is a dummy variable that takes values of 2000, 2002… 2021, with 2011, the time of policy implementation, as the base year. ${\beta }_t$ denotes a series of estimates from 2000 to 2021, and the other variables and coefficients are consistent with Formula (2).

Placebo Test

To ensure robustness, we conducted a placebo test to rule out the possibility that observed effects on TFPG are driven by random factors rather than CHSF. In this test, we randomly assigned CHSF values across provinces to create a virtual experimental group (counties with a higher degree of CHSF based on random assignment). This randomization process was repeated 500 and 1000 times, respectively, to generate a distribution of estimated effects under the null hypothesis of no treatment effect.

Other Robustness Tests

In addition to parallel trend and placebo tests, we employed multiple approaches to test for robustness. First, we compared the estimation results of ordinary standard errors and heteroskedasticity–serial correlation–cross-section correlation robust standard errors separately to explore the robustness of the benchmark regression results.

Second, we lagged the explanatory variables by one period to address potential reverse causality between CHSF and TFPG. As CHSF implementation requires time, the policy likely has a lagged effect on TFPG. Hence, we lagged the core explanatory variable by one period to handle the endogeneity problem.

Third, we adjusted the scope of the sample selection. The dataset includes four municipalities—Beijing, Tianjin, Shanghai, and Chongqing—whose agricultural endowments and development structures differ significantly from other provinces. These municipalities exhibit extremely high urbanization levels—87.5% for Beijing, 89.3% for Shanghai, 84.5% for Tianjin, and 70.3% for Chongqing in 2020—compared to the national average of 63.9% [49]. Similarly, in terms of agricultural importance, the agricultural GDP shares of Beijing, Shanghai, and Tianjin were all below 2% in the same year, far lower than those of major grain-producing provinces such as Henan or Heilongjiang (over 10%) [49]. Additionally, arable land per capita in Beijing and Shanghai was below 0.03 hectares, compared to the national average of 0.1 hectares [50]. These regions also implement differentiated rural and fiscal policies that diverge from national standards, potentially leading to biased estimations. Therefore, to ensure comparability and robustness, we excluded these four municipalities from the analysis.

Fourth, we minimized potential interference from other policies. Other policies implemented during CHSF adoption may have influenced TFPG. For instance, in 2013, China introduced a land rights policy, focusing on the registration and certification of rural land contracting and management rights. Since land tenure reforms can affect agricultural production decisions, we excluded post-2014 data to exclude the interference of the land tenure policy on the TFPG.

3.2.3. Mechanism Analysis

To understand how CHSF influences TFPG, we explore two key channels, technological change (TC) and agricultural scale management (ASM), as we hypothesized in H2.

For technological change, we follow the approach of scholars like Zhong et al. [51] and measure the allocation of production factors (APF) using the average ratio of input factors—such as pesticides, fertilizers, and seeds—to labor in grain production. The mechanization level (ML) is quantified as the number of agricultural machines per unit of arable land.

Agricultural scale management (ASM) includes agricultural land scale management (ALSM) and service scale management (SSM). ALSM is designed to optimize farm size to maximize farmers’ planting benefits. We use the count of farms operating more than 10 acres as an indicator of ALSM, with a higher count reflecting a greater degree of agricultural scale management. SSM is assessed using the Herfindahl–Hirschman Index (HHI), a composite measure of industry concentration ranging from 0 to 1, where values closer to 1 signify a higher degree of specialization and labor clustering. The measurement formula is as follows:

$$HH{I}_{it}={\sum }_{n=1}^N(S_{itn}{)}^2={\sum }_{n=1}^N{\left({\displaystyle \frac{x_{itn}}{x_{it}}}\right)}^2$$

(4)

In Formula (4), N indicates the type of grain crop, and $S_{itn}$ represents the proportion of the sown area $x_{itn}$ of the n-th grain crop (including wheat, maize, and rice in this study) to the total sown area $x_{it}$ of grain crops in the t-th period in the i-th province.

3.2.4. Heterogeneity Analysis

To further investigate the variation in the relationship between CHSF and TFPG, we conduct heterogeneity analyses based on functional grain production areas and topographical differences.

Figure 3 illustrates the proportion of grain production across China’s main producing areas, showing variability ranging from 3% to 22%.

To examine differences across functional grain production areas, we introduce a production functional area dummy variable. Based on the “Outline of the Medium- and Long-Term Plan for China’s Food Security” (2008–2020), sample provinces (including cities and regions) were classified as main grain production areas—covering 13 provinces, namely Liaoning, Jilin, Heilongjiang, Inner Mongolia, Hebei, Shandong, Anhui, Jiangsu, Jiangxi, Henan, Hunan, Sichuan, and Hubei—with all other provinces classified as non-main grain production areas. This variable takes a value of 1 if the province is a main grain production area and 0 otherwise.

Given China’s diverse geography and topographical differences, we also classify the 31 sample provinces into plain and non-plain regions. Plain terrain includes the 16 provinces (cities and regions) within the Northeastern Plain, North China Plain, and the middle and lower reaches of the Yangtze River: Beijing, Tianjin, Hebei, Inner Mongolia, Liaoning, Jilin, Heilongjiang, Shanghai, Jiangsu, Zhejiang, Anhui, Jiangxi, Shandong, Henan, Hubei, and Hunan. All other provinces are considered to have mountainous terrain. This classification variable is assigned a value of 1 for plain regions and 0 otherwise.

4. Empirical Analysis

4.1. Baseline Results

Table 3. Effect of CHSF on TFPG: baseline regression results.

Variables	Explained Variable: TFPG
	(1)	(2)
Highi × Iitpost	0.077 * (0.043)	0.083 ** (0.040)
CLAP		0.026 (0.022)
PS		0.096 (0.060)
PIGI		−0.145 (0.026)
DR		−0.036 ** (0.015)
RDD		0.003 (0.003)
FEA		−0.198 ** (0.074)
UR		0.001 (0.001)
IL		0.112 (0.067)
TIL		0.012 ** (0.005)
Province fixed effects	Yes
Year fixed effects	Yes
Number of observations	682	682
R2	0.370	0.463

Note: *, ** and indicate significance at the 10%, and 5% levels, respectively. Control variables include cultivated land area per capita (CLAP), planting structure (PS), per capita income from the grain industry (PIGI), disaster rate (DR), rainfall deviation degree (RDD), fiscal expenditure on agriculture (FEA), urbanization rate (UR), industrialization level (IL), and transport infrastructure level (TIL), as defined in Table 2.

presents the results of the baseline regression of the CHSF on the TFPG. Standard errors are clustered at the provincial level. The results show that the estimated coefficients of the interest variable High_i × I_it^post are all positive. Specifically, the estimated coefficient is 0.077, which is significant at the 10% level when no control variable is included. After adding in the control variables, the coefficient increases to 0.083, which is significant at the 5% level. These findings provide evidence that CHSF positively influences TFPG, thereby supporting Hypothesis H1.

4.2. Robustness Test Results

4.2.1. Parallel Trend Test Results

In Figure 4, after controlling for province- and year-fixed effects, the 95% confidence intervals for ${\mathsf{\beta }}_{\mathrm{t}}$ for all years up to the year of CHSF policy implementation (2011) include 0, which means that the CHSF effect on TFPG was not significant at the 5% level. This shows that the TFPG of the provinces satisfied the parallel trend testing before the implementation of the policy. In addition, there was no significant impact on TFPG in the first five years after the implementation of the CHSF policy, with a positive and significant impact occurring only in the sixth year and continuing in subsequent years.

4.2.2. Placebo Test Results

The regression coefficients and probability densities of CHSF are shown in Figure 5a,b for placebo test with 500 and 1000 repetitions, respectively. As shown in Figure 5, most of the regression coefficients for CHSF were approximately 0, indicating that the variables produced results that were not random, further validating the robustness of the results to support Hypothesis H1.

4.2.3. Other Robustness Tests Results

First, for the alternative standard errors, columns (1), (2), (3), and (4) of

Table 4. Robustness tests.

Variables	Ordinary Standard Error		Heteroskedasticity–Serial Correlation–Cross-Section Correlation Robust Standard Errors		Core Variables Lagged by One Period		Adjusting the Range of Sample Selection		Removing Other Policy Distractions
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
Highi × Itpost	0.077 *** (0.012)	0.083 *** (0.010)	0.077 * (0.043)	0.083 ** (0.040)	0.061 *** (0.012)	0.070 *** (0.012)	0.048 * (0.027)	0.060 ** (0.030)	0.055 ** (0.018)	0.038 ** (0.017)
Cons—	0.989 *** (0.010)	0.822 *** (0.058)	1.016 *** (0.007)	0.832 *** (0.087)	0.988 *** (0.009)	0.848 *** (0.058)	1.007 *** (0.003)	0.987 *** (0.117)	0.997 *** (0.011)	0.951 *** (0.085)
Control variables	No	Yes	No	Yes	No	Yes	No	Yes	No	Yes
Province fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	682	682	682	682	651	651	594	594	465	465
R2	0.370	0462	0.170	0.295	0.373	0.458	0.310	0.431	0.348	0.421

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Control variables include cultivated land area per capita (CLAP), planting structure (PS), per capita income from the grain industry (PIGI), disaster rate (DR), rainfall deviation degree (RDD), fiscal expenditure on agriculture (FEA), urbanization rate (UR), industrialization level (IL), and transport infrastructure level (TIL), as defined in Table 2.

show that the estimated coefficients of the core explanatory variables, High_i × I_t^post, remain positive and statistically significant above the 10% level, regardless of the standard error adjustments, confirming model robustness.

Second, after lagging the core variables by one period, columns (5) and (6) in Table 4 show that there is a positive impact of CHSF policy implementation on TFPG, which still remains significant at the 1% level after the control variables are added, addressing reverse causality issue.

Third, after adjusting the scope of sample selection, columns (7) and (8) in Table 4 show that the effect of High_i × I_t^post on TFPG remains positive and significant, supporting the robustness of the benchmark regression results.

Fourth, after removing other potential policy interference, columns (9) and (10) of Table 4 show that CHSF still has a statistically significantly positive impact on TFPG, supporting the robustness of the results.

4.3. Mechanism Analysis

4.3.1. Technological Change (TC)

Columns (1) and (2) of

Table 5. Effect of CHSF on TPFG: mechanism analysis.

Variables	TC				ASM
	APF		ML		ALSM: Number of Farm Households Operating More than 10 Acres of Arable Land		SSM: HHI
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Highi × Itpost	0.036 (0.053)	0.021 (0.053)	44.071 *** (7.796)	33.959 *** (7.299)	−17.855 (20.210)	−21.358 (19.939)	0.005 (0.018)	−0.003 (0.016)
Cons—	0.613 *** (0.042)	1.171 *** (0.266)	6.193 (6.355)	300.657 *** (36.521)	37.485 ** (16.476)	5.473 (99.764)	0.438 *** (0.014)	0.326 *** (0.081)
Control variables	No	Yes	No	Yes	No	Yes	No	Yes
Province fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	682	682	682	682	682	682	682	682
R2	0.455	0.488	0.596	0.672	0.730	0.757	0.931	0.946

Note: **, and *** indicate significance at the 5%, and 1% levels, respectively. Control variables include cultivated land area per capita (CLAP), planting structure (PS), per capita income from the grain industry (PIGI), disaster rate (DR), rainfall deviation degree (RDD), fiscal expenditure on agriculture (FEA), urbanization rate (UR), industrialization level (IL), and transport infrastructure level (TIL), as defined in Table 2.

present the regression results for the effect of CHSF on the allocation of production factors (APF). While the coefficients for CHSF are positive, they are not statistically significant. In contrast, columns (3) and (4) in Table 5 show positive and statistically significant coefficients for CHSF’s effect on mechanization level (ML), indicating that CHSF contributes to increased ML and, consequently, to TFPG, lending partial support to Hypothesis H2.

4.3.2. Agricultural Scale Management (ASM)

Columns (5) and (6) of Table 5 present the regression results for the impact of CHSF on agricultural land scale management (ALSM). The regression results show a negative coefficient for CHSF, but this effect is not statistically significant. Columns (7) and (8) show the effects of CHSF on service scale management (SSM), specifically the Herfindahl–Hirschman Index (HHI). The results indicate a negative coefficient for CHSF with the inclusion of control variables, but they are not statistically significant. These results suggest that the effect of CHSF on agricultural scale management may be limited or influenced by other unobserved factors. The implications of these results and their alignment with Hypothesis H2 will be further examined in the Section 5.

4.4. Heterogeneity Analysis

4.4.1. Differences in Functional Production Areas

The regression results in columns (1), (2), (3), and (4) of

Table 6. Heterogeneity analysis: by functional grain production area and topography.

Variables	TFPG
	Major Grain-Producing Area		Non-major Grain-Producing Areas		Plains Area		Non-Plain Areas
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Highi × Iitpost	0.027 (0.020)	0.050 ** (0.023)	0.064 *** (0.013)	0.068 *** (0.014)	0.118 *** (0.013)	0.110 *** (0.013)	0.059 ** (0.027)	0.068 ** (0.024)
Cons—	1.011 *** (0.011)	0.911 *** (0.047)	1.008 *** (0.010)	0.816 *** (0.042)	1.024 *** (0.009)	0.909 *** (0.043)	1.030 *** (0.009)	1.067 *** (0.058)
Control variables	No	Yes	No	Yes	No	Yes	No	Yes
Province fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	682	682	682	682	682	682	682	682
R2	0.027	0.654	0.064	0.535	0.302	0.417	0.031	0.620

Note: **, and *** indicate significance at the 5%, and 1% levels, respectively. Control variables include cultivated land area per capita (CLAP), planting structure (PS), per capita income from the grain industry (PIGI), disaster rate (DR), rainfall deviation degree (RDD), fiscal expenditure on agriculture (FEA), urbanization rate (UR), industrialization level (IL), and transport infrastructure level (TIL), as defined in Table 2.

indicate that CHSF has a significant positive effect on TFPG in both main and non-main grain production areas, with a more pronounced effect in non-main grain production areas.

4.4.2. Topographical Differences

The regression results in columns (5), (6), (7), and (8) of Table 6 show that CHSF positively impacts TFPG in both plain and non-plain areas, with a more substantial effect observed in plain regions, irrespective of the inclusion of control variables.

5. Discussion

This study investigates the impact of CHSF on TFPG, focusing on both its underlying mechanisms and regional heterogeneity. This section interprets the findings, explores their broader implications for agricultural modernization, and acknowledges the study’s limitations.

5.1. Discussion of Parallel Trend Test Results

The study identifies a policy lag in the impact of CHSF on TFPG. The parallel trend test shows that CHSF did not have a statistically significant impact on TFPG in the first five years after implementation, with positive and statistically significant effects emerging only in the sixth year and persisting thereafter. This pattern suggests a temporal delay in the policy’s effectiveness.

Several factors likely contribute to this lag. First, the construction and completion of CHSF infrastructure typically span multiple agricultural seasons. Initial years may involve planning, land leveling, irrigation system installation, and coordination among stakeholders. As a result, tangible productivity benefits are unlikely to materialize immediately.

Second, farmers may need time to adapt to the new production conditions enabled by CHSF, such as changes in crop structure, mechanization, or irrigation practices. Behavioral inertia, lack of immediate knowledge about how to utilize the upgraded land efficiently, and initial uncertainty about policy continuity may further delay productivity responses.

Third, many of the benefits associated with CHSF are inherently cumulative. Improvements in soil quality, water retention capacity, and yield stability under stress conditions may take several years to fully manifest. This is consistent with findings in the literature on long-term agricultural infrastructure investment, which often documents gradual returns rather than immediate impacts [52,53].

Taken together, these factors help explain the observed lag in productivity gains from CHSF investment, highlighting the importance of allowing adequate time for policy maturation and for policy impacts to fully emerge when evaluating long-term agricultural interventions.

5.2. Discussion on Mechanism Analysis Results

The mechanism analysis on the technological level offers valuable insights into how CHSF influences TFPG. The results indicate that CHSF positively impacts mechanization levels (MLs), suggesting that infrastructure improvements such as field remediation, soil enhancement, road construction, and farmland management create favorable conditions for mechanization. Improved mechanization enhances agricultural efficiency and contributes directly to TFPG. However, the effect of CHSF on the allocation of production factors (APF) was not statistically significant. This may be because CHSF primarily enhances the farmland infrastructure and production conditions to improve crop growing environments and yields, rather than directly influencing how production factors are allocated. The allocation of production factors often depends on farmers’ individual decisions and market conditions, which may not respond uniformly to infrastructure improvements alone.

The mechanism analysis of agricultural scale management (ASM) presents more complex results. The findings show a negative but statistically insignificant relationship between CHSF and both agricultural land scale management (ALSM) and the Herfindahl–Hirschman Index (HHI). This suggests that while CHSF improves farmland quality and production conditions, it may not necessarily promote larger-scale agricultural operations or industrial agglomeration.

One possible explanation is that CHSF’s primary objectives are to enhance farmland productivity, quality, yield, and sustainability, focusing on improvements such as crop variety, soil conservation, irrigation systems, and agricultural technology. These initiatives aim to boost per-acre productivity and optimize agricultural resources, rather than expanding farm size or fostering market concentration. Furthermore, large-scale farms may already have access to advanced technologies and have likely reached a certain operational scale with structured production systems, making them less sensitive to CHSF and infrastructure-driven improvements. Additionally, policy efforts may prioritize improving overall farmland quality and efficiency rather than expanding the number of large-scale farms. Resources might be more effectively allocated toward promoting innovative technologies, improving infrastructure, and providing training to enhance overall farm quality.

The statistically insignificant relationship between CHSF and HHI suggests that agricultural industrial agglomeration is influenced by factors beyond farmland conditions alone, which may include market demand, technological innovation, policy support, and infrastructure. CHSF addresses only one aspect of agricultural production, while effective industrial clustering relies on a broader set of interconnected factors, such as processing and marketing within the agricultural value chain.

Consequently, while the mechanism analysis shows that CHSF supports technological change (TC) by enhancing mechanization (ML), CHSF’s influence on agricultural scale management is limited and may depend on additional market and policy factors. Thus, these results provide only partial support for Hypothesis H2.

5.3. Discussion on Heterogeneity Analysis Results

The heterogeneity analysis underscores the importance of regional differences in the effectiveness of CHSF. The results reveal that CHSF has a more pronounced positive effect on TFPG in non-main grain-producing areas compared to main grain-producing areas. This disparity may reflect the fact that non-main grain-producing regions typically face greater constraints in land use, agricultural technology, and infrastructure. As a result, the marginal benefits of CHSF are higher in these areas, where improvements in infrastructure and mechanization can significantly enhance productivity. Conversely, main grain-producing areas may have already achieved higher levels of agricultural productivity and technological adoption, leading to smaller incremental gains from CHSF.

Similarly, the impact of CHSF on TFPG varies by topography. The findings indicate that CHSF generates stronger productivity gains in plain regions compared to mountainous areas. Plains generally offer more fertile soil and favorable conditions for large-scale mechanized farming, which enhances the effectiveness of CHSF-related infrastructure improvements. In contrast, the challenging terrain and limited arable land in mountainous regions may restrict the scope for mechanization and productivity gains, thereby dampening the impact of CHSF. This highlights the need for region-specific policy adjustments to account for geographical constraints and maximize the benefits of CHSF.

5.4. Comparison with Other Studies

The findings of this study align with similar research conducted in other countries, where infrastructure improvements and land consolidation have been shown to enhance mechanization and productivity. For example, studies in India and Vietnam have reported that land improvement policies increase agricultural productivity through mechanization and better land use [54,55]. However, some of the findings differ from previous studies on CHSF. For instance, this study found that the CHSF policy does not affect the allocation of production factors (APF) or agricultural scale management (ASM), which contradicts the findings of Gong and Zhang’s study [34]. This discrepancy may stem from differences in how the APF and ASM variables were measured or variations in the implementation of CHSF across provinces. A future comprehensive comparative analysis of similar policies in other developing countries could provide additional insights into the mechanisms driving productivity gains and the contextual factors influencing policy effectiveness.

5.5. Practical Applications and Institutional Implications

The study’s findings have important policy implications beyond China. The regional variation in CHSF’s effectiveness suggests that a one-size-fits-all approach may not be optimal. Policymakers should consider tailoring agricultural interventions to the specific needs and constraints of different regions. In non-main grain-producing areas, targeted support for mechanization and infrastructure improvements could yield higher productivity gains. In main grain-producing areas, policy efforts could focus on enhancing technological innovation and market access to sustain productivity growth. Similarly, in mountainous regions, infrastructure investments could prioritize improving access to markets and developing crop varieties suited to challenging terrain.

At the institutional level, strengthening local governance and coordination mechanisms could enhance the effectiveness of CHSF. Improved capacity-building for farmers, better alignment between central and local government agencies, and stronger financial and technical support systems could accelerate the adoption of infrastructure-driven improvements. Additionally, promoting public–private partnerships could enhance infrastructure development and improve market access, further amplifying the benefits of CHSF.

The findings suggest that infrastructure-driven agricultural modernization may be replicated in other developing countries with similar agricultural structures and policy frameworks. However, further validation in different geographical, economic, and institutional contexts is necessary. Future research could explore how variations in land ownership systems, agricultural labor markets, and technological capacity influence the effectiveness of CHSF-style policies.

5.6. Limitations

While the results provide robust evidence of CHSF’s positive impact on TFPG, this study has certain limitations.

First, although the study verified that CHSF can promote TFPG by increasing the mechanization level (ML), it did not account for other potential channels, such as technological innovation or labor migration, which may also influence TFPG. Future research could explore these additional pathways to offer a more comprehensive understanding of CHSF’s multifaceted impact.

Second, the heterogeneity analysis only considered functional grain production areas and topographical differences but omitted other factors such as climate variations, which may also affect regional responses to CHSF.

Third, the study relies on provincial-level data due to data limitations, which may mask local-level variations in CHSF implementation and impact. In practice, CHSF projects are implemented at the county or sub-county level and often exhibit heterogeneity in the scope, quality, and management. We acknowledge that the aggregated provincial data may not fully capture the localized differences.

Nonetheless, despite the data limitations, provincial-level analysis still provides a meaningful starting point for evaluating spatial policy effects. To improve spatial precision, future research could employ more granular datasets and adopt techniques such as geographically weighted regression (GWR), which allows for the detection of spatially varying relationships and localized spillover effects.

Additionally, incorporating spatial interdependencies could enhance the robustness and policy relevance of the analysis. The construction of CHSF in one province may influence neighboring regions through channels such as policy diffusion or migration-induced intervention. For instance, provinces may replicate successful CHSF practices due to inter-regional competition or administrative convergence, and infrastructure improvements in one region may attract agricultural investment or labor from adjacent areas. These spillovers could be systematically investigated using advanced spatial econometric models such as the Spatial Durbin Model (SDM) or the Spatial Autoregressive Model (SAR), which can capture both direct and indirect policy impacts. By integrating these spatial linkages, future studies could offer a more comprehensive and nuanced understanding of the broader regional implications of CHSF policies.

Finally, as the study focuses on China’s 31 provinces (cities and regions), the applicability of the findings to other countries with different land ownership regimes, governance structures, and agricultural conditions requires further validation.

6. Conclusions

This study builds on existing research by examining the impact of CHSF on TFPG, focusing on both its underlying mechanisms and regional heterogeneity. While previous research has explored the effects of CHSF on grain production capacity, few studies have assessed its influence on TFPG in China. This study provides empirical evidence that CHSF increased the TFPG by an average of 7%.

Broadly implemented in 2011, the CHSF policy represents a key reform in China’s agricultural sector. Leveraging this policy shift, this study examines the relationship between CHSF and TFPG using balanced panel data from 31 Chinese provinces (cities and regions) spanning 2000 to 2021. Specifically, it establishes a continuous DID model to mitigate endogeneity problem in policy evaluation and accurately identify causal relationships. The main findings are as follows:

(1)

At the provincial level, CHSF significantly contributes to TFPG. The paper further validates the findings of the benchmark regression and lagged policy using a parallel trend test.

(2)

The robustness of the results was demonstrated through a series of robustness tests: using core variables lagged by one period, adjusting for sample selection, excluding other policy disturbances, and conducting placebo tests.

(3)

The results of the mechanism analysis indicated that CHSF could enhance TFPG by increasing the mechanization level (ML), and thus TFPG, contributing to the theoretical framework of China’s rural revitalization policy by linking CHSF to productivity gains through mechanization.

(4)

Heterogeneity analysis reveals significant regional variation in the impact of CHSF across regions. The policy contributed more to TFPG in non-major grain-producing and plain areas than in major grain-producing and non-plain areas, suggesting the need for region-specific policy adjustments.