EN中文

Journal of Resources and Ecology >

2025 , Vol. 16 >Issue 4: 919 - 932

DOI: https://doi.org/10.5814/j.issn.1674-764x.2025.04.001

Ecosystems and Ecosystem Services

The Diffusion Characteristics of China’s Policy of Converting Cultivated Land into Forest (Grassland)

  • LI Na , 1, 2, 3 ,
  • WANG Shuting , 1, 2, 3, * ,
  • WU Xinnian 1, 2, 3 ,
  • MA Yue 1, 2, 3
Expand
  • 1. Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou 730000, China
  • 2. Key Laboratory of Knowledge Computing and Intelligent Decision, Gansu Province, Lanzhou 730000, China
  • 3. Department of Information Resources Management, School of Economics and Management, University of Chinese Academy of Sciences, Beijing 100190, China
*WANG Shuting, E-mail:

LI Na, E-mail:

Received date: 2024-12-20

  Accepted date: 2025-02-10

  Online published: 2025-08-05

Supported by

Philosophy and Social Science Planning in Gansu Province(2023ZD006)

Fold

Abstract

China’s Convert Cultivated Land into Forest and Grassland policy has been gradually emphasized and promoted in the context of advancing China’s Western Development Program. Therefore, based on the perspective of localized practice in China, this study bridges the semantic level of information from the perspective of qualitative analysis and quantitative coding through the qualitative analysis of policy content and text similarity metrics. The study systematically reveals the diffusion characteristics of China’s Convert Cultivated Land into Forest and Grassland policy with respect to spatiotemporal evolution, thematic focus, and the degree of central policy diffusion. The goals are to clarify the mechanisms of policy evolution in a long time span, to analyze the implementation effects of the policy in accordance with local conditions, and thereby to make strong contributions to policy making. The results show five important aspects of the policy diffusion. (1) The policy followed a tendency characterized by “slow-rapid-stable” stages, and its diffusion process can be summarized into four phases. (2) By integrating thematic diffusion characteristics and spatiotemporal evolution characteristics, this study reveals that the Grain for Green policy primarily exhibited a top-down “Hierarchical diffusion mode”. (3) By combining the national land cover change with the geographical distribution data of the Grain for Green policy, this study reveals that the policy has largely achieved its ecological goal of converting cultivated land. (4) Based on empirical research, this study illustrates the relationships between the number of policy issuances, the similarity between central and provincial policies, and the degree of policy diffusion, thereby enriching diffusion theory based on China's localized practical research. (5) This study suggests that national policies appear to reduce vertical pressure, thus inspiring the innovation of regional policy. Meanwhile, efforts should focus on developing distinctive industries to promote improvements in quality and efficiency.

Key words： Grain for Green; policy diffusion; spatiotemporal evolution; qualitative content analysis; text similarity

Cite this article

LI Na , WANG Shuting , WU Xinnian , MA Yue . The Diffusion Characteristics of China’s Policy of Converting Cultivated Land into Forest (Grassland)[J]. Journal of Resources and Ecology, 2025 , 16(4) : 919 -932 . DOI: 10.5814/j.issn.1674-764x.2025.04.001

1 Introduction

The Grain for Green policy, implemented in China in the late 1990s, is a national strategic decision for the long-term and sustainable development of the country. This policy is a significant measure for improving the environment and building an ecological civilization. It aims to address soil erosion and the frequency of natural disasters, fix carbon and increase carbon sinks, and respond to climate change by reducing the area of farmland with low ecological carrying capacity (Duan et al., 2024). China has positively implemented and promoted the Grain for Green policy over the years. Gansu, Sichuan, and Shaanxi were selected as the first pilot provinces in 1999. Since then, a series of national and provincial laws and regulations have been carried out. Benefit from financial support, ecological compensation, and technological assistance, they have provided comprehensive policy and resource guarantees to achieve the objectives of the Grain for Green policy (Liu et al., 2024). The white paper Twenty Years of China’s Convert Cultivated Land into Forest and Grassland (1999-2019), published by the National Forestry and Grassland Administration in 2020, said that: “The Grain for Green policy made a great contribution of more than 4% of the net green growth area globally between 1999 and 2019. It is likely to be a landmark project in the context of China’s ecological civilization construction” (National Forestry and Grassland Administration, 2020). However, the ecological protection and restoration in China is still confronted with several serious challenges. The 2023 edition of the China Soil and Water Conservation Bulletin, released by the Ministry of Water Resources, states that although China has struggled to maintain the tendency of “dual decline” in the area and intensity of soil erosion and the reduction of hydraulic and wind erosion areas, the hydraulic erosion on sloped farmland remains a prominent problem (Ministry of Water Resources of the People’s Republic of China, 2024). That study found that different provinces in the same geographical area often have varying levels of effectiveness in implementing erosion control under the strong control from the national government. For example, Hubei Province had far more area to treat for mild, moderate and severe soil erosion than Hunan Province during 2022-2023 (Ministry of Water Resources of the People’s Republic of China, 2024). That study concluded that the existing issues may be related to the diffusion model of policies across regions. Therefore, this study attempts to elucidate the mechanism of policy evolution through the lens of policy diffusion theory, thereby uncovering the focal points of policy implementation. The aim is to explore the regional differences in policy implementation effects in accordance with local conditions and provide recommendations for both policy formulation and implementation.
Policy Diffusion theory is derived from disciplines such as communication studies, sociology, and information science and others, and it refers to the process in which a policy implemented by government A is subsequently adopted by government B within the same system. It describes the phenomenon where the policy choices of one government are influenced by the policy choices of other governments (Podshibyakina, 2021). Policy Diffusion began in the late 1960s in the United States. Western countries have developed and refined the theoretical framework and analytical system for Policy Diffusion in the subsequent five decades, so now the theory and analytical systems appear to have gradually matured in the development of western countries (Bao, 2021). Compared with western countries, this model began to emerge in China after the country initiated its government performance management practices in the 1990s (Liu, 2020). In 2004, Mao (2004) introduced the “Policy Diffusion Framework” as an important Western policy theory to the Chinese academic community. Chinese scholars gradually adopted Policy Diffusion theory and began conducting related research in the following years. Policy diffusion includes two main types of research, spatiotemporal evolution and content diffusion, which help to trace the policy evolution process and diffusion pathways, respectively. For example, Boushey (2010) summarized various non-incremental temporal diffusion patterns, such as “steep S-shaped”, “R-shaped” and “staircase” curves, based on the policy’s own attributes and the external diffusion environment. In the context of China’s COVID-19 prevention and control efforts, Wang and Zhang (2021) identified a “staircase” curve in the interprovincial diffusion of health codes. Based on Policy Diffusion theory, American scholars Shipan and Volden (2006) revealed that the smoking bans in U.S. states exhibit a bottom-up spatial diffusion characteristic, while Chinese scholars He and Zheng (2023) used a diffusion model to demonstrate that China’s climate policies show both top-down and neighboring effects in their spatial diffusion.
However, Chinese scholars have paid relatively little attention to the application of this theory so far, and the Policy Diffusion methodology developed in the Western cultural context is not fully applicable to China's political development environment. At the same time, existing theoretical research still has some shortcomings, including a disconnection between Policy Diffusion and the policy process (Fundytus et al., 2023), unclear definitions of the outcomes of Policy Diffusion (Zhou et al., 2019), a lack of focus on the policy diffusion influence of provincial governments (Lou et al., 2023), and a lack of localized practical research in China (Bao, 2022). Therefore, this study takes Grain for Green as a typical case. As a pilot policy with a standardized implementation process and a complete engineering system, it is an important blueprint for the study of the localization of policy diffusion theory in China. With the help of quantitative research paradigm for policy, this study explores the temporal and spatial diffusion characteristics of the Grain for Green policy. The major task of this study was to analyze the content and implementation effects of the policy in each province with respect to the diffusion pattern of the policy from a long time-series perspective. Meanwhile, the qualitative analysis of policy content and statistical methods are strong tools for distinguishing the relationships between two factors, the volume of policy issuance and the similarities between the central and provincial policy texts, and the degree of policy diffusion. This study is likely to be a foundational work for subsequent research on localization practices with a new perspective. Furthermore, this study explores the policy diffusion method as it applies to Chinese localities, and provides new ideas for subsequent studies on the diffusion characteristics of China’s large-scale environmental policies, which is of profound practical significance.

2 Data and methodology

2.1 Data sources

Data Retrieval Scope: The Grain for Green Program was initiated in 1999. In 2000, the government issued the Notice on Forwarding the National Development and Planning Commission’s Report on Preliminary Ideas for the Implementation of the Western Development Strategy (2000) (Li and Zhu, 2002) and held the first meeting of the State Council’s Leading Group for Western Regional Development. At that point, the Chinese government officially included Grain for Green as a key component of China’s Western Expansion. Therefore, the time frame for the policy text publication was set as January 1, 1999, to May 1, 2023. The data sources for this study primarily include policy and regulatory documents from the central Ministry of Ecology and Environment and the government websites of provinces, municipalities, and autonomous regions, supplemented by databases such as the WanFang Legal Database, Lawxin Database, and Peking University Law Database. The search keywords were “Convert Cultivated Land into Forest” OR “Convert Cultivated Land into Forest and Grassland” OR “Convert Cultivated Land” OR “Convert Cultivated Land into Grassland” OR “Land Degradation Control” OR “Vegetation Restoration” OR “Ecological Restoration” OR “Ecological Compensation” OR “Soil and Water Conservation” OR “Desertification Control” OR “Sustainable Land Management” OR “Ecological Agriculture” OR “Forest Restoration” OR “Grassland Restoration” OR “Land Use Planning” OR “Farmland Conversion”.
Data Selection Criteria: To ensure the validity of the policy text data and research findings, the following criteria were applied: 1) Exclude texts with low relevance to the Grain for Green content; 2) Exclude irrelevant data types such as notifications, approvals, replies, reports, and decisions; and 3) Compare different versions of policies published by various government websites and select the most complete policy text, while retaining the forwarding documents from provincial governments regarding central policies. A total of 326 valid policy texts were obtained after manual screening, including 45 central policies and 281 local policies at the provincial level.

2.2 Research methodology

Policy diffusion theory is a theory of how policy or project innovations spread (Chen, 2014), including the temporal and spatial diffusion characteristics, content diffusion characteristics, policy diffusion influencing factors, and policy diffusion mechanism analysis. Among them, the study of temporal diffusion characteristics analyzes the evolution of policies over time after their adoption, including the “S-shaped” curve characteristic of progressive growth (De Tarde, 1903) and the diffusion law of “steep S-type, R-type and Step-type” of non-gradual growth (Boushey, 2010). The study of spatial diffusion characteristics is set to find the diffusion effects of policies in the geographic and spatial scopes, including the Hierarchical diffusion mode, Absorptive radiation diffusion mode, Horizontal diffusion mode, and Follow-up policy diffusion mode (Wang and Lai, 2013). The content diffusion characteristics analyze the heterogeneity of policy content and the degree of policy diffusion, including thematic diffusion characteristics and the degree of diffusion.
The Chinese policy texts are characterized by clear logic, accurate language, plain expression, a solemn style, and other features. Consequently, this study employed a word vector-based method to calculate the similarity between policy texts, using the Term Frequency-Inverse Document Frequency (TF-IDF) algorithm to compute the cosine similarity between central and provincial policy texts, to judge the associations between the policy texts at different levels. This type of method is widely used in Chinese policy text similarity calculation, and it has high accuracy and relative maturity, but there is still the problem of insufficient semantic revelation of the policy texts (Wei et al., 2024). This study qualitatively analyzed the policy text data at each stage using Nvivo12 software, with the aim of creating coding units of policy texts for the purpose of thematic categorization in order to mine the policy focus and its core categories (Hu et al., 2019). This study combined the qualitative policy content analysis method with the text similarity metric to bridge the semantic level of information from the perspectives of qualitative analysis and quantitative coding in order to explore the content diffusion features of the policies at a deeper level.
In the process of qualitative content analysis, the policy texts were encoded with keywords. First, open coding was used to identify the attributes and dimensions of conceptual categories. Then, axial coding was applied to organize and establish connections between those conceptual categories. Subsequently, selective coding was employed for further integration and refinement, with the aim of identifying the core categories of keywords and uncovering the evolutionary trends and diffusion characteristics of policy themes. In the policy diffusion analysis, central and provincial policies were first classified according to the primary issuing departments, and policies from departments with similar attributes were then subjected to a similarity comparison analysis. Finally, the mean similarity score for each province was calculated, the calculation formula is as follows.
S¯=∑i=19992023Si/25
where S¯ represents the average similarity between central and provincial policies from 1999 to 2023, while ∑i=19992023Si denotes the total sum of all similarity values over the 25 years. This formula aims to explore the extent of central policy diffusion from a long-term temporal perspective (Figure 1).
Figure 1 Research framework of this study

3 Diffusion characteristics of the Grain for Green policy

3.1 Temporal diffusion follows the characteristic pattern of “slow-rapid-stable”

The policy texts were sorted according to their first publication date, to acquire the full picture of Grain for Green. In total, 20 provinces, four municipalities directly under the central government and five autonomous regions had deployed relevant policies on Grain for Green during the 25 year period from 1999 to 2023. The provinces and regions not represented were Fujian Province, Guangdong Province, Taiwan Province, Hong Kong Special Administrative Region and Macao Special Administrative Region. Therefore, based on the temporal dimension of policy diffusion, the cumulative number of policy texts released by the 29 provincial governments in different years was statistically analyzed (Figure 2), revealing a classic “S-shaped” diffusion curve. This indicates that the diffusion trend of the Grain for Green policy among provincial governments followed the characteristic pattern of “slow-rapid-stable” (Rogers, 1983; De Tarde, 1903). Accordingly, the diffusion trend of the Grain for Green policy is inferred to be currently in the late stage of the “S-shaped” curve, i.e., the steady diffusion phase.
Figure 2 Time series of the cumulative number of provincial-level policies for returning farmland to forests
Figure 2 shows that the cumulative number of the Grain for Green documents in China increased from 1999 to 2005, with the fastest growth in 2006-2010, which decelerated in 2011-2013 and further decelerated in 2014-2023. Furthermore, key policy milestones included the pilot work of the Grain for Green in 1999, the national spread of the Grain for Green in 2002, the new goal for the Grain for Green from 2014 to 2020, and the new target of improving quality and efficiency by 2030 proposed in 2023. Therefore, Grain for Green can be divided into four phases according to the curve inflection points and landmark policy events in the Policy Accumulators.
(1) Phase I: The period of embryonic development (1999-2001)
To ensure the sustainable development of the economy and society in the western regions, the Chinese government initiated pilot projects for the Grain for Green policy in 1999 in Sichuan, Shaanxi, and Gansu provinces. The policy measures were initially designed to transfer the farmland to forest and grass, food-for-work, and individual contracting. This was the starting point for the Grain for Green Project. In 2001, the State Council included Grain for Green in the Outline of the Tenth Five-Year Plan for National Economic and Social Development of the People's Republic of China (2001), greatly promoting the pilot projects for the conversion of cultivated land into forest. The release of these policy measures pushed provincial governments across the country to implement the Convert Cultivated Land into Forest and Grassland program. However, at the local level, only the 10 provincial governments of Guizhou, Henan, Hubei, Qinghai, Shandong, Shaanxi, Sichuan, Beijing, Shanghai, and the Guangxi Zhuang Autonomous Region issued clear pilot policy measures for the Convert Cultivated Land into Forest and Grassland program. This phase lacked systematic and comprehensive policy planning across the country, with only a few governments leading the implementation. Therefore, this period can be regarded as the period of embryonic development prior to the formal diffusion of the policy.
(2) Phase II: The period of slow diffusion (2002-2005)
In January 2002, the State Council announced the nationwide launch of the Grain for Green Program via a video conference, marking the formal commencement of the policy diffusion phase. In April 2002, the State Council issued the agreement Several Opinions on Further Improving the Policy Measures for Converting Cultivated Land into Forest (Grassland) (2002), which clarified the principles and relevant policy measures in the following years. In December of the same year, the Regulations on Converting Cultivated Land into Forest (2002) were officially promulgated and set to be enforced from January 20, 2003. Compared with the period of embryonic development, this phase saw the release of more detailed policy measures, and over the next four years, more detailed policies and measures were released during this stage. The speed of proliferation gradually accelerated within the four years of this period. The survey revealed that a total of 21 provincial governments released policies and measures on the Grain for Green. The number of policy texts increased from 21 to 68. During Phase II, the Grain for Green Project experienced the transformation from ‘demonstration and pilot project’ to the slow proliferation period of ‘central policy guidance’.
(3) Phase III: The period of rapid diffusion (2006-2013)
In 2006, the National Development and Reform Commission issued the Notice on Requesting the Submission of Relevant Information on Converting Cultivated Land into Forest (Grassland) (2006) to comprehensively understand the status of Grain for Green implementation across various regions, with the aim of further studying and improving related policies. In 2010, the Ministry of Finance issued the Notice on Issuing the Budget Management Measures for Financial Funds of Converting Cultivated Land into Forest (2010), which sought to enhance the management of financial resources, improve the integrity of the budget, and accelerate the budget execution process. Compared with the period of slow diffusion of policies, more detailed management methods for financial subsidies and policy measures for acceptance and verification of the results of the project were released at this stage. The proliferation speed increased dramatically within this eight-year period. In Phase III, 23 provincial governments released policies and measures for the Grain for Green Project, and the number of policy texts launched increased from 68 to 171. This Phase marked the transition of the Grain for Green policy from the “central policy guidance” stage of slow diffusion to the “simultaneous verification and development” stage of rapid diffusion.
(4) Phase IV: The period of stable diffusion (2014-2023)
The Chinese government have attached great importance to the work of the Grain for Green Project. Since the 18th National Congress of the CPC incorporated the construction of ecological civilization into the overall layout of the ‘Five-in-one’, the goal of ‘stabilizing and expanding the scope of returning farmland to forests and grasslands’ has been taken as one of the key tasks for comprehensively deepening the reforms of the 18th CPC Central Committee. Since 2014, several national government documents have required consolidating the results of the Grain for Green Project and expanding the scale of returning farmland to forests and grasslands. The Notice on Issuing the Overall Plan for the New Round of Converting Cultivated Land into Forest and Grassland (2014), issued by the National Development and Reform Commission, proposed that by 2020, approximately 42.4 million acres of sloped farmland and severely desertified farmland nationwide would be converted into forest and grassland. This marked the initiation of a new round of the Converting Cultivated Land into Forest and Grassland project in China. In 2023, the State Forestry and Grassland Administration issued the Notice from the National Forestry and Grassland Administration on Enhancing the Quality and Efficiency of the Converting Cultivated Land into Forest and Grassland Program (2023), which proposed a new target to fully complete the two rounds of the program and improve its quality and efficiency by 2030. This formally concluded the construction of the new rounds of Grain for Green Projects that began in 2014. Compared with the previous stage, the policies in this phase were more systematic and comprehensive. A series of policies were designed to solve several long-standing issues, such as the “sustainable livelihood mechanisms for returning farmers” and the “small scale and slow progress of the Grain for Green Projects”. Over the following decade, the diffusion rate gradually slowed down, with 20 provincial-level governments issuing policy measures for the Grain for Green Project. The number of policy documents published decreased from 171 in the previous phase to 66 in this phase. However, the policy transitioned from a phase of “concurrent assessment and development” in the rapid diffusion period to a phase of “consolidating results” in the stable diffusion period.

3.2 The spatial diffusion primarily presents a top-down “Hierarchical diffusion mode”

Considering the time divisions of each phase, the missing data of forest land, and farmland data in 2023, the data for 2001, 2005, 2013, and 2022 were used to analyze the relationship between farmland area and the number of policy releases (Figure 3), as a reference for evaluating diffusion effects①. Based on a geographical division of Northwest, Northeast, Northern China, Central China, Eastern China, Southern China, and Southwest, the geographical distribution characteristics of the policy of returning cultivated land to forests was explored in terms of spatial diffusion by taking into account the releases of policies at different stages of diffusion.
Figure 3 Relationships between cultivated land area and policy release quantity in relevant provinces

3.2.1 Phase of embryonic development: Follow-up policy diffusion mode

In this stage, the Grain for Green Project exhibited the features “Follow-up policy diffusion mode”. First, in 1999, the Chinese government initiated pilot projects for the Grain for Green Program in Sichuan, Shaanxi, and Gansu provinces. Second, in March 2000, the government expanded the pilot projects to 13 provinces (autonomous regions and municipalities) and 174 counties (townships and state farms) along the middle and upper reaches of the Yellow River and Yangtze River. Third, following discussions and reviews of the pilot projects, the government summarized the experiences and issued the Several Opinions on Further Improving the Pilot Work for Converting Cultivated Land into Forest and Grassland (2000) in September 2000 to improve the policy. Finally, in August 2001, the State Forestry Administration established the Convert Cultivated Land into Forest and Grassland Project Management Center to systematically promote and supervise the implementation of the project. In summary, the whole process of the embryonic development period followed the logical pattern of ‘experimentation-absorption-propagation’, which is consistent with the ‘Follow-up policy diffusion mode’ among regions at different levels of development.

3.2.2 Phase of slow policy diffusion: Hierarchical diffusion mode

Under the Guidance of Central Policy, the Regulations on Converting Cultivated Land into Forest (2002), issued by the State Council in December 2002, served as a systematic document issued by the central government to guide the nationwide implementation of the Grain for Green Project. Concerning regional diffusion in this stage, provincial governments developed their policies based on the central government’s requirements, in line with the hierarchical features of a “leader-follower” relationship. This exhibited a top-down “Hierarchical diffusion mode”. In terms of policy diffusion outcome evaluation, since the goals of the Grain for Green policy are to control soil erosion, reduce soil degradation, and combat land degradation, nationwide land cover change was used as a common evaluation indicator.
During this period, under the premise of national land reclamation and development, the total area of cultivated land decreased by 5.53 million ha. At the same time, the area of forest land increased by 6.55 million ha. Therefore, the planned objectives for the conversion of cultivated land into forest were achieved overall, and the provinces that adopted the policy innovation all reached varying degrees of success in implementing the goals of the Grain for Green policy.

3.2.3 Phase of rapid policy diffusion: Hierarchical diffusion mode and Horizontal diffusion mode

During the period of rapid diffusion, the Grain for Green policy exhibited both a top-down “Hierarchical diffusion mode” and a “Horizontal diffusion mode” between regions or departments at the same level. 1) In the top-down “Hierarchical diffusion mode”, provincial governments primarily used central policies as the decision-making blueprint for the Grain for Green policy, exhibiting a “leader-follower” characteristic. For example, in the Northwest region, Qinghai Province implemented measures to improve the Grain for Green in 2007 based on the State Council’s Notice on Improving the Policy for Converting Cultivated Land into Forest (2007) and local practices. 2) For the “Horizontal diffusion mode” at the same level, the ‘proximity effect’ of policy innovation clustering occurred among provincial governments in some neighboring regions. For example, in the Northwest region, after the large-scale snow and ice storms in January 2008, Gansu and Shaanxi Province issued the Notice on Strengthening Post-Disaster Recovery, Reconstruction, and Achievement Consolidation in the Converting Cultivated Land into Forest Program (2008) in May and June, respectively. Both provinces implemented post-disaster recovery, reconstruction, and consolidation of results for the Grain for Green Project. In terms of evaluating the outcome of policy diffusion, the area of arable land nationwide increased by 13080700 hectares under the premise of national land reclamation and development. However, the area of forested land maintained its growth trend, with its increase of 17512800 ha being 2.67 times the increase in the area of forested land during the period of slow diffusion of the policy. Overall, the planned objectives for the project were achieved, and the provinces adopting policy innovations showed varying degrees of success in meeting their targets when conducting checks and evaluations of their implementation results.

3.2.4 Phase of stable policy diffusion: Hierarchical diffusion mode

In the stage of stable diffusion, provincial governments formulated policies in accordance with the requirements of central government policies. This was consistent with the “leader-follower” hierarchical characteristics, presenting a top-down “Hierarchical diffusion mode” driven by the central government to promote the adoption of policy innovations by provincial governments. In evaluating the policy diffusion results, under the premise of adhering to the arable land red line and improving the quality of cultivated land through national land reclamation and consolidation, the total cultivated land area nationwide declined by 7.5834 million ha during the phase of stable diffusion. At the same time, the area of forest land continued to show an upward trend, with an increase of 30.2921 million ha, which was 1.73 times the increase in forest land during the phase of rapid diffusion. Overall, the goal of the Grain for Green Project was achieved, and the provinces that adopted policy innovations demonstrated varying degrees of success in meeting their goals for the Grain for Green Project during the inspection and acceptance of their outcomes.

3.3 Thematic diffusion characteristics

3.3.1 Phase of embryonic development: Hierarchical diffusion mode

The national government primarily guides local governments in conducting pilot practices for the Grain for Green policy, so the concerns of both are highly compatible. From the perspective of policy themes, both central and local policies focus on “subsidy measures” and “inspection and acceptance” (Table 1). Furthermore, the national government focuses more on macro-guidance of local pilot practices while the local governments are more concerned with how to implement the action according to the local conditions and consider technical training and technical contracting as priorities. In terms of the comparative analysis of policy themes, the ‘seedling fee subsidies’ for the level of provincial governments has broken down into subsidy measures and ‘inspection and acceptance’. Therefore, policy diffusion during this period exhibited a top-down ‘cascading diffusion pattern’ in which the central government guided the implementation of policies at the local level.
Table 1 Central and provincial policy priorities
Period Central government Local government
Indicator Keywords Indicator Keywords
Embryonic development period Subsidy measure Food subsidies; Provide food aid instead of relief Subsidy measure Food subsidies; Arrange grain sources locally and proximally
Reduction and exemption of agricultural taxes Reduction and exemption of agricultural taxes
Financial subsidies Cash subsidies; Subsidy for seedling costs
Examine and acceptance County-level self-inspection; Provincial-level re-inspection; National-level verification Examine and
acceptance		 Standard; Quality
Standard Supervision; Re-inspection
Regional
development		 Policy pilot Policy pilot Combined with reality
Management Technology contracting responsibility system; Technical training
Slow
diffusion period		 Subsidy measure Cash subsidies Subsidy measure Cash subsidies
Food subsidies; Cash subsidy disbursement Food subsidies
Subsidy for seedling-based afforestation Subsidy for seedling-based afforestation
Examine and acceptance Standard; Supervision; Quality Examine and
acceptance		 Quality-specific inspection; Standard
Seedling quality; Task design; File management Provincial-level re-inspection; Data archiving
Fund
management		 Fund allocation and reimbursement system Fund management Hierarchical management; Hierarchical
responsibility; Designated funds for specific purposes
Designated funds for specific purposes Supervisory management
Technology Technical promotion; Training; Practical
technology		 Regional
development		 Land contracting
In accordance with local conditions; Technology demonstration site Technology contracting responsibility
system; Technical promotion
Information management Statistical reporting system Economic development; Ecological
conservation
Information network construction In accordance with local conditions
Rapid
diffusion period		 Subsidy measure Financial subsidies Subsidy measure Cash subsidies, Food subsidies
Subsidy standards; Subsidy period Subsidy standards; Subsidy period
Examine and acceptance Standard; Quality Examine and
acceptance		 Standard; Quality
Provincial-level inspection and acceptance; National-level key inspection and acceptance Enhance supervision
Fund
management		 Separate accounting; Designated funds for
specific purposes		 Regional
development		 Economic forest construction; Clean energy construction; Industrial structure adjustment
Strengthen supervision and inspection Technology contracting responsibility
system; Technical training
Disbursement of funds based on the verification results In accordance with local conditions
Stable
diffusion period		 Subsidy measure Food subsidies; Financial subsidies Subsidy measure Specialized funds subsidies
Subsidy for seedling afforestation costs Cash subsidies
Examine and acceptance Remote sensing interpretation Examine and
acceptance		 Provincial-level re-inspection
On-site verification standard; Approach
Engineering design Field survey; Office design Technology
application		 Technical training; Technical promotion
Document preparation; Filing and
implementation		 Technical guidance and services
File management Unified leadership; Hierarchical management High-quality development
Digitalization, informatization, and networking construction Regional
development		 Economic forest construction
Consolidation of achievements Extension of the subsidy period Industrial base construction
Precise management; Strengthen responsibility implementation Protection of the rights and interests of
farmers returning cultivated land to forests

3.3.2 Phase of slow policy diffusion: Hierarchical diffusion mode and absorptive radiation diffusion mode

This study found that central policies were further improved through the summarization of pilot practice experiences, while local policies, under the macro-level guidance of central policies, were refined according to specific provincial conditions. From the perspective of themes, both central and local policies focused on three key areas: “subsidy measures”, “inspection and acceptance” and “fund management”. On this basis, the central government prioritized “technology” and “information management”, proposing the formulation of scientific and technological support programs and the implementation of scientific and technological safeguards in accordance with local conditions, as well as strengthening the information networks at all levels of government throughout the country. On the other hand, local governments prioritized “regional development”, through measures such as giving priority to the development of energy projects in fallow areas, adjusting the industrial structure to build new pillars of the industrial system, and developing planting, breeding and processing industries in accordance with local conditions. In the comparative analysis of policy themes, after the embryonic development phase, local governments refined “seedling cost subsidies” and the central government mentioned “seedling subsidies” in the Regulations on Converting Cultivated Land into Forest (2002), thus, showing a bottom-up “absorbing and diffusing pattern”. At the same time, while archiving the information, the central policy of self-inspection was still implemented at the county level, reviewed at the provincial level, and verified at the national level during the period of embryonic development, reflecting the central government’s macro-guidance of the implementation of the policy by the local governments. In summary, during this period, policy diffusion exhibited both a top-down Hierarchical diffusion mode and a bottom-up Absorptive radiation diffusion mode.

3.3.3 Phase of rapid policy diffusion: Hierarchical diffusion mode

The central government identified the management and use of funds as a key issue for special rectification, while local governments developed more detailed and comprehensive regional development policies. From the perspective of policy themes, national and local policies both focused on “subsidy measures” and “inspection and acceptance”. The difference is that the national policies took “fund management” on the outstanding place, aiming to consolidate the results of the Grain for Green Project, while the local government regarded “regional development” in the first place, seeking to balance the development of the regional economy and the environment. In the comparative analysis of policy themes, the central policy emphasized the inspection and acceptance process, combining “comprehensive provincial-level inspections” with “key national-level inspections”. Local policies, under the macro-guidance of the central government, organized self-inspections and coordinated with provincial and central government inspection efforts. Therefore, at this stage, policy diffusion exhibited a top-down Hierarchical diffusion mode, with guidance from the central government driving the local implementation of policies.

3.3.4 Phase of stable policy diffusion: Hierarchical diffusion mode

Based on the successive achievements of the Converting Cultivated Land into Forest (Grassland) policy goals, the central government initiated the planning for a new round of the Converting Cultivated Land into Forest and Grassland project. Guided by the central government’s macro-level policies, local governments assigned new tasks for the project while consolidating previous outcomes. In the overall analysis of policy themes, both central and local policies focused on “subsidy measures” and “inspection and acceptance”. On this basis, the central government prioritized “achievement consolidation, project design, and archival management”, while local governments emphasized “technical application” and “regional development”. Local governments also highlighted “protecting the rights and interests of returning farmers” and allocated funds to difficult-to-reclaim areas and their farmers. The aim was to strengthen the management and protection of reforested areas, thereby promoting the high-quality development of the Converting Cultivated Land into Forest and Grassland project. In the comparative analysis of policy themes, the central policy emphasized that national-level inspections should combine “remote sensing interpretation” and “on-site verification”, while local governments strictly implemented the central government’s regulatory measures for inspection and acceptance. Therefore, the policy diffusion during this period displayed a top-down “Hierarchical diffusion mode” guided by the central government in the implementation of the policy by local governments.

3.4 Extent of central policy diffusion

The empirical study of Wu and Dong (2024) proved that the textual similarity of central and provincial policies is inversely proportional to the degree of central policy diffusion, while the number of provincial policy issuances is positively proportional to the degree of central policy diffusion. Meanwhile, the range for the data in this study was divided according to the average similarity and the number of policy releases (Figure 4). Since the distribution of policy similarity is in the range of [0.271,0.717], those in the range of [0.271,0.499] are classified as low similarity, those in the range of [0.500,0.599] are classified as average similarity, and those in the range of [0.600,0.717] are classified as high similarity. Since the number of policy releases is in the range of [1,76], the distribution of values indicates a low number of releases in the range of [1,5], an average number of releases in the range of [6,29], and a high number of releases in the range of [30,76].
Figure 4 Textual similarity between central and provincial policies
An analysis combining regional divisions and the number of policy publications revealed the key findings for four regional groups. 1) Northwest Region: Shaanxi demonstrates a relatively low similarity, while Gansu, Ningxia, Qinghai, and Xinjiang exhibit moderate similarity. The overall degree of policy innovation and diffusion is high.
Notably, although Shaanxi has the highest number of policy publications, it shows a low similarity to central policies, indicating that the province exhibits high levels of innovation and diffusion regarding the Grain for Green policy. 2) Northeast, Northern China, Central China, and Southern China Regions: Policies in these regions are mostly either forwarded from central policies or based on central guidelines tailored to the provinces. This suggests that the overall levels of policy innovation and diffusion are relatively low. 3) Eastern China Region: Jiangsu and Zhejiang show relatively low similarity, Anhui shows moderate similarity, and Jiangxi, Shandong, and Shanghai show high similarity. Among them, Jiangsu, Zhejiang, Shandong, and Shanghai have each published only one policy, indicating a low degree of policy diffusion. 4) Southwest Region: Chongqing demonstrates moderate similarity, while Sichuan, Guizhou, Yunnan, and Tibet show high similarity. Notably, although Tibet published fewer policies, its policy similarity is the highest. The two policies in Tibet were consistent with the guidelines issued by the State Council and the National Development and Reform Commission, suggesting low levels of policy innovation and diffusion. Based on the empirical study of the Grain for Green policy, the relationships between the number of policy publications, the similarity of central and local policy texts, and the degree of policy innovation and diffusion are shown in Table 2.
Table 2 Relationships between policy publication volume, similarity between central and provincial policies, and the degree of policy diffusion
Number of policy documents issued Policy similarity between central and local governments Degree of policy diffusion
Limited Moderate Extensive Relatively low Moderate Relatively high
√ √ Moderate
√ √ Relatively low
√ √ Low
√ √ Relatively high
√ √ Moderate
√ √ Relatively low
√ √ High
√ √ Relatively high
√ √ Relatively low

Note: If there is a correlation between variables, it is indicated with a √ in the table; otherwise, the cell is left in blank.

First, to verify the relationship between the number of policy publications, the similarity of central and local policies, and the degree of policy diffusion, the two nominal variables—number of publications and central-local policy similarity—were assigned values of 1, 2, and 3 in ascending order. This transformation aimed to convert them into numerical variables for subsequent analysis. The results (Table 3) show a significantly high positive correlation between the number of publications and the degree of policy diffusion, while the similarity between central and local policies exhibits a significantly high negative correlation with the degree of diffusion. These results confirm the conclusions presented in Table 2. Second, the multiple regression model is statistically significant with a good fit. Specifically, the variance inflation factors (VIF) for “number of publications” and “central-local policy similarity” are both 1.004, indicating no severe multicollinearity between the variables, which allows for the construction of the regression model (Berry and Berry, 1990). The P-values for “number of publications” and “central-local policy similarity” are both less than 0.001, which is below the 0.05 significance level, demonstrating that both independent variables are statistically significant and the overall regression model is valid. The adjusted R2 value is 0.708, meaning that the independent variables account for 70.8% of the variation in the “degree of policy diffusion”, indicating a good model fit. Finally, the regression analysis results (Table 4) reveal that “number of publications” significantly positively predicts “policy diffusion degree”, while “central-local policy similarity” significantly negatively predicts “policy diffusion degree”, further confirming the conclusions shown in Table 2. In summary, the relationships between the number of publications, the similarity of central and local policies, and the degree of policy diffusion align with the findings presented in Table 2.
Table 3 Verification by correlation analysis
Variable Number of policy
documents issued		 Policy similarity between central and local governments Degree of policy
diffusion
Number of policy documents issued 1 -0.065 0.660**
Policy similarity between central and local governments -0.065 1 -0.583**
Degree of policy diffusion 0.660** -0.583** 1

Note: ** indicates a significant correlation at the 0.01 level.

Table 4 Verification by linear regression model
Variable Regression coefficient Standard error Statistical significance VIF value
Number of policy documents issued 0.039 0.006 <0.001 1.004
Policy similarity between central and local governments -4.439 0.838 <0.001 1.004
Constant term 4.664 0.486 <0.001

Note: The dependent variable is “Degree of policy diffusion”.

4 Discussion and conclusions

4.1 Discussion

The findings above lead to five main conclusions. 1) To promote the effective diffusion of central policies, local governments should be guided to formulate policies in accordance with local conditions, thereby reducing the barriers caused by the similarity between central and local policies. 2) By considering regional specificities and the macro guidance of central policies, local governments should be encouraged to create diverse, targeted policies, which will help increase the degree of central policy diffusion. 3) Reducing vertical pressure from the central government on local governments and encouraging mutual learning among local governments within similar geographical regions can promote the agglomeration of policy innovation in neighboring areas. 4) Focusing on the effective utilization of local resource endowments, developing region-specific industries based on local conditions, and fostering emerging industries such as forest rehabilitation and eco-tourism can improve quality and efficiency, effectively consolidating the results of the Grain for Green policy. 5) Over the 25 years of implementing the Grain for Green policy, measures such as policy subsidies and the development of ecological industries have not only effectively driven regional economic development but have also provided valuable practical experience for the innovation and diffusion of other domestic policies, demonstrating its significant policy spillover effects.
Based on these conclusions and inferences, the following recommendations are proposed for the future formulation of Grain for Green policies.
(1) Integrating Regional Resource Endowments and Development Stages. The formulation of central policies and their implementation at the local level must be grounded in the specific conditions of each region and take into account the local economic, political, historical, social, cultural, and geographical factors to develop feasible the Grain for Green policy proposals. Key factors include regional natural resource conditions, ecological environmental status, and socio-economic development. Efforts should be made to protect ecologically fragile areas and focus on dynamic poverty reduction in key impoverished regions based on the assessment of relevant indicators across different regions. Given that China is currently in the stable diffusion phase of the Grain for Green policy, future policy formulation should focus on consolidating policy achievements, expanding the scope of implementation, and improving quality and efficiency.
(2) Regional Cooperation to Drive Policy Innovation. Given China’s vast territory and diverse natural geographical conditions, excessive reliance on the macro guidance of central policies may hinder the achievement of region- specific ecological land retirement goals. Therefore, actively encouraging innovative cooperation among provinces with similar natural geographic conditions is essential. This cooperation should involve the flow and sharing of resources such as capital, technology, and talent, to optimize resource allocation and improve resource use efficiency. To establish a long-term cooperation mechanism, regular regional seminars should be held to exchange effective policy practices and innovative technologies, while jointly addressing challenges such as environmental pollution, ecological protection, and poverty alleviation. This collaborative effort can promote balanced regional economic development through policy synergies, thus enhancing overall competitiveness.
(3) Empowering Industrial Transformation with New Types of Productive Forces. Through the application of technological innovations, industrial structure adjustments, and market mechanism guidance, it is possible to effectively promote the deep transformation of industries and further the objectives of the Grain for Green policy. First, innovations in biotechnology, genetic improvement techniques, and modern agricultural technologies such as smart farming can reduce dependence on arable land. Second, the development of green industries such as forestry, grassland management, and eco-tourism aims to replace traditional agriculture and increase the income of farmers. In addition, the establishment of new agricultural models, such as ecological agriculture, can enhance agricultural value added. Finally, the establishment and improvement of ecological product market systems and the development of carbon trading markets can transform the ecological benefits of the Grain for Green policy into economic benefits.
Compared to existing research, the integration of qualitative and quantitative methods in this study addresses a major gap in current literature, which tends to focus more on instantaneous diffusion nodes. This approach based on information science research methods provides new insights for future studies. From the perspective of localized research on China’s practices, this study helps fill the gap in research related to policy innovation diffusion theories as they are applied to China. In addition, given the lack of attention in existing studies to the role of provincial governments in policy diffusion, this study explores the degree of policy diffusion by comparing the similarities and differences in the focal points of central and provincial governments at different diffusion stages. Furthermore, the current research on the evaluation of diffusion results based on land cover change data lacks precision. Future studies should incorporate methods such as artificial intelligence and machine learning to improve the accuracy of diffusion result evaluations.

4.2 Conclusions

The policy diffusion characteristics of China’s typical environmental policy, the grain to green project, were measured by qualitative content analysis and similarity metrics, which systematically revealed the diffusion characteristics of that policy. This study was constructed on the foundation of the process of that policy, and deeply defined the relationship between the innovation diffusion behavior of central and provincial governments. The results indicated four key aspects of the policy diffusion. 1) In terms of the temporal dimension, the Grain for Green policy in China exhibits an “S-shaped” diffusion curve, following the diffusion trend of “slow-rapid-stable”. Specifically, this process can be divided into four stages: the embryonic development period, the slow diffusion period, the rapid diffusion period, and the stable diffusion period. 2) Combining the spatial-temporal evolutionary characteristics with thematic diffusion features revealed that during the embryonic development period, the policy followed a “Follow-up policy diffusion mode” and a top-down “Hierarchical diffusion mode”; during the slow diffusion period, the diffusion followed a “Hierarchical diffusion mode” and a bottom-up “Absorptive radiation diffusion mode”; during the rapid diffusion period, the diffusion exhibited a “Hierarchical diffusion mode” and a “Horizontal diffusion mode” between regions or departments at the same level; and during the stable diffusion period, it followed a “Hierarchical diffusion mode”. This reflects the central government's dominant role in policy formulation and diffusion. 3) Combining national land cover changes with regional distribution data of the Grain for Green policy revealed that the policy has essentially achieved its ecological land retirement goals. 4) The empirical research and statistical analysis results confirmed that the “number of publications” is significantly positively correlated with the “degree of central policy diffusion”, while “central-local policy similarity” is significantly negatively correlated with “degree of central policy diffusion”. This suggests that when provincial governments actively respond to central policies during the policy formulation process, the diffusion effect is enhanced. However, when provincial governments are constrained by the central policy framework and fail to adapt it to local conditions, the innovation diffusion of the policy is inhibited.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Bao W H. 2021. A review of policy diffusion theory studied abroad: Trends, attentions and prospects. Journal of Inner Mongolia University (Philosophy and Social Sciences), 53(4): 82-89. (in Chinese)

[2]
Bao W H. 2022. Research on the types of horizontal intergovernmental policy diffusion. Local Governance Research, (4): 2-13, 77. (in Chinese)

[3]
Berry F S, Berry W D. 1990. State lottery adoptions as policy innovations: An event history analysis. American Political Science Review, 84(2): 395-415.

[4]
Boushey G. 2010. Policy diffusion dynamics in America. Cambridge, UK: Cambridge University Press.

[5]
Chen F. 2014. Evolvement of public policy diffusion. Chinese Public Administration, (6): 99-104. (in Chinese)

[6]
De Tarde G. 1903. The laws of imitation. New York, USA: Henry Holt.

[7]
Duan X M, Han M, Kong X L, et al. 2024. Spatiotemporal evolution and simulation prediction of ecosystem carbon storage in the Yellow River Basin before and after the Grain for Green Project. Environmental Science, 45(10): 5943-5956. (in Chinese)

[8]
Fundytus K, Santamaria-Plaza C, McLaren L. 2023. Policy diffusion theory, evidence-informed public health, and public health political science: A scoping review. Canadian Journal of Public Health, 114(3): 331-345.

[9]
He Y J, Zheng S M. 2023. Paths and mechanisms of climate policy diffusion in China. China Population, Resources and Environment, 33(9): 26-37. (in Chinese)

[10]
Hu Y P, Liu Q L, He J X. 2019. Influencing factors of intelligence intervention in the prevention of decision failure—Qualitative analysis based on Nvivo 11. Library and Information Service, 63(11): 80-87. (in Chinese)

[11]
Li J, Zhu X H. 2002. Seven books closely related to the ecological economy. Ecological Economy, (6): 75-78. (in Chinese)

[12]
Liu W. 2020. The theory, practice and development of policy diffusion. Beijing, China: Scientific and Technical Documentation Press. (in Chinese)

[13]
Liu Y W, Deng N, Yan J J, et al. 2024. Evaluation of the ecological effect of returning farmland to forest and grassland in the former concentrated contiguous impoverished area of China. Acta Ecologica Sinica, 44(11): 4719-4732. (in Chinese)

[14]
Lou S, Sun Z, Zhang Y. 2023. To join the top and the bottom: the role of provincial governments in China’s top-down policy diffusion. Journal of Chinese Governance, 8(2): 161-179.

[15]
Mao S L. 2004. The theoretical development path of western public policy and its implications for localization research. Jiangsu Social Sciences, (1): 143-148. (in Chinese)

[16]
Ministry of Water Resources of the People’s Republic of China. 2024. 2023 China soil and water conservation bulletin. http://www.mwr.gov.cn/sj/tjgb/zgstbcgb/202403/t20240329_1708287.html. Viewed on 2024-09-25. (in Chinese)

[17]
National Forestry and Grassland Administration. 2020. China has returned farmland to forests and grasslands for 20 years (1999-2019). https://www.forestry.gov.cn/html/main/main_195/20200630085813736477881/file/20200630090428999877621.pdf. Viewed on 2023-02-01 (in Chinese)

[18]
Podshibyakina T A. 2021. Policy diffusion: Theory, concept, practice. Tomsk State University Journal of Philosophy, Sociology and Political Science, 61: 254-262.

[19]
Rogers E M. 1983. Diffusion of innovations. New York, USA: Free Press.

[20]
Shipan C R, Volden C. 2006. Bottom-up federalism: The diffusion of antismoking policies from US cities to states. American Journal of Political Science, 50(4): 825-843.

[21]
Wang F S, Zhang H P. 2021. Research on explosive policy diffusion under the background of major public crisis—Analysis of event history based on inter-provincial diffusion of health code. E-Government, (1): 21-31. (in Chinese)

[22]
Wang P Q, Lai X J. 2013. A study on the model and mechanism of public policy diffusion in China. Journal of Peking University (Philosophy and Social Sciences), 50(6): 14-23. (in Chinese)

[23]
Wei W, Ding X X, Guo M X, et al. 2024. A review of text similarity calculation methods. Computer Engineering, 50(9): 18-32. (in Chinese)

DOI

[24]
Wu K B, Dong Y. 2024. Study on the diffusion characteristics of China’s artificial intelligence policies integrating the innovation value chain and text relevance mining. Journal of Modern Information, 44(3): 12-22. (in Chinese)

[25]
Yang J, Huang X. 2021. The 30 m annual land cover dataset and its dynamics in China from 1990 to 2019. Earth System Science Data, 13(8): 3907-3925.

DOI

[26]
Zhou Y N, Huang S, Song X M. 2019. Literature review on policy diffusion and a research agenda. East China Economic Management, 33(5): 150-157. (in Chinese)

Options
Outlines

/

〈 〉