农业社会化服务创新试点政策能激励粮食生产吗OACHSSCD
Can the pilot policies for agricultural socialized service innovation incentivize grain production?A causal inference based on county-level quasi-natural experiments and double machine learning
保障国家粮食安全面临"谁来种地"的现实挑战,农业社会化服务被视为破题关键.2021年农业农村部启动的农业社会化服务创新试点政策,为评估服务政策成效提供了准自然实验场景.研究旨在评估该政策对县域粮食生产的激励效应、动态轨迹及作用机制.文中以2021年试点政策为准自然实验,基于2015-2023年200个县的面板数据,综合运用双重差分模型与双重机器学习方法进行因果推断.依据地理邻近、农业生态相似和经济发展水平相近原则,为100个试点县匹配100个非试点县,缓解系统误差和选择偏误.研究发现:1)农业社会化服务创新试点政策显著促进县域粮食生产,这一结论经稳健性检验后依旧成立.2)政策效果呈现短期脉冲特征,实施当年效应显著,但后续年份不显著,政策的激励机制尚不稳固.3)异质性分析表明,政策效应在中等收入地区最显著呈倒U型;在非粮食主产省的效应强于粮食主产省;在湿润半湿润地区显著,干旱半干旱地区不显著.4)机制检验显示,增产效应主要依赖规模扩张路径,而效率扩张路径未能发现.上述发现揭示,当前试点政策仍以劳动替代型服务为主,技术集成与知识溢出效应尚未充分释放.建议推动服务内涵从扩规模向提效率转型,实施差异化区域支持政策,并建立市场化长效机制,以夯实粮食安全根基.
Ensuring national food security is confronted with the practical challenge of"who will farm the land",and agricultural socialized services are regarded as the key to solving this problem.The agricultural socialized service innovation pilot policy initiated by the Ministry of Agriculture and Rural Affairs in 2021 provides a quasi-natural experimental scenario for evaluating the effectiveness of the service policy.Aiming to assess the incentive effect,dynamic trajectory,and mechanism of this policy on county-level grain production,this study takes the 2021 pilot policy as a quasi-natural experiment,uses a double difference model and double machine learning methods for causal inference based on panel data from 200 counties from 2015 to 2023.According to the principles of geographical proximity,similar agricultural ecology,and similar economic development levels,100 pilot counties are matched with 100 non-pilot counties to alleviate systematic errors and selection biases.The study finds that:1)The agricultural socialized service innovation pilot policy significantly promotes county-level grain production,and this conclusion remains valid after robustness tests.2)The policy effect shows a short-term impulse characteristic,being significant in the year of implementation but not in subsequent years,indicating that the incentive mechanism of the policy is not yet stable.3)Heterogeneity analysis shows that the policy effect is the most significant in medium-income regions,presenting an inverted U-shaped curve,stronger in non-grain-producing provinces than in grain-producing provinces,and significant in humid and semi-humid regions but not in arid and semi-arid regions.4)Mechanism tests reveal that the production increase effect mainly relies on the scale expansion path,while the efficiency expansion path has not been discovered.These findings suggest that the current pilot policy still mainly focuses on labor-replacement services,and the effects of technology integration and knowledge spillover have not been fully released.It is recommended to promote the transformation of service connotations from scale expansion to efficiency improvement,implement differentiated regional support policies,and establish a market-oriented long-term mechanism to consolidate the foundation of food security.
邓国权;熊媛媛;余国新;朱娜
新疆农业大学经济管理学院,乌鲁木齐 830000新疆农业大学经济管理学院,乌鲁木齐 830000新疆农业大学经济管理学院,乌鲁木齐 830000新疆农业大学经济管理学院,乌鲁木齐 830000
管理科学
粮食生产政策激励效应双重差分模型双重机器学习
grain productionpolicy incentive effectdouble difference modeldouble machine learning
《干旱区资源与环境》 2026 (7)
32-44,13
国家自然科学基金项目(72163032)资助.
评论