数字技术能提升粮食生产韧性吗?——基于双重机器学习的因果推断——许佳彬等《智慧农业(中英文)》2026年第3期
核心期刊论文,运用双重机器学习方法探究数字技术对粮食生产韧性的因果效应,方法新颖,数据详实,具有较高学术价值,但时效性较低。
许佳彬、秦亚茹、温浩洋、仇焕广、崔钊达在《智慧农业(中英文)》2026年第3期发表《数字技术能提升粮食生产韧性吗?——基于双重机器学习的因果推断》论文。研究聚焦数字技术对粮食生产韧性的因果效应,基于中国省级面板数据,运用双重机器学习(DML)方法有效解决高维控制变量与潜在混杂问题,准确估计数字技术对粮食生产韧性的处理效应。研究发现,数字技术显著提升粮食生产韧性,主要通过要素重组效应、数据要素与传统生产要素的协同配置等中介机制实现;数字技术的赋能效应呈现区域异质性,在粮食主产区和数字基础设施较好的地区效应更为显著。基于研究结论,建议构建数字农业社会化服务新体系,依托数字平台的规模效应,有效弥合小农户与现代农业之间的数字鸿沟,推动粮食生产韧性持续提升。
欢迎您访问《智慧农业(中英文)》官方网站!English
Toggle navigation智慧农业
Smart Agriculture ›› 2026, Vol. 8 ›› Issue (3): 176-189.doi: 10.12133/j.smartag.SA202601011
数字技术能提升粮食生产韧性吗?——基于双重机器学习的因果推断
许佳彬 1, 秦亚茹 1, 温浩洋 1, 仇焕广 2, 崔钊达 3(
)
-
- 东北农业大学经济管理学院,黑龙江 哈尔滨 150030,中国
-
辽宁大学经济学院,辽宁 沈阳 110036,中国
-
山东农业大学经济管理学院,山东 泰安 271018,中国
收稿日期:2026-01-08出版日期:2026-05-30基金项目:国家自然科学基金项目(72503031); 中国博士后科学基金面上项目(2025M772469); 黑龙江省自然科学基金项目(LH2024G001); 黑龙江省哲学社会科学基金项目(25JYC030); 山东省自然科学基金青年项目(ZR2025QC734)作者简介:许佳彬,博士,副教授,研究方向为农业绿色转型。E-mail:[email protected]通信作者:崔钊达,博士,讲师,研究方向为粮食安全。E-mail:[email protected]
Can Digital Technology Improve the Resilience of Grain Production: Causal Inference Based on Dual Machine Learning
XU Jiabin 1, QIN Yaru 1, WEN Haoyang 1, QIU Huanguang 2, CUI Zhaoda 3(
)
-
- College of Economics and Management, Northeast Agricultural University, Harbin 150030, China
-
School of Economics, Liaoning University, Shenyang 110036, China
-
College of Economics and Management, Shandong Agricultural University, Taian 271018, China
Received:2026-01-08Online:2026-05-30Foundation items:National Natural Science Foundation of China(72503031); General Project of China Postdoctoral Science Foundation(2025M772469); Natural Science Foundation of Heilongjiang Province(LH2024G001); Heilongjiang Province Philosophy and Social Science Fund Project(25JYC030); Shandong Province Natural Science Foundation Youth Project(ZR2025QC734)About author:XU Jiabin, E-mail: [email protected]Corresponding author:CUI Zhaoda, E-mail: [email protected]
在线阅读 4
知网下载
本地下载 20
摘要/Abstract
摘要:
【目的/意义】 运用数字技术提升粮食生产韧性对保障粮食安全、构建现代化农业保障体系具有重要意义。本研究旨在探究数字技术对粮食生产韧性的影响效应、内在机制及异质性特征,为数字技术赋能粮食生产韧性提升提供参考。 【方法】 采用2011—2023年中国30个省份的面板数据,采用熵值法构建粮食生产韧性综合评价指标体系并测算其水平。选取支持向量机作为基学习器的双重机器学习模型识别数字技术对粮食生产韧性的因果影响。运用中介效应模型验证数字技术通过“人-地-服”三条机制路径实现要素重组的驱动作用。通过异质性检验,分析粮食产区、地形起伏度及经济发展水平对该赋能效应的调节差异。 【结果和讨论】 数字技术显著促进了粮食生产韧性提升,且在不同细化维度呈现结构性差异。数字技术显著增强了调节适应力与转型创新力,但受技术依赖与资本挤占效应影响,对风险抵抗力表现出负向影响。机制分析表明,数字技术通过“人-地-服”三条路径实现要素重组,即促进劳动力非农转移带来的资本回流、降低交易成本推动的土地规模经营,以及发挥技术创新集成作用的农业生产性服务。异质性分析发现,该赋能效应在地形起伏度较高、经济发展水平较低及粮食产销平衡区与主销区更为显著。 【结论】 应促进数字技术集成以打通全产业链数据链,创新“数字反哺”横向补偿机制以弥补省际数字鸿沟,培育数字化新型农业社会化服务体系,并因地制宜提升自然条件劣势地区的数字化短板,从而多路径提升粮食生产韧性。
关键词:粮食生产韧性,数字技术,要素重组,双重机器学习,支持向量机
Abstract:
[Objective] Leveraging digital technology to enhance grain production resilience is pivotal for ensuring food security and constructing a modernized agricultural support system. The purposes of this study are to: (1) Define the connotation of grain production resilience and construct an evaluation index system across three dimensions: risk resistance, regulatory adaptability, and transformative innovation; (2) Identify the impact and mechanisms of digital technology as a new production factor on grain production resilience; (3) Reveal how digital technology influences resilience through the reorganization of "People-Land-Service" factors; (4) Analyze the heterogeneity of these effects across diverse geographical environments and economic endowments. [Methods] Based on panel data from 30 Chinese provinces spanning 2011 to 2023, a rigorous empirical analysis was conducted. First, the entropy method was employed to measure the levels of digital technology and grain production resilience. Second, a dual machine learning model, utilizing support vector machine as base learners with 5-fold cross-validation and double residualization, was constructed for causal inference to address the "curse of dimensionality" and specification biases inherent in traditional linear models. To ensure robustness, an instrumental variable approach was applied, using the product of the number of fixed-line telephones in 1984 and the national internet users of the previous year. Third, a mediation effect model was introduced to test three pathways: non-farm labor transfer, large-scale land management, and agricultural productive services. [Results and Discussions] Mechanism analysis indicated: (1) In the "People" dimension, digital technology promoted non-farm labor transfer by reducing information search costs, thereby inducing capital reflux to alleviate financial constraints; (2) In the "Land" dimension, digital platforms lowered transaction costs for land transfer, facilitating contiguous large-scale farming and improving the marginal efficiency of digital equipment; (3) In the "Service" dimension, digital technology empowered agricultural service organizations, enhancing resilience through technology diffusion and professional intervention. Heterogeneity analysis showed that the empowerment effect was more pronounced in grain production-marketing balanced zones, high-relief terrains, and economically underdeveloped regions. This highlighted the "gap-filling" nature of digital technology, where precision tools like drones and remote sensing compensated for the limitations of traditional machinery in complex terrains. Notably, while digital technology significantly drove overall resilience, it primarily bolstered regulatory adaptability and transformative innovation while exerting a significant negative impact on initial risk resistance. [Conclusions] Digital technology is not a mere superposition of factors; rather, it catalyzes a fundamental transformation in the dynamics of grain production through the systemic restructuring of elements. To this end, differentiated strategies should be implemented: Major grain-producing areas should prioritize full-chain data integration, whereas production-marketing balanced zones and ecologically disadvantaged regions should focus on overcoming digital infrastructure bottlenecks. Furthermore, a horizontal "digital feedback" compensation mechanism should be innovated, encouraging major consumption areas to shift from "blood transfusion" (passive aid) to "blood-making" (capacity building) through technology licensing and targeted talent support. Finally, a new system of digitalized agricultural social services should be cultivated, leveraging the scale effects of digital platforms to effectively bridge the gap between smallholder farmers and modern agriculture.
Key words:grain production resilience,digital technology,restructuring of elements,dual machine learning,support vector machine
中图分类号:
引用本文
许佳彬, 秦亚茹, 温浩洋, 仇焕广, 崔钊达. 数字技术能提升粮食生产韧性吗?——基于双重机器学习的因果推断[J]. 智慧农业(中英文), 2026, 8(3): 176-189.
XU Jiabin, QIN Yaru, WEN Haoyang, QIU Huanguang, CUI Zhaoda. Can Digital Technology Improve the Resilience of Grain Production: Causal Inference Based on Dual Machine Learning[J]. Smart Agriculture, 2026, 8(3): 176-189.
使用本文
[收藏文章](javascript:;) /[推荐](javascript:void(null))
导出引用管理器EndNote|Ris|BibTeX
链接本文:https://www.smartag.net.cn/CN/10.12133/j.smartag.SA202601011
https://www.smartag.net.cn/CN/Y2026/V8/I3/176
[图/表 10](javascript:;)
图1
粮食生产韧性概念示意图

图1
图2
数字技术提升粮食生产韧性的机制框架图

图2
表1
数字技术影响粮食生产韧性的变量含义与描述性统计
| 变量类别 | 变量名称 | 变量赋值 | 平均值 | 标准差 | 最大值 | 最小值 |
|---|---|---|---|---|---|---|
| 被解释变量 | 粮食生产韧性 | 熵值法计算 | 0.188 | 0.068 | 0.544 | 0.079 |
| 核心解释变量 | 数字技术 | 熵值法计算 | 0.157 | 0.010 | 0.704 | 0.029 |
| 机制变量 | 劳动力非农转移/% | (农村从业人员-第一产业从业人员)/农村从业人员×100% | 0.043 | 0.019 | 0.106 | 0.012 |
| 土地规模经营 | 家庭承包耕地流转总面积取对数 | 6.967 | 0.512 | 7.839 | 5.216 | |
| 农业生产性服务/% | 农林牧渔专业及辅助性活动产值/农林牧渔业总产值×100% | 0.042 | 0.019 | 0.106 | 0.012 | |
| 控制变量 | 城镇化率/% | 城镇人口/总人口×100% | 0.606 | 0.120 | 0.896 | 0.350 |
| 地区产业结构/% | (第二产业增加值+第三产业增加值)/地区生产总值×100% | 0.905 | 0.052 | 0.998 | 0.748 | |
| 农产品价格指数 | 采用上一年度农产品生产价格指数(指数,上年=100) | 4.646 | 0.059 | 4.955 | 4.472 | |
| 农村人均用电量/(kw·h/人) | 农村用电总量/乡村人口数 | 6.913 | 0.979 | 10.628 | 4.897 | |
| 公路里程数/(万km) | 区域内公路通车总里程 | 15.674 | 8.424 | 41.830 | 1.210 | |
| 农田有效灌溉率/% | 有效灌溉面积/农作物播种面积×100% | 0.440 | 0.178 | 1.234 | 0.172 | |
| 自然灾害受灾率/% | 农作物受灾面积/农作物播种面积×100% | 0.139 | 0.181 | 2.980 | 0.001 | |
| 人均可支配收入 | 农村居民人均可支配收入取对数 | 10.088 | 0.443 | 11.349 | 9.044 |
表1
表2
粮食生产韧性评价指标体系构建
| 一级指标 | 二级指标 | 二级指标解释 | 属性 |
|---|---|---|---|
| 风险抵抗力 | 粮食产量波动率/% | 当年粮食总产量-过去5年粮食总产量移动平均值 | |
| 粮食减损潜力/(台/万亩) | 粮食每万亩脱粒机械年末拥有量 | + | |
| 耕地资源水平/(千hm 2/万人) | 耕地面积/总人口 | + | |
| 粮食病虫草鼠害受灾挽回情况/% | 粮食病虫草鼠灾害挽回数量/粮食病虫草鼠灾害受灾数量 | + | |
| 调节适应力 | 农业劳动生产率/(元/人) | 农林牧渔业总产值/第一产业从业人员数 | + |
| 粮食生产综合机械化水平/% | 粮食作物机耕率×0.4+机播率×0.3+机收率×0.3 | + | |
| 土地生产率/(万t/千hm 2) | 粮食总产量/粮食播种面积 | + | |
| 转型创新力 | 种质创新水平/个 | 农业植物新品种权申请数量 | + |
| 科技研发潜力/(人/千hm 2) | 公有经济(国有)企事业单位农业技术人员/农作物播种面积 | + | |
| 科技研发投入/% | 区域研究与试验发展经费投入强度/国内生产总值 | + |
表2
表3
数字技术评价指标体系构建
| 一级指标 | 二级指标 | 二级指标解释 | 属性 |
|---|---|---|---|
| 数字通信基础 | 农村有线广播电视普及率/% | 农村有线(广播)电视用户占家庭总户数比重 | + |
| 互联网业务普及率/% | 开通互联网业务的行政村占行政村总数比重 | + | |
| 基础通信网络水平/km | 长途光缆线路总长度 | + | |
| 计算机普及率/(台/每百户) | 农村居民平均每百户年末家用计算机拥有量 | + | |
| 智能手机普及率/(部/每百户) | 农村居民平均每百户年末移动电话拥有量 | + | |
| 数字应用设施载体 | 数字基地规模/个 | 区域认定的淘宝村数量 | + |
| 物联网应用支撑/km | 农村寄递物流长度 | + | |
| 农业气象观测能力/个 | 农村气象观测业务站个数 | + | |
| 邮政服务网点分布/(个/处) | 区域内邮政营业网点数量 | + | |
| 数字创新转化服务 | 农业数字创新产出/个 | 农业科技相关专利申请数量 | + |
| 数字金融广度(指数) | 数字普惠金融指数覆盖广度 | + | |
| 数字金融深度(指数) | 数字普惠金融指数使用深度 | + |
表3
表4
数字技术影响粮食生产韧性的基准回归结果
| 变量名称 | (1)粮食生产韧性 | (2)粮食生产韧性 | (3)风险抵抗力 | (4)调节适应力 | (5)转型创新力 |
|---|---|---|---|---|---|
| 数字技术 | 0.234***(0.039) | 0.309***(0.032) | -0.195***(0.027) | 0.895***(0.066) | 0.671***(0.065) |
| 控制变量一次项 | YES | YES | YES | YES | YES |
| 控制变量二次项 | NO | YES | YES | YES | YES |
| 年份固定效应 | YES | YES | YES | YES | YES |
| 省份固定效应 | YES | YES | YES | YES | YES |
| 观测值 | 390 | 390 | 390 | 390 | 390 |
表4
表5
数字技术影响粮食生产韧性的稳健性检验结果
| 变量名称 | (1)缩尾1% | (2)缩尾5% | (3)剔除直辖市 | (4)1∶2 | (5)1∶7 | (6)弹性网络回归 | (7)套索回归 |
|---|---|---|---|---|---|---|---|
| 数字技术 | 0.327*** (0.028) | 0.441*** (0.027) | 0.252*** (0.025) | 0.305*** (0.033) | 0.307*** (0.031) | 0.413*** (0.096) | 0.416*** (0.097) |
| 控制变量一次项 | YES | YES | YES | YES | YES | YES | YES |
| 控制变量二次项 | YES | YES | YES | YES | YES | YES | YES |
| 年份固定效应 | YES | YES | YES | YES | YES | YES | YES |
| 省份固定效应 | YES | YES | YES | YES | YES | YES | YES |
| 观测值 | 390 | 390 | 338 | 390 | 390 | 390 | 390 |
表5
表6
数字技术影响粮食生产韧性的内生性检验结果
| 变量名称 | Ⅳ:历史期1984年每百人固定电话数 | Ⅳ:滞后一期数字技术 |
|---|---|---|
| 数字技术 | 0.600*** (0.075) | 0.308*** (0.034) |
| 控制变量一次项 | YES | YES |
| 控制变量二次项 | YES | YES |
| 年份固定效应 | YES | YES |
| 省份固定效应 | YES | YES |
| 观测值 | 390 | 360 |
表6
表7
数字技术影响粮食生产韧性的机制检验结果
| 变量名称 | 粮食生产韧性 | 非农劳动力转移 | 土地规模经营 | 农业生产性服务 |
|---|---|---|---|---|
| 数字技术 | 0.309*** (0.032) | 0.935*** (0.091) | 0.874*** (0.152) | 0.051*** (0.014) |
| 控制变量一次项 | YES | YES | YES | YES |
| 控制变量二次项 | YES | YES | YES | YES |
| 年份固定效应 | YES | YES | YES | YES |
| 省份固定效应 | YES | YES | YES | YES |
| 观测值 | 390 | 390 | 390 | 390 |
表7
表8
数字技术影响粮食生产韧性的异质性分析结果
| 变量名称 | 粮食生产区 | 地形起伏度 | 经济发展水平 |
|---|---|---|---|
| 主产区 | 主销区 | 平衡区 | 高 |
| 数字技术 | 0.272***(0.038) | 0.351***(0.056) | 0.386**(0.113) |
| 控制变量一次项 | YES | YES | YES |
| 控制变量二次项 | YES | YES | YES |
| 年份固定效应 | YES | YES | YES |
| 省份固定效应 | YES | YES | YES |
| 观测值 | 169 | 91 | 130 |
表8
[参考文献 38](javascript:;)
[1]李国祥. 新时代国家粮食安全的目标任务及根本要求——学习习近平关于国家粮食安全论述及十九届六中全会相关精神的体会[J]. 中国农村经济, 2022(3): 2-11.
LI G X. On the policy goals and fundamental requirements of national food security in a new era: To learn from General Secretary Xi Jinping's exposition on national food security and the relevant spirit of the sixth plenary session of the 19 th CPC central committee[J]. Chinese Rural Economy, 2022(3): 2-11.
[2]丁存振, 徐宣国. 国际粮食供应链安全风险与应对研究[J]. 经济学家, 2022(6): 109-118.
DING C Z, XU X G. Research on security risks of international food supply chain and countermeasures[J]. Economist, 2022(6): 109-118.
[3]鲍国良, 姚蔚. 我国粮食生产现状及面临的主要风险[J]. 华南农业大学学报(社会科学版), 2019, 18(6): 111-120.
BAO G L, YAO W. Current situation and main risks of grain production in China[J]. Journal of South China Agricultural University (Social Science Edition), 2019, 18(6): 111-120.
[4]MARTIN R. Regional economic resilience, hysteresis and recessionary shocks[M]// Papers in Evolutionary Economic Geography (PEEG). Utrecht University, Department of Human Geography and Spatial Planning, Group Economic Geography, 2010.
[5]HODBOD J, EAKIN H. Adapting a social-ecological resilience framework for food systems[J]. Journal of Environmental Studies and Sciences, 2015, 5(3): 474-484.
[6]HUANG Y, REN F W, WANG Y W. Evaluation and pathways for achieving agricultural resilience under the framework of climate-smart agriculture[J]. Humanities and Social Sciences Communications, 2026, 13: 105.
[7]COOPMANS I, BIJTTEBIER J, MARCHAND F, et al. COVID-19 impacts on Flemish food supply chains and lessons for agri-food system resilience[J]. Agricultural Systems, 2021, 190: 103136.
[8]朱满德, 张青. 农业生产性服务与粮食生产韧性: 影响机制与实证检验[J]. 湖南农业大学学报(社会科学版), 2024, 25(6): 1-11.
ZHU M D, ZHANG Q. Agricultural productive services and grain production resilience: Influencing mechanisms and empirical tests[J]. Journal of Hunan Agricultural University (Social Sciences), 2024, 25(6): 1-11.
[9]LAMICHHANE P, MILLER K K, HADJIKAKOU M, et al. Resilience of smallholder cropping to climatic variability[J]. Science of the Total Environment, 2020, 719: 137464.
[10]李少林, 马里. 国家现代农业示范区建设影响粮食生产韧性的作用机制及政策效应[J]. 深圳大学学报(人文社会科学版), 2025, 42(3): 5-15.
LI S L, MA L. The mechanism and policy effects of national modern agricultural demonstration zone construction on grain production resilience[J]. Journal of Shenzhen University (Humanities Social Sciences), 2025, 42(3): 5-15.
[11]WARD F A. Enhancing climate resilience of irrigated agriculture: A review[J]. Journal of Environmental Management, 2022, 302: 114032.
[12]刘佳, 秦芳. 高标准农田建设对粮食生产韧性的影响研究——基于高标准农田建设试点政策的准自然实验[J]. 农业经济与管理, 2024(6): 77-90.
LIU J, QIN F. Research on impact of high-standard farmland construction on grain production resilience: Based on quasi-natural experiment of high-standard farmland construction pilot policy[J]. Agricultural Economics and Management, 2024(6): 77-90.
[13]周密, 牛浩, 魏超, 等. 农业保险保障对粮食生产韧性的影响研究[J]. 中国农业资源与区划, 2024, 45(8): 44-55.
ZHOU M, NIU H, WEI C, et al. The impact of agricultural insurance guarantee on the food production resilience[J]. Chinese Journal of Agricultural Resources and Regional Planning, 2024, 45(8): 44-55.
[14]SHILOMBOLENI H, EPSTEIN G, MANSINGH A. Building resilience in Africa's smallholder farming systems: Contributions from agricultural development interventions: A scoping review[J]. Ecology and Society, 2024, 29(3): art22.
[15]FANG D, CHEN J Q, WANG S G, et al. Can agricultural mechanization enhance the climate resilience of food production? Evidence from China[J]. Applied Energy, 2024, 373: 123928.
[16]高鸣, 杨新宇. 农业增效益: 数字技术赋能粮食产业高质量发展的实践进路[J]. 学习与探索, 2025(2): 23-32.
GAO M, YANG X Y. Increasing agricultural benefits



## 可视化 0