生态经济

气候变化已成为影响中国粮食安全和农业可持续发展的重要外部冲击，频发的极端气候事件持续考验粮食生产系统的稳定性、适应性和恢复能力。论文基于2005—2023年中国31个省份的省级面板数据，整合国家气象信息中心、NOAA和《中国气候变化蓝皮书》等多源气象资料，构建气候物理风险指数（CPRI），并采用熵值法测度粮食生产韧性。在双向固定效应框架下引入双重机器学习方法，识别气候风险对粮食生产韧性的影响，并结合全局莫兰指数、局部LISA检验和空间杜宾模型考察其空间关联与溢出效应。研究发现：气候物理风险显著抑制粮食生产韧性，该结论在调整交叉验证折数、缩尾处理、加入控制变量二次项和替换被解释变量等稳健性检验后依然成立；粮食生产韧性和气候物理风险均存在显著空间自相关，空间杜宾模型结果进一步表明，CPRI的直接效应、间接效应和总效应均显著为负，说明气候风险不仅影响本地区的粮食生产韧性，也会对邻近地区的粮食生产韧性产生负向空间溢出；机制分析表明，气候风险会通过抑制农业社会化服务削弱粮食生产韧性，同时也会推动农地流转和土地要素重新配置，农地流转水平提高有助于增强粮食生产系统的适应性和恢复能力，表现出一定补偿性作用；区域异质性结果显示，东部地区气候风险对粮食生产韧性的抑制效应更为明显。研究结论可为完善气候适应型农业政策、强化跨区域风险协同治理和提升粮食生产体系韧性提供经验依据。

Climate change has become a major external shock affecting China’s food security and agricultural sustainability. Frequent extreme weather events continue to challenge the stability， adaptability， and recovery capacity of food production systems. Using provincial panel data for 31 provinces in China from 2005 to 2023 (excluding Hong Kong， Macao， and Taiwan)， this study integrates meteorological data from multiple sources， including the National Meteorological Information Center， NOAA， and the China Climate Change Blue Book， to construct a Climate Physical Risk Index (CPRI). Food production resilience is measured using the entropy-weight method. Within a two-way fixed-effects framework， the study incorporates double machine learning to identify the impact of climate risk on food production resilience. Global Moran’s I， local LISA tests， and a spatial Durbin model are further employed to examine spatial correlations and spillover effects. The results show that climate physical risk significantly inhibits food production resilience. This finding remains robust after adjusting the number of cross-validation folds， winsorizing variables， adding quadratic control terms， and replacing the dependent variable. Both food production resilience and climate physical risk exhibit significant spatial autocorrelation. Results from the spatial Durbin model further indicate that the direct， indirect， and total effects of CPRI are all significantly negative. This suggests that climate risk not only undermines food production resilience within a given region but also generates negative spatial spillovers for neighboring regions. Mechanism analysis shows that climate risk weakens food production resilience by suppressing agricultural socialized services. At the same time， it promotes farmland transfer and the reallocation of land factors. Higher levels of farmland transfer enhance the adaptability and recovery capacity of food production systems， demonstrating a certain compensatory effect. Regional heterogeneity analysis indicates that the inhibiting effect of climate risk on food production resilience is more pronounced in eastern China. This study provides empirical evidence for improving climate-adaptive agricultural policies， strengthening cross-regional collaborative risk governance， and enhancing the resilience of China’s food production system.