Analysis of the Drivers of Farmers’ Willingness To Exit Rural Homesteads from a Sustainable Livelihood Perspective
摘要
Based on the perspective of sustainable livelihoods, this study constructs a two-dimensional analysis framework of ‘livelihood capital-geospatial resource endowment’, combines this framework with data from the 2020–2022 land economic survey of Jiangsu Province, China, and adopts the combined machine learning approach of Random Forest and SHAP to systematically explore the decision-making mechanism underlying farmers’ paid withdrawal from rural homesteads. The study reveals that: (1) Geospatial resource endowment exerts a significantly stronger influence than livelihood capital on explaining farmers’ willingness to withdraw; (2) The drivers exhibit an effect characterized by ‘low promotion, high inhibition’; (3)There is heterogeneity in interaction strength across different indicator categories,, and complex synergistic relationships exist among the drivers.This study proposes a spatially differentiated governance approach for rural homesteads based on regional resource endowments. Implementing property-based utilization strategies can enhance resource efficiency while preserving farmers’ ownership rights. Moving beyond the conventional single-capital paradigm, this study offers theoretical and practical insights for rural homestead reform in developing countries seeking context-specific governance pathways.