Abstract
Driven by comparative returns, non-grain use of cultivated land (NGUCL) has intensified, posing risks to food security. This study approaches the problem by employing a risk transfer valuation framework, integrating a multi-regional input–output model with a synthetic risk index to establish China’s virtual cultivated land risk transfer network. Complex network analysis was utilized to explore its features while a temporal exponential random graph model was used to identify driving factors of its formation. Results indicate that fewer provinces took on additional pressures and risks. Despite differing motifs, transfer patterns showed little variation. Block analysis showed increasing net recipient relationships (from four to nine) and variable block divisions. Economic development and industrial structure are negatively associated with outgoing transfers, whereas population, production capacity and resource endowment are positively associated with them. The network exhibits time-dependent stability, with few new risk transfer paths forming. This study provides a theoretical basis and new ideas for optimizing land resource efficiency, re-shaping risk transfer patterns and maintaining food security.