The Impact of Stock Land Redevelopment Level on Urban Resilience and its Conditional Patterns: Evidence from 289 Chinese Cities

As urban development shifts toward the redevelopment of existing built-up space, understanding how stock land redevelopment relates to urban resilience is increasingly important. Yet whether this relationship varies across redevelopment states and urban contexts remains insufficiently understood. Using land transaction records and panel data for 289 Chinese cities from 2014 to 2023, this study combines fixed-effects estimation with XGBoost-SHAP analysis to examine the average association and conditional predictive patterns. Higher redevelopment levels are positively associated with overall urban resilience in the baseline estimates. Under the baseline two-year specification, the dimension-specific association is strongest for economic resilience, although the relative strength of the economic and ecological associations varies across alternative lag structures. Among the redevelopment features, scale intensity and cumulative extent receive the largest SHAP and permutation-importance attributions. More importantly, the predictive contribution of scale intensity is greater at medium cumulative extent than at low cumulative extent but declines at high cumulative extent. A similar non-monotonic pattern emerges across project-density states, while the magnitude of these conditional differences varies across regions and urban-agglomeration tiers. These findings characterize stock land redevelopment as a state-dependent process and highlight the relevance of redevelopment conditions and urban context to differentiated land governance.