Socio-economic Inequality and Behaviour Heterogeneity Drive the Flood Exposure Trap

Flood risk disproportionately impacts socially and economically marginalized households, creating feedback loops that reinforce cycles of poverty and limit long-term resilience. Most flood risk management strategies have traditionally focused on understanding the physical drivers of flooding thereby limiting the risk mitigation to structural flood protection measures, which have in-turn resulted in unintended consequences like levee effect. While socio-hydrological assessments of risk and vulnerability indicators exist, most studies assume the exposed populations to be behaviourally homogeneous, thereby failing to explain how flood risk is persisted, redistributed, and entrenched across different sections of the society. The current study addresses this gap by simulating the migration decisions of nearly 100,000 households in an agent-based model, conditioning agent behaviour on socio-economic backgrounds to capture divergent pathways of migration, in-situ adaptation, and long-term risk persistence. The households are classified into four behavioural archetypes grounded in critical socio-economic indicators including social stratification, asset ownership, income source and literacy.

Our analysis reveals the ‘flood exposure trap’ is driven by the intersection of resource constraints and behavioural immobility. The least mobile groups remain critically exposed, experiencing prolonged entrapment in high-hazard zones for over a decade of repeated flood events. These households absorb cumulative losses that further erode their capacity to recover, effectively locking them into a cycle of poverty. In contrast, high-mobility groups successfully reduce their exposure under historical flood conditions by relocating; however they fail to prevent escalated flood exposure under unprecedented, climate change-driven extremes. Thus, proactive migrants eventually face renewed exposure as hazard magnitudes exceed historical precedents. The results indicate that long-term flood resilience is not merely a function of hazard intensity, but is fundamentally governed by social inequality and behavioural heterogeneity. Our work emphasizes the need for equity-sensitive flood risk management strategies that explicitly account for the heterogeneous behavioural constraints of vulnerable populations.