How might we assess whether a climate adaptation policy will benefit a population — and at what economic cost — before it is implemented? This talk presents a bottom-up, data-intensive approach based on agent-based simulation: computational models in which hundreds of thousands of virtual households and firms make individual decisions, so that aggregate economic and social outcomes emerge from the interactions of many heterogeneous actors rather than imposed assumptions.

The central idea is calibration against real-world micro-data. By grounding these “artificial societies” in household surveys, flood maps, and geolocated climate variables such as temperature and precipitation, we can construct synthetic populations that closely reproduce the structure of specific territories. Within this controlled virtual environment we can then run counterfactual, “what-if” experiments — introducing a candidate climate policy and tracing its effects on the simulated economy and population over time.

Building these artificial societies draws on several fields: economics and psychology shape how agents behave, geography and climate science define the physical environment they inhabit, and data science and high-performance computing make the simulation possible — offering policymakers a low-risk laboratory for evaluating climate and development choices.

20 October 2026, 12:00 CEST
To join the webinar, register here