ESG climate risk stress testing with flood risk factor in mortgage modeling: an agent-based approach

This project provide a agent-based model that studies the flood impact on the mortgage portfolios.

Repository structure:

agents.py - contains the mesa agent class

model.py - contains the mesa model class

run.py - the function to excute running. A separated .job or .sh file if the simulation is supposed to be run in a visual machine.

utils.py - contains helper functions

config.py - contains the GEV parameters to generate recurrent flood with different intensity (different settings)

data preparation:

The flood risk map can be found: Flood Risk map The initialization of the agents is based on the distribution extracted by KDE method with private data, the features per agent are random sampled from this distribution.