In June 2022, BIDS’ Tim Thomas (Research Training Lead for Berkeley Computational Social Science Program) and his team at Berkeley’s Urban Displacement Project (UDP) released a first-of-its-kind machine learning model that predicts the risk of low-income renter displacement at the neighborhood level. With over a year of development, the Estimated Displacement Risk (EDR) Model uses a decade’s worth of household level data, controlling for over 600 census, market, housing, and demographic variables to predict which neighborhoods are at the greatest risk of displacement.
For its initial launch, the UDP applied the EDR to the state of California, which is helping local and state agencies identify housing vulnerability and understand where displacement risk is highest. Planning is under way to continue expanding the project to conduct analyses across the rest of the country by early 2023, and the model will continue to be improved as new data sources are gathered and integrated. Throughout, Thomas plans to link unique eviction and employment data to the EDR in order to improve UDP’s Housing Precarity Risk Model, which helps local, state, and federal agencies understand drivers of vulnerability and recommend policies to deter displacement.