Marco Pangallo from Sant’Anna School of Advanced Studies, Pisa will present his (life only) talk on Friday, July 1, 2022 at 3 PM in the Salon.
If you would like to join the talk, please send us an email.
Title: “Data-driven economic agent-based models”
Agent-based models (ABMs) are deterministic-stochastic maps that iterate the state of a system forward in time. In standard practice, ABM microstates -representing heterogeneous, interacting units- are initialized randomly and evolved following internal dynamics that are not anchored to real-world time series. In recent years, however, more and more researchers have started to initialize their ABM microstates with real-world data and attempted to reproduce real-world dynamics.
In this talk, I will give an overview of my research on data-driven economic ABMs. I will discuss theoretical problems such as latent variable estimation and consistency between micro and macro economic statistics, and show two applications. In the first application, I will address some of the most debated issues related to epidemic-economic tradeoffs by introducing a detailed ABM that simulates infections and unemployment at the level of 500,000 synthetic individuals in the New York area. This ABM is initialized from census data, regional and national accounts and input-output tables, and cell phone mobility data.
In the second application, I will discuss the effect of beliefs about sea level rise on the housing market, by building an ABM of the housing market of Miami that uses detailed property, transaction, mortgage and demographic data. Overall, these results suggest that ABMs are the ideal tool to bridge big data and theoretical modeling, although a lot of research is still needed to develop standard techniques and practices.