A perspective on urban economics and geographical agent-based models. (Keynote Speech)

In this presentation, we will argue that there are many benefits to further liaise urban economic models and geographically -aware agent-based simulation models for further understanding the structure and dynamics of cities, especially at the scale of city regions but also, eventually, for contributing micro understanding of changes in a system of cities. At first sight it appears that these two fields of urban research are very separated: urban economics relies mostly on mathematical proofs, a parsimonious set of interactions, static equilibria and a rather homogeneous space and set of agents; while geographical agent-based models emphasize the dynamics and complexity of cities, the heterogeneity of agents and space, and rely solely on computational methods. The presentation will showcase a series of theoretical and applied research to stress the idea that the two can fruitfully be combined and that results are then palatable to both urban economists and geographers and can therefore participate to consolidating a wider and formalised urban theory. The geographical turn towards agent-based approaches, compared to more aggregate models and behaviourally-blind geocomputational methods, is actually a chance for geographers to formalise market as well as non-market interactions in a manner that is consistent with urban economic theory and enhance the treatment of space in this theory. Relying on a formal description of agents utility permits to discuss normative instruments where environmental and welfare impacts can be traded-off by policy makers. Furthermore,, land or housing market outcomes of the models and the analytical efforts actually help to reduce parametric space and calibrate or validate models against independent data.