Simulating artificial cities in a GIS environment: urban growth under alternative regulation regimes

This paper reports on an attempt to combine neo-classical urban economic theory with complex systems methods. The innovative feature of our model from the point of view of conventional economic theory lies in its explicit treatment of spatial relationships and time sequence. From the perspective of raster or cellular GIS models of urban processes, the work is innovative in that it replaces the more usual heuristic cell-transition rules with micro-economic theory. The mix of modelling paradigms is not unproblematic, however, and we discuss the challenges encountered at this research frontier. These notwithstanding, our hybrid model has the potential to be used as a GIS-based laboratory for exploring micro-economic propositions, particularly those relating to urban processes that are path dependent. The version of the model reported simulates spatially equilibriated path dependent futures of a city governed by local development decisions that are at partial equilibria in the neo-classical sense. Two simulations are described which permit visual and economic exploration of (a) an explicitly spatial version of the economic theory of externalities and (b) a new theory of densification. The dual paradigm (Cellular Automata-neo-classical economics) leads to an interesting class of simulations in terms of stability. Economically our simulated cities become increasingly efficient, in terms of private and social product. The long-run economic equilibrium is achieved by many individually efficient negotiations based only on local information. There is no parallel long-run spatial equilibrium however. The spatial configuration of land uses is constantly shifting as a result of randomness in the land use bidding process. The spatial instability is, however, limited by the self-organised drive for greater overall economic efficiency. In economic terms, the model's spatial instability represents random re-allocation of land-use within a set of Pareto-efficient spatial configurations-an intriguing result that we intend to follow up in future work.