Resumen
Urban and regional models have been developed for different policy fields at different levels of spatial and temporal resolution. But it has become apparent that policies interact across space and time and need to be modelled together. The first urban and regional models were aggregate in space and comparative-static in time. More recently, new data sources and computing techniques have stimulated ever more disaggregation in space and time culminating in agent-based, activity-based microsimulation despite its significant even larger data needs, computing requirements and theoretical problems. This paper argues for models that are instead multi-level and multi-scale in space, time and subsystems. This paper starts with a brief history of urban models and the experience of the authors with the highly integrated urban microsimulation model ILUMASS. Based on this experience, it discusses the benefits and pitfalls of microsimulation and proposes a three-level model system of spatial development, ranging from the European to the local level. The paper closes with new challenges for urban models posed by climate change, energy scarcity, new social problems and new technologies and argues that they make multi-level, multi-scale models even more important and illustrates this by ongoing work with the multi-level model for cities in the Ruhr.