Effects of Neighborhood Choice on Housing Markets
The main goal of the overall project is to develop a new method to measure and forecast the effects on urban development, housing markets and quality of life combining a new microsimulation tool (Interaction Spaces, a new type of mathematical model for complex systems developed by the Modeling and Applications of Complex Systems Laboratory (Macs-Lab), University of Lugano) and integrating it with an RP-SP based residential choice model in a recursive way. The secondary objective consists in modeling neighborhood segregation processes and studying sense of community in small urban areas. In order to develop the new model, an interdisciplinary approach between mathematics and economics and psychology seems appropriate, because simulation models risk to be too idealized in terms of validation on existing reality, and economic stated choice models, even if embedded in a psychological frame, might remain too far from realistic calibration and are not able to produce future forecasts. The combination of both approaches in a new and recursive model should attribute more relevance to both, introducing revealed as well as hypothetical behavioral parameters into simulations on the one hand, and enhancing stated preference designs integrating variable values extracted from simulations on the other. Concerning the secondary objective, it has to be noted that neighborhood does not refer only, as usual in European applications to environmental quality, to infrastructure and public services, but to the psychological sense of community in the neighborhood and characteristics of the neighbors in a literal sense. This focus, more popular in US applications, permits to study the impacts of immigration, increasingly relevant also in the reality of relatively small Swiss urban conglomerations, on the dynamics of spatial segregation processes. Though the scope is not to explain the complexities of social segregation as such, but its spatial and economical impacts, it is essential to complement the study with a psychological module. This will permit to evaluate and enhance the relevance of the economic rationale behind the choice experiment. The first step will be the development of a model for housing market in the framework of IS, a new type of mathematical model for complex systems developed by the Macs-Lab at the University of Lugano. This type of model would enable to join the approach of microsimulations with more classical analytical approach typical of Econophysics, obtaining a very detailed model for housing markets with the possibility to study its dynamics thanks to its sound mathematical properties. Using fuzzy modeling methods it is possible, in the framework of IS, to realistically describe agents’ decision processes of a multi-agents like system (MAS), including memory effects, long range spatial interactions, neighborhood effects on prices, etc. Moreover extensive state variables, typically used in IS, always verify suitable differential equations for the time dynamics, realizing in this way the above mentioned connection between MAS and analytical methods, and enabling to study a MA like system as a continuous dynamical system. In this research project, we propose a stepwise approach to calibration and validation of our microsimulation model via RP and SP surveys. First we use the reliable hedonic prices and multinomial choice models constructed from the RP survey in order to obtain an input “learning set” for the calibration and validation of the prices sector of the dynamical model. In this way the output prices of the microsimulation model can reach a level of validity comparable to the one of the hedonic method, which are well-known strongly reliable at the time of data collection. Other validation methods (comparison between simulated and empirical frequencies of events and validation by means of experts’ opinion, see the enclosed Vancheri et al. 2006 II) will be used to calibrate and validate all the parameters which are not included in the “prices sector” of the dynamical model. At this stage, future scenarios can be produced using the microsimulation model. A meaningful set of local and global future urban configurations can be selected from these scenarios and used in the subsequent SP survey. It is important to point out that, unlike usual design of SP survey, we do not use here hypothetical and simple variations of given situations. Indeed, fuzzy logic indicators used in the model can be employed to produce strongly contextualized and meaningful descriptions of local and global urban configurations using natural language (virtual reality tools are also planned as a support to visualize spatial and land use related aspects). For example it is possible to propose a situation which is not just the present one plus an added feature (e.g. a new shopping centre), but also all the probable consequences of this added feature. Moreover, these scenarios belong to a probable future of the system and hence extend in a relevant way the validity of the SP survey beyond the current time. It is possible to focus on critical configurations far away from the present one, e.g. near to bifurcation points (in our approach it is possible to investigate these aspects using random and ordinary differential equations, see the enclosed Vancheri et al. 2006, I). In the last stage, the results of the SP will be used to calibrate and validate the dynamical model at a higher level of reliability. More precisely the SP extends the initial learning set to future critical situations and hence increases the validity of the dynamical model. This use of the SP results is made possible by the use of fuzzy logic based methods which model agents’ decisional behavior in terms of goals and constraints, and this makes possible the translation of SP results into the dynamical model.