Abstract
In the context of a land use – transport simulation for the Greater Zurich area that is currently being developed at ETH Zurich, multinomial logit models of residential location choice have been estimated. The estimations were based on two datasets: a household survey conducted in 2005 and real estate offers collected from the Internet in 2004 and 2005. Drawing on these data sources, a number of different variable combinations were tested in order to obtain models with satisfactory explanatory power. In addition, estimation results for specific household types were compared. A core set of variables was used as a starting point to explore the significance of various characteristics of household, dwelling and location. This core set was then tentatively extended by adding further variables. Variables that proved significant comprised mostly location-related attributes like various densities or travel time to city centre but also characteristics of municipalities like the rate of vacant rental units or the tax index and housing unit’s features such as price and size. Interaction terms with sociodemographic attributes of the decisionmaking households were introduced to improve the explanatory power of the models. More detailed results were obtained by estimating separate models for different household types. Those types were formed regarding sociodemographic and socioeconomic features. For choice set selection, random sampling was applied. This approach might later be extended to include stratified sampling strategies making use of similarities between chosen and nonchosen alternatives. For carrying out the sampling procedures a custom made Java programme was developed. The models resulting from the estimation process are presented and evaluated in the paper at hand.
Paper version not known (
Free)
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call