Reference-Dependent Residential Location Choice Model within a Relocation Context

Abstract

This paper presents a reference-dependent model for residential location choice. The key contribution of the model is its incorporation of reference dependence that explicitly recognizes the role of the status quo and captures asymmetric responses toward gains and losses in making location choice decisions. The study uses a retrospective residential search survey and a dwelling supply data set from the Toronto Real Estate Board in Ontario, Canada, to estimate the model at the elemental level of individual dwelling units. The study applies a mixed logit formulation that captures unobserved heterogeneity and avoids imposing independence of irrelevant alternatives restrictions on the choice probabilities. Several types of variables, including dwelling characteristics, land uses and other zonal attributes, accessibility measures, and household socio-demographics, are tested in the model. Although the current dwelling is assumed to be the reference point in framing evaluation of alternative dwellings, all gains and losses are measured by a comparison of current and prospective dwellings in the modeling framework. The results reveal that households prefer gains in the number of bedrooms, but they are more sensitive to the equal amounts of losses. A similar loss aversion attitude is observed for the percentage of open areas and unemployment rate. It is also found that decision makers are sensitive only to the losses for the level of service attributes. The reference-dependent model performs better than a conventional location choice model in terms of model fit and provides important behavioral insights.

The role of location in residential location choice models: a review of literature

“…Habib and Miller (2009) observe a positive impact for more rooms in their reference dependent model. Eliasson (2010) presents an approach similar to a two-step regression to measure the number of rooms per floor space.…”

“…However, due to differences in cities' historic heritages and different structures in residential real estate markets (i.e., differences between public and private housing, tenant protection, and mortgage types), different preferences are to be expected. Vyvere et al (1998) find a general preference for houses in Belgium; Habib and Miller (2009) find a negative preference for attached houses in Toronto. Lee and Waddell (2010a,b) differentiate between household types and find that single-person households and renters prefer multi-family houses; households with children favor single-family buildings.…”

“…Price captures various location characteristics as seen in hedonic regression models (i.e., models, like Löchl and Axhausen 2010), but any models that include property price mention its significant negative impact. While some studies implement price as untransformed value (Andrew and Meen 2006;Kim et al 2005;Vyvere et al 1998;Zolfaghari et al 2012), others include a logarithmic price transformation (de Palma et al 2005(de Palma et al , 2007Habib and Miller 2009;Lee and Waddell 2010a,b). Several studies interact price with household income (Habib and Miller 2009;Zolfaghari et al 2012).…”

“…While some studies implement price as untransformed value (Andrew and Meen 2006;Kim et al 2005;Vyvere et al 1998;Zolfaghari et al 2012), others include a logarithmic price transformation (de Palma et al 2005(de Palma et al , 2007Habib and Miller 2009;Lee and Waddell 2010a,b). Several studies interact price with household income (Habib and Miller 2009;Zolfaghari et al 2012). Zolfaghari et al (2012) see a positive price impact when constraining the choice set based on price, indicating that households choose the best available alternative within a certain range.…”

“…Geo-databases allow combinations of various datasets, as well as adding a spatial reference point. With the increasing availability of such data, ways to represent residential alternatives and their environment have changed; recent models consider buildings or units as choice alternatives and include building-and location-specific attributes (Habib and Miller 2009;Lee and Waddell 2010a).…”