Time

2:00 - 3:00pm

Location

1065 Kemper Hall

Speaker(s)

Dr. Xinyu (Jason) Cao, Assistant Professor at Urban and Regional Planning program at Humphrey School of Public Affairs at the University of Minnesota

Abstract

Planners have long advocated land use policies like transit-oriented development, smart growth, and new urbanism in order to reduce the problems caused by auto use. But do such policies induce people to drive less? Researchers have argued that estimates of the built environment’s effects on travel behavior may be incorrect if those estimates do not control for the unobserved process of residential sorting by travel preferences–a process referred to as “residential self-selection.” Scholars have offered several competing arguments about both the likely effects of this process on estimates of built environment effects, and about its implications for predicting the effects of land use policies. Some have argued that ignoring residential self-selection causes overestimates of built environment influences and undermines advocacy for land use policies to change travel patterns. Others have suggested that if residential self-selection is significant and ongoing, land use policies to expand the supply of alternative development could enable more residents who prefer to take transit, walk, or use other alternative modes to move to places that better match their preferences and hence reduce their driving. In this study we distinguish between estimation and prediction, showing how each informs the other. We explain theoretical reasons to expect either over- or under-estimates based on four direct factors and one indirect factor. The direct factors are the base level of modal use for households with different travel preferences, their elasticity of travel demand with respect to built environment characteristics, the representation of such preferences in the population, and the extent to which residents with different travel preferences successfully sort themselves into different development types. The indirect factor, critical in explaining interactions of the direct factors, is the extent to which such development is undersupplied or oversupplied. We explore some scenarios that illustrate how different assumptions under the residential self selection hypothesis will bias estimates of the built environment’s effects on travel, and we discuss the implications of these different factors in predicting how land use policies might affect travel patterns.