How will smart growth land-use policies affect travel? A theoretical discussion on the importance of residential sorting

Do policies to encourage compact, mixed use, pedestrian-friendly land-use patterns reduce driving? Not necessarily. Understanding how the built environment affects travel patterns is complex, not least because households may choose their neighborhoods on the basis of how they expect to get around. Some scholars have argued that ignoring this process of residential sorting, or self-selection', causes overestimates of built-environment influences and leads to false optimism about the efficacy of land-use policies in influencing travel. But others have suggested that residential self-selection provides a strong argument for using land-use policies to expand the supply of development that may facilitate lower automobile use. We argue that previous work on both sides of the argument has neglected to think through the myriad ways that residential choice could affect estimates of built-environment effects. In this paper we provide a more rigorous theory of residential self-selection, identifying a set of five household, market, and policy factors that are critical to understanding the residential self-selection problem, along with research questions that correspond to these factors. We explain why observed relationships between travel and the built environment could be misleading, causing either overestimates or underestimates, depending on the nature and context of residential choice. We illustrate with scenarios that show how different plausible assumptions about residential choice will bias, in different directions, estimates of the built environment's effects on travel; and we argue the need for research to focus not just on those independent estimates but, critically, upon the market and policy context that influences residential sorting.