Sustainable metropolitan growth strategies : exploring the role of the built environment

摘要

The sustainability of metropolitan areas has been considered one of the most significant social challenges worldwide. Among the various policy options to achieve sustainable metropolitan growth, smart-growth strategies attract increasing interests due to their financial and political feasibility. Leveraging the interconnection between land use and transportation, smart-growth strategies aim to improve urban life and promote sustainability by altering the built environment with such mechanisms as transit-oriented development, mixed-use planning, urban-growth boundary, etc. My focus in this study is to understand the role that the built environment can play in sustainable metropolitan growth. Unlike previous studies that rely primarily on household survey data in the land use-transportation research, I explore the potential for utilizing spatially detailed administrative data to calibrate urban models and support metropolitan planning. I structure this study in three separate essays. In these essays, with several newly available fine-grained administrative datasets and advanced Database Management System (DBMS) and Geographic Information Systems (GIS) tools, I compute a set of improved indicators to characterize the built environment at disaggregated level and incorporate these indicators into quantitative models to investigate the relationships between the built environment, household vehicle usage and residential property values. I select the Boston Metropolitan Area as the study area. The focus of the first essay is to understand the built-environment effect on household vehicle usage as reflected by the millions of odometer readings from annual vehicle safety inspections for all private passenger vehicles registered in the Boston Metropolitan Area. By combining the safety inspection data with fine-grained GIS data layers of common destinations, land use, accessibility, and demographic characteristics, I develop an extensive and spatially detailed analysis of the relationship between annual vehicle miles traveled (VMT) and built-environment characteristics. The empirical results suggest that there are significant associations between built-environment factors and household vehicle usage. In particular, distance to non-work destinations, connectivity, accessibility to transit and jobs play significant roles in explaining the VMT variations. The research findings can help analysts understand the environmental implications of alternative regional development scenarios, and facilitate the dialogue among regional planning agencies, local government and the public regarding sustainable regional development strategies. In the second essay, I investigate the built-environment effect on residential property values with a cross-sectional analysis. The major dataset is the single-family housing transaction records from city and town assessors in the Boston Metropolitan Area assembled by the Warren Group. I use factor analysis to extract several built-environment factors from a large number of built-environment variables, and integrate the factors into hedonic-price models. Spatial econometric techniques are applied to address the spatial autocorrelation. The empirical results suggest that the transaction price of single-family properties is positively associated with accessibility to transit and jobs, connectivity, and walkability, and negatively related to auto dominance. The built-environment effects depend on neighborhood characteristics. In particular, households living in neighborhoods with better transit accessibility tend to pay a higher premium for smart-growth type built-environment features. The research findings suggest that most smart-growth strategies are positively associated with residential property values. Although built-environment characteristics advocated by smart-growth analysts do not have universal appeal to households, they no doubt satisfy an important market segment. In the third essay, I examine the role that selectivity and spatial autocorrelation could play in valuing the built environment. Using transaction and stock data for single-family properties in the City of Boston from 1998 to 2007, I integrate a Heckman-selection model and spatial econometric techniques to account for sample selection and spatial autocorrelation, and estimate the willingness-to-pay for built-environment attributes. The empirical results suggest that the built environment can influence both the probability of sale and transaction price of properties. Failing to correct for sample selection and spatial autocorrelation leads to significant bias in valuing the built-environment. The bias might misguide policy recommendations for intervening urban development patterns and distort estimations of the value-added effect of infrastructure investment for land-value-capture programs.