The Influence of Neighborhood Design on the Sustainability of US Suburbs

TL;DR

The paper investigates whether the sustainability challenges of US suburbs stem primarily from remoteness or from the specific Garden City Design (GCD) of neighborhoods. It introduces a nationwide composite GCD index, derived from four street-network and block-configuration features, and links historical adoption to contemporary social and environmental outcomes using OLS, propensity score matching, and an IV strategy based on national design waves. Empirical results show that higher GCD exposure is associated with greater social isolation, more sedentary time, and higher per-capita greenhouse gas emissions, with IV estimates indicating sizable causal effects. The analysis suggests that GCD accounts for roughly 27–38% of the adverse costs of suburbanization, highlighting that urban form, not just remoteness, substantially shapes sustainability; policy implications point to retrofitting and updated development standards as complementary to core-decentralization efforts.

Abstract

The growth of suburbs in the US has led to significant sustainability challenges; yet, it remains unclear whether these challenges stem from the remoteness of suburbs from city centers or the specific designs used to develop them. This paper examines how Garden City Design (GCD) -- one of the most influential suburban design paradigms since the early 20th century -- impacts the social and environmental outcomes of neighborhoods. I first introduce a composite measure of GCD, derived from street layouts and block configurations, to quantify its nationwide adoption. I use this measure combined with mobility and emissions data to estimate the impact of GCD on neighborhood outcomes using complementary identification strategies, including ordinary least squares (OLS), matching estimators, and an instrumental variables (IV) approach that exploits historical variation in GCD adoption. Results show that GCD leads to worse sustainability outcomes, including increased greenhouse gas emissions, greater social isolation, and higher sedentary behavior. The prevalence of GCD accounts for 27-38% of the adverse effects associated with suburbanization, underscoring the crucial role that neighborhood design plays in shaping urban sustainability.

Figures (10)

• Figure 1: Temporal and Spatial Dynamics of GCD. Panel A illustrates GCD trends in 5-year intervals from 1800 to 2015. Panel B plots the relationship between GCD and urban sprawl, measured as distance to the primary city within each MSA, for neighborhoods established during four distinct periods: 1900–1925, 1926–1950, 1951–1975, and 1976–2000. For visual clarity, the graph focuses on neighborhoods within 45 km of city centers, capturing 90% of the sample. The error bands represent 95% confidence intervals.
• Figure 2: Variation of GCD in Urban Areas. Panel A illustrates the average GCD index across urban areas (N=22,494). Panel B shows the GCD index for neighborhoods in five selected urban areas: Philadelphia, Boston, Sacramento, Phoenix, and Salt Lake City.
• Figure 3: Benchmarking the Effects of GCD versus Remoteness on Neighborhood Outcomes. Panel A presents results for social isolation, Panel B for daily wake-time spent at home, and Panel C for annual greenhouse gas emissions per capita. The figures plot predicted outcomes for high-GCD, low-GCD, and the average neighborhood, computed for 10-km rolling windows of distance to the main city center. These are computed from regressions in each bin using the specification from Column 5 of Table \ref{['table: all discrete']}. Predicted outcomes for high-GCD neighborhoods are calculated as $\text{Predicted outcome}=\text{Mean outcome in window}+\hat{\beta}\; (1-\text{Mean garden design in window}),$ and for low GCD neighborhoods as $\text{Predicted outcome}=\text{Mean outcome in window}-\hat{\beta}\; \text{Mean garden design in window},$ where $\hat{\beta}$ denotes the estimate of Equation \ref{['eq:ols']} in that distance window.
• Figure 4: Map of Urban Areas. The map shows the location of the urban areas defined using the 250x250 raster data by leyk_2018. 22,494 urban areas and 60,421 neighborhoods within these areas were identified.
• Figure 5: Geographical Distribution of Neighborhoods for GCD Validation. Location of iconic Garden City neighborhoods using data from wheeler_evolution_2008, talen_2022, and SalazarMiranda2021.