Network Structure and City Size

Recently published: Levinson, David (2011) Network Structure and City Size. PLoS One PLoS ONE 7(1): e29721, January 12, 2012 [doi]

Network structure varies across cities. This variation may yield important knowledge about how the internal structure of the city affects its performance. This paper systematically compares a set of surface transportation network structure variables (connectivity, hierarchy, circuity, treeness, entropy, accessibility) across the 50 largest metropolitan areas in the United States. A set of scaling parameters are discovered to show how network size and structure vary with city size. These results suggest that larger cities are physically more inter-connected. Hypotheses are presented as to why this might obtain. This paper then consistently measures and ranks access to jobs across 50 US metropolitan areas. It uses that accessibility measure, along with network structure variables and city size to help explain journey-to-work time and auto mode share in those cities. A 1 percent increase in accessibility reduces average metropolitan commute times by about 90 seconds each way. A 1 percent increase in network connectivity reduces commute time by 0.1 percent. A 1 percent increase in accessibility results in a 0.0575 percent drop in auto mode share, while a 1 percent increase in treeness reduces auto mode share by 0.061 percent. Use of accessibility and network structure measures is important for planning and evaluating the performance of network investments and land use changes. Keywords: Connectivity, Network Structure, Transportation Geography, Network Science, City Size, Scaling Rules, Accessibility, Travel Behavior, Mode Share, Journey-to-Work

This paper has several features:

  1. The paper includes a ranking of 50 US cities by estimated accessibility (Table 3). This estimate is macroscopic, though I think quite plausible, and shows the variation in the 10 minute vs. 20 minute … vs. 60 minute and composite accessibilities. The composite numbers are more or less what you expect, but some small cities are quite fast, so have high 10 or 20 minute accessibilities by car. Lots of work remains to be done on this (both multiple modes and multiple points in time) but this should be a valuable metric.
  2. Larger cities are better connected. They are also more productive. This research suggests a hypothesis (which further research will need to test) that variations in network structure may explain variations of economic output. More connected cities are more efficient. It is not simply how many people are in the city (the classic economy of agglomeration argument) but how they are connected that affects their productivity.

I will also comment about the publication itself. It was published in PLoS One, a first for me. PLoS ONE is a newish, open content journal across part of the Public Library of Science family that aims to represent all fields of study. I did this as an experiment as much as anything. The paper is out less than 4 months after submission, and 2 months after revision. This is *fast*, much faster than for-profit publishers offer. The journal is interdisciplinary, and does not winnow for importance (letting the field do that), instead winnowing for quality of the work and its description. Everyone in the field knows how arbitrary publication is when paper is a constraint. This seems an improvement.