A global take on congestion in urban areas

TL;DR

This paper analyzes global urban congestion data, revealing that peak-hour congestion scales with the square root of population density, highlighting the need for better theoretical models and a focus on activity distribution rather than density alone.

Contribution

It provides a novel empirical analysis linking congestion growth to population density using scaling arguments and data fitting across numerous cities.

Findings

Congestion during peak hours scales as the square root of population density.

Density alone may be misleading in understanding urban congestion.

A better theoretical understanding of congestion phenomena is urgently needed.

Abstract

We analyze the congestion data collected by a GPS device company (TomTom) for almost 300 urban areas in the world. Using simple scaling arguments and data fitting we show that congestion during peak hours in large cities grows essentially as the square root of the population density. This result, at odds with previous publications showing that gasoline consumption decreases with density, confirms that density is indeed an important determinant of congestion, but also that we need urgently a better theoretical understanding of this phenomena. This incomplete view at the urban level leads thus to the idea that thinking about density by itself could be very misleading in congestion studies, and that it is probably more useful to focus on the spatial redistribution of activities and residences.