Abstract
Using a newly assembled global database on motor vehicle travel speed from nearly 1 billion simulated trips in 1200 large cities containing more than 90% of the world’s urban population outside China, we calculate comparable city-level indices of speed and congestion and document three novel facts. First, urban travelers in rich countries experience speeds two to three times faster than travelers in poor countries. Second, variation in travel speed across cities, both between and within countries, is driven primarily by uncongested speed, not congestion. Third, a novel decomposition, which unifies several prominent decomposition methods, shows that several infrastructure features, including more grid-like networks, more dual carriage and major roads, and fewer intersections can explain a large share of both the GDP-speed correlation across countries and within-country variation. These results suggest that enhanced urban mobility is an important and previously undocumented feature of economic development, and one that is limited more by poor infrastructure than by congestion. Since slow travel is costly regardless of its cause, existing indices emphasizing congestion are missing most variation in speed