The Brookings Institution's Metropolitan Policy Program just released a report on the continuing decentralization of employment in U.S. metropolitan areas. Unfortunately, while a useful empirical note on the changing geography of U.S. urban areas, it provides little relevant for policy purposes.
"Job Sprawl Revisited" laments the fact only 21 percent of employees in the top 98 metropotlian areas work within three miles of downtown, while 45 percent work more than 10 miles away (in 2006). Moreover, 95 of the 98 metropolitan areas found the share of employment within three miles of downtown fell from 1998 to 2006, so this trend is broad-based.
Why is this a problem? According to the Brookings report, decentralization increases infrastructure costs, lengthens commutes, creates a greater mismatch between jobs and low income workers, reduces innovation, and increases energy consumption. Unfortunately, these are presumptions, and the report does not explicitly tie decentralization of jobs to these effects.
More importantly, is this true? No, at least not based on the evidence in the report. The substance of the report simply measures where jobs are located based on its distance from a the center of a census-defined central "place". It then assumes that further distance from the CBD means urban productivity is lower (because growth is in lower density locations). It does not evaluate productivity or income growth. In fact, even though it would have been easy to do, the report does not even present eveidence on the relationship between job growth and decentralization.
I suspect the results wouldn't be quite as negative if they had. For example, the top five most centralized metropolitan areas are: Virginia Beach, New York, Salt Lake, Las Vegas, and Boston. Salt Lake and Las Vegas are fast growing metro areas, but they are also small. New York, while large, is growing very slowly.
The top five most decentralized metros are: Detroit, Chicago, Dallas-Fort Worth, Los Angeles, and Philadelphia. Philadelphia and Detroit are not growing like gangbusters, but almost all the job growth in Detroit is west of the traditional CBD and around Ann Arbor (home to the University of Michigan). LA and Dallas, of course, are big regions, big job generators and centers of rising productivity in the national economy. Ironically, Chicago is considered a decentralized region, but its traditional downtown, the "loop," is among the nation's largest (with more than 16,000 residents) and has been growing.
Two other flaws stand out.
First, the report presumes that distance is important to regional growth. In fact, as we point out in our recent book Mobility First: A New Vision for Transportation in a Globally Competitive 21st Century, travel speed is more important. It's access to labor and resources, not linear distance, that impacts productivity.
Second, the idea that a CBD is the center of income and regional job growth is a quaint 20th century notion of the urbanized area, but ignores all the relevant changes to urban spatial structure of the past one hundred years. The idea that a single (or two) employment center(s) is "optimal" ignores the actual economics of urban areas where multiple employment centers have emerged in places such as Phoenix, Houston, Los Angeles, and San Francisco. This decentralization has been a central source of productivity growth.
If low-density suburban areas are so unproductive, why are those areas the centers for regional job growth? Alas, the Brookings report doesn't ask this question, let alone answer it.