The Landscape of Innovation
Economies happen in places, and they have an effect on the physical form of places – as the new ghost ‘burbs of the real-estate driven economies of Florida and Nevada show us. If the next American economy will be export-oriented, lower-carbon, and innovation-driven, as Larry Summers has posited (my colleagues and I agree), what kind of landscape will result? What will be the successor to the exit-ramp office parks, tangled residential cul-de-sacs, relentless highways, and reliably repetitive strip malls that were the built environment for the consumption-fueled, cheap-energy economy we’re struggling to come out of?
The short answer: denser and better connected. An export-oriented economy will need a robust port infrastructure, which the U.S. more or less has, but also a freight network, guided by an intelligent, coordinated freight strategy, which the U.S. sorely lacks, so American manufactured goods can make it to the ports as fast as possible. (Howard Rosen has lucidly explained why goods exports are so critical to economic growth, particularly after the recession – the upshot: we still need to make stuff to sell abroad).
A lower-carbon economy will also increase the value of denser places, where transit makes sense, and where a detached single-family home isn’t the only residential option.
But what’s jaw-dropping to me is how much innovation also relies on density – measured in blocks, not miles. Ed Glaeser and others, going as far back as Alfred Marshall in 1890, have noted the gains from agglomeration, or geographically clustered activities. But they’ve generally looked at agglomeration at the metropolitan scale, and metros tend to be quite far-flung places, stretching across several counties. A couple of papers have described how dramatically the benefits of innovation and agglomeration fall off once people spread out beyond walking distance.
Stuart Rosenthal and William Strange found that the intellectual spillovers that drive innovation and employment drop off dramatically as firms and people move more than a mile apart. As they note, “Information spillovers that require frequent contact between workers may dissipate over a short distance as walking to a meeting place becomes difficult or as random encounters become rare.” These effects are staggering. As you move out beyond just one mile, the power of intellectual ferment to create another new firm or even another new job drops to one-tenth or less of what it is closer in, or as the economists say in their inimitable prose:
Agglomeration economies attenuate with distance. The initial attenuation is rapid, with the effect of own-industry employment in the first mile up to 10 to 1000 times larger than the effect two to five miles away….
To gain a sense of the magnitude of these estimates, consider first the software industry for which the localization effects are among the most pronounced. For that industry, adding one hundred software workers to the 1-mile ring would generate 0.04 births and 1.117 additional employees at new establishments. Adding one hundred additional employees to the 2-5 mile ring would result in 0.005 births and 0.08 employees, while adding one hundred additional employees to the 6-10 mile ring would lead to 0.004 births and 0.08 workers.