Ever since the economic crisis of 2008, and perhaps even before, a growing chorus of economists have argued that the pace of American innovation has slowed.
Tyler Cowen of George Mason University contends that America and other advanced nations are entering a prolonged "great stagnation," in which obvious technological ideas have been largely exhausted and the rate of innovation and economic growth have slowed. In a much-cited paper, "Is US Economic Growth Over?" the economist Robert J. Gordon of Northwestern University concurs, providing empirical evidence that recent epochs of technological advance pale in comparison the great boom of the Industrial Revolution a century ago.
Others counter that a new period of innovation is emerging. The late economist Christopher Freeman, invoking Joseph Schumpeter, long ago argued that economic crises set the stage for great bursts of innovation and technological advance.
The graph below by Deborah Strumsky, a leading expert on patents and innovations at the University of North Carolina Charlotte, takes an empirical look, charting the pattern of innovation measured by patent applications from 2001 to 2011. Strumsky used patent applications, as not enough time has elapsed to document accepted patents (the two are highly correlated). Despite their limitations, economists use patents as the standard measure of innovation.
The pace of innovation was clearly slowed by the onset of the crisis. Patent applications were literally cut in half from a pre-crisis peak of 155,802 in 2007 to a low of 75,060 in 2010. The good news is patent applications have bumped up since then, reaching 106,711 in 2011.
But the geography of innovation is substantially uneven, clustered and spiky; It concentrates much more in some places than others. As background for my recent feature for The Atlantic, I asked Strumsky to chart the changing geography of innovation – measured by patent applications – across America’s more than 350 metros. She divvied up patenting activity into three time periods: pre-crisis (2001-2007), crisis (2008-2010) and post-crisis (2011). My Martin Prosperity Institute colleague Zara Matheson mapped the data.
The first map shows the absolute number of patent applications by metro area for the post-crisis year, 2011. The San Jose-Sunnyvale-Santa Clara, California, metro area is far and away the country’s largest center for innovation. In 2011, this Silicon Valley region produced 8,551 applications, beating out next-best San Francisco by over 2,000 (the San Francisco-Oakland metro accounted for 6,260 applications). At its peak in 2007, the San Jose region produced more than 10,000 patent applications, accounting for 7 percent of the nationwide total. Other leading patenting centers include: New York (6,181 applications), Los Angeles (4,766), Seattle (4,302), and Boston-Cambridge (4,089). While these are all large metros, the pattern does not follow population size precisely. In 2011, the two leading metros in terms of patenting, San Jose and San Francisco, ranked 11th and 34th in terms of population, ahead of the nation’s two largest metros New York and Los Angeles, which ranked third and fourth on patenting. The nation’s third and fourth largest metros, Chicago and Dallas, ranked 8th and 12th on patenting.
The chart above shows the trend for the five leading patenting metros – San Jose, New York, San Francisco, Los Angeles and Boston – from 2001 to 2011. Together, these five metros accounted for 30.2 percent of patent applications over the course of this decade. All five mimic the national trend with a substantial dip from 2008 to 2010 and then an uptick in 2011. But, the graph also shows San Jose increasing its lead over the others. San Francisco also experienced a more dramatic recovery than New York, L.A. or Boston-Cambridge, as shown by the steeper final segment of its line between 2010 and 2011.
To better compare the geography of innovation in the pre- and post-crisis periods, Strumksy used a metric called a Location Quotient, or LQ. An LQ is conventional urban metric that charts how the level of activity in a city or metro compares to its expected share of the national economy. An LQ of greater than one indicates that the metro produced proportionally above the national average, based on the size of the local economy, while an LQ of less than one shows the opposite. She calculated LQ for the pre-crisis period 2001 to 2007, the crisis period 2008 to 2010, and the post-crisis period of 2011.
The above map shows the LQ for patent applications across all U.S. metros for the pre-crisis period. San Jose takes the first spot by a wide margin. Corvallis, Oregon – which includes the campus of Oregon State University – and Rochester, Minnesota – home to both the Mayo Clinic and a large IBM campus – come in the next two slots. Two other large college towns – Boulder, Colorado, and Burlington, Vermont – come next. Rounding out the top ten are Austin, Texas; Fort Collins, Colorado; Ann Arbor, Michigan; Rochester, New York; and San Francisco, California.
This map charts the same metric for the crisis period. Again San Jose takes the top spot. Rochester, Minnesota, moves up from third to second. Boulder moves up from fourth to third. Corvallis falls from second to fourth. Ann Arbor moves up from 8th to 5th. Burlington, San Francisco, Santa Cruz, Austin, and San Diego round out ten top ten. Rochester, New York, just misses the top ten during this period, and Fort Collins, Colorado fell significantly, from 7th to 19th. Santa Cruz and San Diego, both in California, climbed into the top ten.
The last map charts the pattern for the post-crisis period. Again San Jose takes the top spot by a large margin. Boulder overtakes Rochester, Minnesota, for second. The remainder of the top ten includes the smaller college towns of Ann Arbor, Corvallis, Burlington, Ithaca and well-known tech hubs like San Francisco, Seattle, Austin and the North Carolina Research Triangle. The only change in the top ten is the reappearance of Rochester, New York, displacing Austin.
Strumsky then compared the change in patenting before and after the crisis. To do so, she simply calculated the change in the LQ over the course of these two periods. The San Jose metro saw the biggest gain, as its LQ increased from 10.3 to 12.5. Several other leading knowledge centers and tech hubs also saw considerable gains, including Santa Cruz, Seattle, Ann Arbor, San Francisco, and San Diego. Among larger metros, Louisville, Detroit, Albany, Charlotte and Houston saw considerable gains. A number of smaller metros that are not known as tech centers also saw substantial gains, such as Greenville-Anderson-Mauldin, South Carolina; Dubuque, Iowa; Peoria, Illinois; Niles-Benton Harbor, Michigan; Elmira, New York; and Rockford, Illinois.
On the flip side, the metros that saw the biggest declines in patenting activity (measured by the change in their LQs) were mainly smaller college towns. Corvallis, Oregon, home to Oregon State University, saw the biggest overall decline, as its LQ dropped from 9.3 to 3.7. Other metros that experienced big declines include Fort Collins, Colorado; Boise City, Idaho; Oshkosh-Neenah, Wisconsin and Appleton, Wisconsin. Among large metros, Austin saw its LQ for patenting activity slip considerably, from 3.9 to 2.7
Though the nationwide pace of innovation has fluctuated overall, the specific geography of innovation has remained fairly stable over the course of the crisis and recovery. The Bay Area – and especially Silicon Valley – remains the nation’s unquestioned leader, with innovation strongly concentrated in these knowledge centers. And when looked at on a relative basis (in terms of location quotients), leading college towns like Boulder and Ann Arbor stand out as similar high-performers.
Innovation in America remains spiky. Across the decade from 2001 to 2011, the top 20 patent-producing metros accounted for more than 60 percent (63.7 percent) of all patent applications, the top 10 generated more than 40 percent of the entire nation’s innovative output, and the Bay Area alone produced 13.6 percent of the country's innovative output. Innovation is a main driver of economic growth, so this trend may only serve to reinforce the increasingly uneven development of the U.S. economy in coming years.