Great universities are a necessary but insufficient condition for high-tech growth.
Americans were up in arms last month when a Princeton alum wrote a letter to the school's "young women" urging them to spend their college years seeking an ideal mate from among the pre-screened student body.
Many of them will, in fact, do just that. But this isn't a sorting that occurs by virtue of college admissions offices. It reflects the even deeper geographic sorting of Americans by talent and skill which, as Ross Douthat pointed out recently in his New York Times column, is "one of the striking social facts of the modern meritocratic era" (kindly citing my 2006 Atlantic essay, "Where the Brains Are").
Douthat's colleague and Times' Washington bureau chief David Leonhardt weighed in recently with a provocatively titled post, "How the Location of Colleges Hurts the Economy." Leonhardt points out that America's elite colleges and universities no longer match up with its leading centers of economic activity. Most of America's elite universities were founded a century a more ago, he notes, well before the birth of many of the nation's most important companies — Apple, Amazon, Facebook, Google, Microsoft, Home Depot, and Walmart.
Moreover, America's elite colleges are concentrated in a relatively small number of states that do not necessarily align with more recent, rapidly-growing areas of the U.S. economy. Based on data compiled by the Times from Barron's Profiles of American Colleges, he notes that:
The country's most selective 236 colleges – colleges that tend to have more resources and much higher graduation rates than others – are especially concentrated in New England and, to a lesser extent, the mid-Atlantic. The Midwest, Southeast and California have fewer such colleges than the Northeast, but still noticeably more than Florida and Texas. The Mountain West has just a handful.
This isn't just a convenience problem. He cites research that points out that young people who live near colleges are more likely to actually go to college and that students who live close to selective colleges are more likely to go to selective colleges. As he points out:
Because students who attend a selective college are more likely to graduate than similar students who attend a community college or nonselective four-year college, the geographic dispersion of colleges creates winners and losers. Students who live near selective colleges benefit – and those who do not pay a price. The economy also pays a price, because so many talented teenagers, many of them in smaller metropolitan areas in the Sun Belt and Mountain West, fall far short of their potential.
I'm a case in point: A working class kid from New Jersey, I benefited immeasurably from a Garden State scholarship that made it possible for me to go to nearby Rutgers College.
The connection between universities and economic development has long been of interest to me. So, I thought it both interesting and useful to do some analysis of the geographic connection between elite universities and state economies. With the help of my colleague Charlotta Mellander, we ran a correlation analysis between Leonhardt's measures of elite colleges and universities and a variety of state-level economic, demographic, and other factors. We removed Washington, D.C., from the analysis, because it is an extreme outlier with a much larger number of elite colleges and universities than any other state. D.C. really isn't comparable to other states, after all: Its relatively small geographic size and high density makes it more like a metro than a typical state.
As usual, I point out that correlation does not equal causation and that it points only to associations between variables. This is especially true of the complex, multi-faceted relationship between universities and economic development where many other factors surely come into play.
As numerous economic studies have pointed out, it is especially hard to disentangle causality between universities and economic growth and the precise cluster of factors that mediate the relationship between the two. It's also important to note that the connection between universities and economic development, according to most studies of the subject, involves the kind of proximity that can be best gauged at the metro as opposed to the state level. Even so, the findings of our analysis are interesting across several dimensions.
First off, per Leonhardt's contention, Mellander's analysis picks up no statistically significant association between the location of elite colleges and universities and two key indicators of state economic performance: economic output per person and the share of high-tech companies. This suggests a disconnect of some sort between elite universities and the economy. On the flip side, Mellander's analysis finds a positive correlation between elite colleges and universities and average wages (.39) at the state-level. This suggests that having relatively more elite colleges and universities is associated with higher state wages.
What might be behind these divergent results?
For one, the reason that universities have a bigger effect on some places than others may have something to with the place they are located as well as the status or activities of the university itself. Elite universities have clearly played a big role in innovation and entrepreneur-fueled economic development in places like the Bay Area, Cambridge-Boston-Route 128 corridor, greater Seattle, North Carolina's Research Triangle, greater San Diego, and Austin, Texas. In some of these places, great universities preceded economic growth; in others, local growth provided the resources required to bolster the university.
But there are many other places with elite universities that have seen far less dynamic high-tech growth. There are other regions that have experienced substantial economic decline even though they are home to great universities. My own take on this is simple: Great universities are a necessary but insufficient condition for high-tech growth. Economic success also requires a broader innovation ecosystem that can absorb and utilize the innovations generated by local universities.
If not, the technology produced by local universities, however elite, will simply flow elsewhere. As I pointed out more than a decade ago:
Michael Fogarty and Amit Sinha of Case Western Reserve University in Cleveland have examined the outward flow of patented information from universities and have identified a simple but illuminating pattern: There is a significant flow of intellectual property from universities in older industrial regions such as Detroit and Cleveland to high-technology regions such as the greater Boston, San Francisco, and New York metropolitan areas. Their work suggests that even though new knowledge is generated in many places, it is only those regions that can absorb and apply those ideas that are able to turn them into economic wealth.
The key mediating variable here is talent or human capital. "Universities have been naively viewed as 'engines' of innovation that pump out new ideas that can be translated into commercial innovations and regional growth," I wrote in the same essay, adding that the conventional technology push view of university-driven growth "misses the larger economic picture: Universities are far more important as the nation's primary source of knowledge creation and talent."
In fact, Mellander's analysis finds a close relationship between elite universities and talent at the state level. There is a substantial positive correlation between elite colleges and universities and the share of state residents that are college graduates (.46). My take: even though elite universities are not related to higher levels of economic output per person, they operate on state economies indirectly through their relationship to greater human capital accumulation (which, as many studies have shown, is in turn associated with higher wage levels).
Elite universities are also closely connected to the occupational structure and class composition of the workforce — additional evidence in support of their talent-oriented role, Mellander's analysis finds a positive correlation between elite universities per capita and the share of the workforce in knowledge-based and creative occupations (.49). Conversely, there is a negative correlation between elite colleges and universities and working class economic structures (-.33). This connection provides another indirect channel through which elite universities bolster state wages.
The relationship between elite universities and state economic development may also turn on politics. Elite universities hew closely to America's red state/blue state divide. Elite colleges and universities are positively and significantly correlated with the share of state residents who voted for Obama (.59) and who identify as political liberals (.55), and negatively and significantly associated with the share of residents who voted for Romney (.-59) and who identify as political conservatives (-.58). This, in my view, provides a missing link of sorts in explaining how and why elite universities relate to state economies. Bluntly put, blue states appear more willing to invest in elite colleges and universities; red states less so. Of course, all of this is interrelated and mutually reinforcing. Liberal states tend to spend more on higher education; states with more elite universities tend to accumulate higher levels of human capital; higher levels of college grads encourage more spending on higher education, and so on.
All of this suggests that the economic effects of elite universities are etched fairly deeply into America's economic landscape and that they are highly path-dependent and long-lasting.
Leonhardt suggests that forces may be causing higher education to change in ways that better align universities to America's evolving economic landscape. "Philanthropists in Birmingham, Miami, Oklahoma City, Phoenix or Salt Lake City might one day decide to follow the path of the 19th-century donors who founded Stanford and the University of Chicago," he writes. Call me cynical, but I'm quite a bit less optimistic that America's more recently growing states possess the political will (and underlying demographic makeup) to mobilize the needed investments.
Leonhardt concludes by predicting that "higher education will change more over the coming century than it did over the last century." I don't doubt that. But my hunch is that the direction of change is likely to lead toward even greater concentration and geographic "spikiness" of elite universities than today.