This is good—really good. I like Robin’s persistence. He makes me think. And that’s always good. Let me take some time to reply.
First off, I think the passage Robin quotes from my book speaks for itself. There is a correlation between these two things, the gay index and the bohemian index and (1) innovation (measured by patents) as well as (2) high-tech industry concentration (measured by the Milken Index.) The point of that passage is that these are empirical results, and that people went berserk over them in ways that just blow me away. I’m still amazed at the overreaction. I have been a researcher for nearly three decades. I find it frankly amazing that of all the many empirical findings in my research over this time period, it is these two results that have generated a massive debate, or dare I say overreaction. As I said, in independent research both Marcus Noland and Ronald Inglehart have found similar correlations.
As I said in the original article, I did not start out with any sort of agenda to identify these variables or the resulting empirical findings. I have a long history as a student of innovation and economic growth and was plugging and chugging along as usual until I started an interview-based project which began to suggest that factors like “amenities” and lifestyle mattered to the location choices of highly skilled or highly educated people. Then I read Ed Glaeser’s paper on the consumer city and I thought there’s something interesting here.
The problem with a lot of the literature on amenities, location, and economic growth (not Glaeser’s work, by the way) is that the measures of amenities are frankly problematic, unsystematic, and just plain poor. So, my graduate students and I said, “What if we can come up with some better measures?” We surmised that one of these would be a measure of cultural producers—artists, musicians, designers, writers, and what not. This measure is systematic and also has the advantage of reflecting the “revealed locational preferences” of cultural producers. We called it the “Bohemian index,” figuring that these kinds of professions could be somewhat wittily identified as “bohemians.” We were amazed by the statistical associations and wrote them up for an economic geography journal.
The origin of the gay index is a related story I have told before, but which is worth telling again. As I was pursuing the research on amenities and location, my Dean at Carnegie Mellon, Mark Kamlet, mentioned to me that there was a graduate student in the program, Gary Gates, who was looking at amenities and location decisions in his work with Lowell Taylor, Seth Sanders, and others in the context of the demography and location of the male gay community. I said to him, “Huh? What could this have to do with my work on high-tech innovation and growth?” But as a curious colleague I thought, “OK, I’ll meet with Gates.” Gates, who had been doing this research and was already sensitized to its “hot-button” nature, asked me to name my top 5 high-tech centers. I rattled off a list like: San Francisco, Boston, Austin, San Diego, and Seattle. He said, “Rich, you hit the jackpot. You just named 5 of the top 10 leading places for gay men.” The ice was broken and we had a good laugh. After a time we decided to try to see what was there, and ultimately we wrote up a research paper on our findings for the Brookings Institution.
The point is: I did not set out to go after culture issues or weigh in on the culture wars. These two factors—the locational concentration of artistic and cultural producers and of gay men—turned out to have a much more powerful statistical association to concentrations of high-tech innovation and high-tech firms than I ever would have expected.
So I’ve spent the better part of a decade now trying to figure out why. The best I can tell is not that innovative people are bohemians or gay, but rather that locations that are open to diverse groups of people, open to diverse sets of ideas, and that enable self-expression tend to have underlying eco-systems that encourage and attract innovative people generally, including those who would like to apply their talents to scientific, technological, and economic innovation. I really believe there is something to this. I plan to keep on exploring it. And, as I said before, others, like Marc Noland and Ron Inglehart, are generating complementary findings.
There is a bigger debate to be had here, and it is a robust one, with all sorts of people, like Ed Glaeser of Harvard and Terry Clark of the University of Chicago, weighing in on the direction of causality, and also on the scale and unit of analysis.
Terry says the Bohemian index results mainly hold, but that the gay index results are less robust at the county level. We use the Consolidated Metropolitan Area level in our analysis. We agree with Terry: the gay index result is much stronger at the Consolidated Metropolitan Area level—the level we prefer—than at the county level. The reason is that commuting distances are longer these days. Gay people who are technologically inclined, but want to live near significant concentrations of other gay people, can commute to Silicon Valley from San Francisco, or from DuPont Circle to Northern Virginia. Similarly, family people who work in urban centers also can commute from Silicon Valley suburbs to San Francisco, or from Northern Virginia to DC.
Glaeser’s critique is different and quite interesting. In fact, we sent him the data he used in his analysis. Ed ran a simple linear regression with a series of variables: human capital (using educational attainment), the Bohemian index, gay index, and so on. He found that human capital was the most powerful explanatory factor by far.
There are two points I’d like to make here. First, I offer my creative class measure as an alternative to traditional education-based measures of human capital. I say that human capital matters, following Lucas, Glaeser, and others, but I believe that occupation-based measures are useful and perhaps more effective. People like Michael Dell and Bill Gates fail to meet the criteria of the educational-attainment measure, which is bachelor’s degree and above. Interestingly, a study by economists at Utrecht University, which tested the standard educational measure against my occupational measure, found that the occupational measure did substantially better in accounting for economic growth of Dutch regions. I would encourage researchers out there to test the two of these in side-by-side comparisons. I think the occupational measure also has a lot to offer for regional development analyses because it can help identify real regional strengths and weaknesses and can be used to help orient regional growth strategies.
The second issue is somewhat more complicated or arcane, but may be even more important. It has always been self-evident to me that human capital (however measured) has the main effect on innovation and economic growth. But this begs the question, which I mentioned in my earlier posts, of why some locations have more human capital than others. To me, that’s a really big question—perhaps one of the biggest unanswered questions in the literature on growth. Human capital is not a factor that some places are magically endowed with. It’s different from land, raw materials, or natural resources. People are mobile and get to choose the places they want to be. Tiebout pointed this out a long, long time ago. Highly skilled and highly educated people are the most mobile of all. For example, think of the location decisions of university scientists. So, the real question is: Why do some places end up with higher concentrations of human capital than others?
This is the real question my work poses and tries to answer. It seems to me that a very simple and plausible answer is a place’s openness to diversity. I refer to it technically in my work as a place’s having “low barriers to entry” for human capital. That’s the real force I see at work—not coffee shops, or bohemian lifestyles, or gay neighborhoods. But I say as much in my work.
Methodologically, this is very hard to get at. In a paper I wrote for a mainline geography journal, we used path analysis to get at this. And the results of that analysis are pretty clear. Measures like the gay index and the bohemian index, while having a weak direct effect on innovation and growth, have a powerful indirect effect on both, working through the intermediate variable of human capital.
Another way to get at this would be to use a simultaneous system of equations to tease out, first, the effects of openness and other factors on the location of human capital and then, second, the effects of human capital on innovation and growth. This is a line of research I have been working on, and would like to encourage others to probe as well.
The point is that saying cultural factors matter to innovation and economic growth, and trying to show that they do matter in a fact-based, empirical manner, is quite different from trying to stoke the fires of the culture wars. My honest best sense of the culture wars is that in an environment so hotly ideologically and politically charged we need more, not less, objective research on these matters.
Robin, thank you for honestly engaging this debate. Since we’re both at GMU, maybe we can get some of our graduate students to work on the model and the data.
What do others think?