The Power of Density

Density is a key factor in innovation and economic growth. The dense geographic clustering of economic activities was true of the industrial behemoths of the past - steelmaking in Pittsburgh and automotive production in Detroit. And, despite advances in communications technology, it applies even more so today: from high-tech firms in Silicon Valley to film producers in Los Angeles and recording studios and record labels in Nashville. There's no doubt: The geographic concentration of firms, industries, technologies, people, and other economic assets plays a powerful role in innovation and economic growth.

The great economist Alfred Marshall long ago outlined the dynamic of agglomeration - that is, the process by which co-location of related economic activities and assets shapes industries and economic development. Jane Jacobs showed us how the clustering of diverse groups of people, firms, and industries in cities provides the basic engine of innovation and new product development. Harvard's Michael Porter has shown how clusters of related industries, customers, and suppliers power innovation and growth. Density makes it easier for people and firms to interact and connect with one another, and it reduces the effort, friction, and energy that's used to make these connections. Density increases the speed at which new ideas are conceived and diffused across the economy, accelerating the speed with which new enterprises and new industries are created.

The curious thing is that most of our key economic and innovation measures don't take density explicitly into account. Economists, economic geographers, and other social scientists tend to normalize the numbers they're interested in by population, representing the data on a per person or per capita basis. This approach has led to all sorts of important empirical insights and findings. But since density itself is an important factor in certain kinds of economic growth, it's useful and important to develop indicators that take it explicitly into account. For that, we need to look at the distribution of activities and key variables across space. So instead of measuring them on a per capita basis, we can examine them on the basis of land area or per square kilometer.

A while back, I posted about this analysis by Rob Pitingolo (h/t: Don Peck) which looked at the density of human capital. Pitingolo developed an intriguing metric that he called "educational attainment density." Instead of measuring human capital or college degree holders as a function of population, he measures it as a function of land area - that is, as college degree holders per square kilometer. He did this for the primary urban centers of metropolitan areas.

Inspired by this, I worked with my Martin Prosperity Institute colleague Charlotta Mellander to build indicators of density for a wider range of key economic and demographic variables. We conduct our analysis at the metropolitan level. It's important to point out that there are limits to using the metropolitan area as a unit of analysis. Metropolitan areas combine core cities with their suburbs and come in all different shapes and sizes. Some are more concentrated at the core (like Portland), others more sprawling (like Phoenix). Examining the distribution of key economic, social, and demographic variables at the metro scale is admittedly crude. But it is also a useful and important starting point, since the metro level is by far the most common unit of analysis in studies of regional economic development. In our research on the subject, we're interested in developing new, more precise metrics and indicators of density within metropolitan areas - comparing central cities or urban centers to suburbs and probing the distribution of density across Census tracts and zip codes, which I will report in future posts.

We also compare our density measures to population density, to see which metros over- and under-perform relative to their populations. To get at this, Mellander performed a residual analysis - a statistical procedure which systematically compares how metros perform on a given factor compared to what we'd expect based on their population density. We also look at the associations between our various density measures and key metrics for regional economic development - wages, incomes, innovations, and regional economic output. As usual, I'll point out that these are preliminary, exploratory analyses that simply point to associations between variables. We don't make any claims here about the direction of causality, and we acknowledge that intervening variables may come into play.

Over the next couple of weeks, I'll report the key findings from our analysis. Later this week, I'll look at density of human capital - based on the conventional measure of people with a bachelor's degree and above. Then, I'll turn to the density of the creative class - that is, of people employed in science and engineering, business and management, health care and law, and arts, culture, design, media, and entertainment. The fourth post in this series will look at the density of a subset of this group - artistic and cultural creatives. In the fifth post, I'll share our findings on density of innovation and high-tech industry. And, in the final post in the series, I'll bring it all together and sum it up with maps of these density measures.