Innovation Files has moved! For ITIF's quick takes, quips, and commentary on the latest in tech policy, go to

U.S. Cluster Analyses, Circa 1900 and 1909

In perusing the Census Bureau archive at, I found, deeply embedded in multi-thousand-paged censuses of manufacturers for 1900 and 1909, relatively sophisticated analyses of U.S. clusters, organized by industry and locality.

The 1900 version goes on for 25 pages, covering for each of 15 industries, major manufacturing operations by state and city, nominally and in terms of localization (city’s share of the U.S.) and specialization (industry’s share of city’s output).It summarizes these findings into several tables regarding the industry localization and specialization by cities and states. It ends by discussing, in detail, “The Universal Character of the Localization of Industries,” describing how the phenonmenon of industry agglomeration has been well known for centuries and explaining “The Causes of Localization”: nearness to materials, nearness to markets, waterpower, a favorable climate, a supply of labor, capital avialable for investment in manufactures, and momentum of an early start.

I’ve posted the 1900 cluster analysis here:

Download this file

Also available via GW:…

Source materials are here:

The 1909 analysis is shorter, by far, but offers additional reasons for localization, e.g., convenient transportation facilities and the “habit of industrial imitation.”

That analysis is here:

Download this file…

Source materials are here:

A few observations. One, I find it remarkable that the federal government over 100 years ago was putting significant resources into tracking and analyzing our geography of economic activity.

Once upon a time, firms couldn’t grow significantly without being in an agglomerated setting. Transportation and communication technology advances and the development of the branch plant economy changed that to the point that we lost track of the fact that long ago people understood the notion of clustering. In recent decades, it took scholars like Piore, Sabel, and Porter to raise awareness of the importance of agglomeration, but even then at least I haven’t seen the historical antecedents until now.

Two, the studies’ conclusions about the causes of agglomeration, accounting for differences in technology, hold up today.

Three, this work, understandably, is manufacturing focused, given the times. That said, there is a general sense of excitement within the Census of Manufactures documents about the broad growth of the U.S. manufacturing sector across a plethora of industries and the implications for national economic strength. You can see the profound, widespread role played by manufacturing in the economy of the nation, states, and cities.

Four, I didn’t see any effort to link the findings to policy. That may be in other documents.

Fifth, I see a difference between the whole-hearted effort on the part of the early 20th century federal government to understand our economic geography–an interest which largely disappeared after the rise of macroeconomics in the mid-20th century–and the low level of interest now, with a related general lack of understanding of the relation between place, economic policy, and outcomes.

I look forward to your thoughts.

Print Friendly, PDF & Email

  • Mike Rudis

    Quite interesting to find something so old! Just goes to show that the idea that there is a need to connect your key players in any ecosystem is nothing new – and really only becoming more important. However, just as “convenient transportation facilities and the ‘habit of industrial imitation’ ” led to localized clusters, new forces today (the Internet and an even flatter global marketplace) are pushing the need for the next step to foster the growth of truly competitive businesses – a globally networked “virtual cluster”.As part of Global Entrepreneurship Week, my company, Larta Institute, is doing a webinar on this newer approach as it relates to high-tech innovation (Nov 17). If anyone is interested in checking it out, you can get more info an register at