An analysis of urban innovation in the United States

Research question and approach

Innovation activities are strongly clustered in space and this spatial concentration is often self-reinforcing, through path-dependent and place-dependent success-breeds success mechanisms. As such, most countries present innovation maps with peaks of innovation activity in specific places, like in the famous Richard Florida’s “The World is Spiky” picture. Even more specifically, innovation spikes appear to correspond to large metropolitan areas, leading to the claim that urban agglomeration disproportionally drives innovation, as tested in urban scaling studies. The case of the United States has featured prominently in this discourse, as the ultimate example of a country where most innovation happens in a handful of cities with world-leading technological capabilities.

In this project I propose that the current perspective on urban innovation in the United States (U.S). can benefit from moving beyond invention and technological innovation only, towards a focus on commercialized innovation, inclusive of non-technological or so-called ‘soft’ innovation.  I argue that broadening the notion and measurement of innovation allows drawing a richer map of urban innovation across U.S. cities.

Hence, the question that motivates this work is: ‘do the narrative and stylized facts on urban innovation in the United States change if one goes beyond technological invention only?

The question is both a conceptual and empirical one. The overwhelming majority of studies on urban innovation has used patents as the standard innovation metric, hence only focused on technological invention.  As Carlino and Kerr (2015) put it: “Innovation comes in many shapes and sizes, except in economic studies” (p.2). This percolates to policy narratives as well, with a tendency of (local) policymakers to focus on the shiny and exciting initial phases of innovation processes. The same policymakers seem to disregard or rank as inferior, those other innovation activities where local firms can specialize in, for instance, design and product development. Similarly, policymakers and investors in different places tend to be attracted by the same high-tech (manufacturing) sectors and to fixate on Silicon Valley replica’s. This implies that innovation in other sectors remains hidden and is not further leveraged as a ‘real’ source of jobs and prosperity.

I argue that complementing patent metrics with two other metrics, namely trademarks and design rights, allows to capture other stages of the innovation process. All three are intellectual property rights, formal appropriation tools widely used by innovative individuals and companies.  This project can leverage  recent work on validating trademarks as innovation metrics.

An important point is that the three metrics are to a large extent highly correlated, exactly because they refer to different stages of the innovation process. Yet, they also tell a separate story, because: (i) not all firms and not all places master each stage of the innovation process, and (ii) some innovation processes do not follow the same stages, in particular they do not rely on technological novelty discovery but focus on downstream or user-driven learning.

The empirical analysis focuses on U.S. metropolitan statistical areas (MSAs) and leverages data from the USPTO on utility patent, design patent and trademark registrations. The focus is on the years 2010-2015, to compare results with recent studies using patents only.

Each intellectual property right has its own specificity, as summarised in the Table below. Combining all three, one gets a better picture of different stages of the innovation process, but also different types of innovation, in particular ‘softer’ forms of innovation.


Table: Specificities of three innovation metrics,

  Patents Design rights Trademarks
Subject matter Novel technological inventions Novel designs Distinctive symbols
Stage of the innovation process “Technology”





Design and Prototyping

Product development



Product Development


Main legal requirements at USPTO Novelty

Industrial applicability


Industrial applicability


(Intention to) Use in market

Knowledge/occupations involved Technical Design Symbolic
Sectoral context Mostly high tech sectors Specific manufacturing sectors All sectors


Preliminary results

The empirical analysis revolved around reassessing a few stylised facts about urban innovation in the United States. See the Visualizations page to check out the patterns.

Stylised fact 1: Innovation concentrates in the largest metropolitan areas (so-called super-linear scaling of urban innovation, see Bettencourt at al. 2007) in the U.S.

This project’s results: The results broadly confirm urban scaling laws for all three innovation metrics. Urban agglomeration is positively correlated with the ability of local actors to develop innovation along the whole innovation process, for reasons specific to each stage. Yet, the ranking of cities for the three metrics is not the same. Cities can demonstrate capabilities focused on one of the innovation stages (technology, design or market). Hence, using a broader set of metrics allows giving credits to more cities for their innovation than relying on technology-metrics only.

Stylised fact 2: Innovation is more concentrated than invention in the U.S. (see Feldman and Kogler, 2010, p.385).

This project’s results: The analysis provide a more nuanced assessment of the original stylised fact, which was based on earlier studies focusing on product innovation in manufacturing only. Invention, design and innovation activities are all strongly concentrated, but they each concentrate in different U.S. cities.


Stylised fact 3: When looking at inventive intensity, secondary cities in the U.S. turn out to specialise in invention even without access to urban externalities (see O’Huallachain and Douma, 2020)

This project’s results: We also look at innovation intensity using the other two metrics and find that secondary cities can also appear as places that specialize in soft innovation. So not all secondary cities specialize in technology, but we see many other places on the map of innovative specialization than one would not spot as innovation centers with patent metrics. In particular, high innovation intensity can be observed without high invention intensity, since innovation can rely on adoption of technology instead of development of new technology.

The analysis also suggest two new stylized facts.

First, when one compares patents by location of inventor and by location of owner, ones sees a much stronger concentration of invention, even higher than the concentration of (soft) innovation.

(New) Stylized fact 4: Invention is more concentrated in ownership across U.S. cities than innovation.

Second, when one looks at the correlation between invention and innovation intensity, one finds that the two tend to align but not so strongly. Only a few cities specialize in both stages of the innovation process.

(New) Stylized fact 5: Urban specializations in invention and innovation tend to align, but U.S. cities often specialize in one specific phase of the innovation process.


Note: at a later stage of the project the data by US MSA will also be shared on this website.