What the Rural–Urban Innovation Gap Actually Measures
The subject of this series is the U.S. rural–urban innovation gap: the difference in innovation activity reported by firms in rural and urban counties, and what that difference implies for place-based policy. This opening asks a narrower question before it asks how to close the gap: what, exactly, is being compared? The answer gives the reader a practical way to judge the evidence that follows—separating a measurable difference between groups from the much stronger claim that geography itself determines what firms can do.
That distinction matters because “rural innovation” can sound like a contradiction if innovation is imagined only as a patent, a laboratory, or a fast-growing technology company. It is also easy to swing to the opposite mistake and conclude that geography does not matter at all. The available evidence supports neither shortcut. Rural and urban firms operate in different economic environments, but those environments contain both constraints and capabilities. The useful question is how they shape opportunities for firms to innovate, which forms of innovation are visible, and which policy responses are plausible.
The subject is a relationship, not a stereotype
A 2026 Census working paper studies this relationship with firm-level data from the 2018 Annual Business Survey, whose data reference year is 2017. It compares firms in counties classified as metropolitan and nonmetropolitan using the U.S. Department of Agriculture’s Rural–Urban Continuum Codes. In the paper, urban or metro counties are RUCC 1–3; rural or nonmetro counties are RUCC 4–9. That definition is not a timeless boundary between two kinds of America. It is the classification used to make a particular comparison reproducible.
The unit of analysis is just as important as the geography. The study focuses on single-unit, nonfarm employer firms. A multi-unit company may report one location while its innovation practices are spread across establishments in several places. Because the survey does not reveal which establishment carries out a particular innovation, assigning the whole company’s activity to one local context would blur the relationship the study is trying to examine. Restricting the sample makes the location link more defensible, but it also narrows the population to which the results apply.
The subject, then, is not “rural counties” in isolation and not “rural people” as a general category. It is the relationship between a firm’s location and its reported innovation activity, viewed through the conditions surrounding the firm. The paper compares averages across groups; it does not claim that every rural firm performs below every urban firm. A rural average can be lower while individual rural firms are highly innovative, and an urban average can be higher while many urban firms make no meaningful innovation at all.
That relationship matters because innovation affects more than the fortunes of a few technology firms. It can change how businesses produce, organize work, reach customers, use resources, and deliver services. If firms in some places have fewer opportunities to improve products or processes, regional inequality can persist even when national growth continues. The rural–urban gap therefore touches economic growth, local competitiveness, and the ability of communities to participate in new markets. But its importance does not justify treating a single statistic as a complete diagnosis.
This is the basic map for the series. A place does not send an instruction directly to a firm. It influences access to knowledge, labor, infrastructure, capital, customers, and relationships. Those conditions interact with what owners and firms already possess. The result is an observed pattern in one or more innovation measures. A policy can respond to a condition in that chain, but the pattern alone does not prove that changing the condition will produce a particular result.
Why the gap depends on what counts as innovation
The most familiar rural–urban comparison uses patents, research spending, or high-tech firm formation. Those indicators are valuable for some questions, especially questions about formal invention and frontier technology. They are not a sufficient measure of everything an operating business changes. A manufacturer that redesigns its production flow, a distributor that improves logistics, or a local service firm that changes pricing and customer management may innovate without producing a patent.
The Census study therefore looks at three self-reported outcomes. New-to-market innovation concerns a product or service that no other company has provided to the market. Process innovation covers new methods of manufacturing, logistics, distribution, and supporting activities. Organizational and marketing innovation includes new business practices, supply-chain and quality management, design and packaging, and new ways to promote, place, or price a product. These outcomes do not make patents irrelevant; they make clear that patents cover only one slice of business change.
The choice of measure can alter the apparent geography of innovation. Rural economies have different industrial mixes, and rural employment is more concentrated in sectors that patent little or not at all. Dividing patent counts by population can therefore make a place look less inventive partly because the indicator favors the activities and occupations concentrated in metropolitan areas. Earlier work summarized by the study also uses a latent innovation index to detect capabilities that conventional measures miss. Such a measure can reveal potential embedded in local industries and relationships even when patent output is low.
A broader measure does not automatically erase the rural–urban difference. It changes the question from “Where are the most patents?” to “How often do firms report different kinds of innovation, and what conditions are associated with that activity?” The answer may still show an urban advantage on average. It may also show that the advantage is smaller for some forms of innovation, or that rural strengths appear in places a patent count would overlook. Measurement is not a technical preface to the real story; it helps determine what the real story is.
Self-reported measures have their own boundaries. Firms may interpret survey questions differently, and a reported innovation does not by itself tell us how valuable, durable, or commercially successful it became. The indicators are also binary outcomes: they record whether a firm reports a type of innovation, not the full scale or quality of the change. A careful comparison can use them to identify patterns without turning them into a universal ranking of rural capability.
A better question than “urban or rural?”
Once the subject and measurement are clear, the central question becomes more precise: which differences in conditions help produce the observed rural–urban gap, and which differences merely travel with location? The study organizes its explanation around three levels. Owner characteristics include education, age, gender, ethnicity, foreign-born status, and multiple ownership. Firm characteristics include size, manufacturing, in-house research and development, family ownership, and cloud-service adoption. County context includes broadband, inventive occupations, local innovation potential, education, amenities, and social capital.
Calling these factors “endowments” means treating them as resources or conditions that firms and places bring to the innovation process. A county with fewer specialized workers may make it harder to recruit for a technical role. A firm with limited access to broadband or cloud services may have fewer ways to coordinate, test, or reach customers. A community with strong local trust may support cooperation, even while its firms remain farther from outside markets or specialized knowledge. These are different mechanisms, so they should not be collapsed into one label called rural disadvantage.
This framing also leaves room for spatial agglomeration and firm selection. Dense urban areas can offer knowledge spillovers, specialized labor pools, transport, amenities, and networks. Rural firms may face greater distance from those resources and a smaller critical mass of related businesses. At the same time, innovation-oriented start-ups may choose urban locations precisely because those advantages are available, while more mature or routine firms may relocate toward lower-cost peripheral areas. An average difference can therefore reflect both what a place provides and which firms settle there.
The study uses an Oaxaca–Blinder decomposition to examine that distinction. In simple terms, it asks how much of the average difference is associated with rural and urban firms having different observed characteristics, and how much remains in a residual component reflecting different estimated relationships between those characteristics and innovation. The first part can point to an endowment gap. The second may contain structural barriers, omitted factors, measurement differences, or model limitations. “Unexplained” is not another word for “caused by rurality.”
This method supplies a disciplined starting point for policy judgment, not a final verdict. If an observed gap is closely associated with broadband, skills, or firm capabilities, those areas may deserve attention. But the decomposition does not show that a broadband investment, a training programme, or a grant will close the gap. That requires evidence about implementation, take-up, alternative explanations, and outcomes over time.
The practical conclusion for this series is therefore modest but consequential. Geography matters because it shapes the conditions under which rural and urban firms operate. It is not enough to say that rural firms lack innovation, and it is not enough to say that location is irrelevant. The evidence invites a better diagnosis: identify the innovation being measured, identify the endowment or access difference connected to it, look for rural capabilities that offset the constraint, and keep the causal uncertainty visible.
The next step is to examine the first part of that diagnosis more closely: when the indicator changes from patents and high-tech creation to firm-reported products, processes, and marketing practices, what kinds of rural innovation become visible—and what remains outside the measure?