HOW ONE REGULATION STANDARDIZED DRONE INNOVATION

Why did clearer rules make drones look more alike?

Before 2016, flying a drone commercially in the United States required case-by-case approval from the FAA. The fastest way to get approved was to copy a design the agency had already accepted, so early drones looked alike out of necessity. Then the FAA replaced that process with a single national rule, Part 107, built around a simple threshold: drones at or above 55 pounds faced stricter requirements. Companies no longer needed to imitate approved designs, and the natural expectation was that products would begin to diversify.

I tracked 21 years of agricultural drone launches to test that expectation, and the data showed the opposite. New designs clustered closer and closer to the 55-pound line, as if the standard itself had become a magnet. But experimentation did not disappear. It survived in particular places, and where a firm was located turned out to matter more than the rule itself.

What I did

To understand how regulation changed what companies built, I created a 21-year product history of 335 agricultural drone models. I combined an industry database covering more than 1,800 platforms with 2,249 Federal Aviation Administration exemption filings, then recovered launch dates and technical specifications through company records, web archives, emails, and direct outreach.

Working with three research assistants, I tracked how close each new design came to the 55-pound cutoff before and after Part 107 was introduced in 2016. I then compared firms exposed to two different kinds of local influence:

• Industry coordination: The National Agricultural Aviation Association (NAAA) and related associations bring operators and companies together around safety, regulation, and accepted industry practices.

• Farmer-facing knowledge: USDA service centers connect agricultural research with farmers through local workshops, demonstrations, and technical assistance.

What I found

• The regulation became a design magnet. After Part 107 was introduced, new agricultural drones clustered significantly closer to the 55-pound standard. A rule intended to clarify commercial operations became a highly visible design target instead.

• Industry associations strengthened the pull toward a common design. Firms in regions with stronger agricultural aviation associations converged even more closely around the standard. Coordination through groups like the National Agricultural Aviation Association reduces uncertainty, but it also makes one design look safer and more legitimate than the alternatives.

• Farmer-facing knowledge networks kept experimentation alive. The opposite happened near USDA Extension programs, where workshops connect universities, farmers, and technology providers. Firms near these networks produced more divergent designs, and this effect was several times larger than the pull of industry coordination.

• The two influences operated at different geographic scales. Industry coordination worked at the state level, where associations set shared expectations. Knowledge-driven experimentation was local, concentrated in the cities where firms, researchers, and farmers actually exchange information.

So what

If you want to predict where unconventional products will emerge in a regulated market, do not map the regulation. Map the places where builders and users learn together.

For policymakers, this means extension programs and workshops are not just channels for spreading finished technology. They are innovation policy. For firms, the same map works as an early-warning system: neighborhoods dense with hands-on learning are where the next product category is most likely to appear.

Coordination helps an industry agree on what works. Local learning helps firms discover what the standard leaves out.

Solo-authored working paper. Presented at the Ronald Coase Institute Workshop on Institutional Analysis (2022) and invited seminars at New York University and Tsinghua University (2023), Syracuse University (2024), and the University of Waterloo (2025). Supported by a National Science Foundation Doctoral Dissertation Research Improvement Grant, Award 2035625.