More than half of US-origin technologies in a newly constructed research network build on at least one technology that first appeared abroad, according to an October working paper examining the international inputs to American invention.
Tarek Alexander Hassan, Aakash Kalyani and Juan Cruz Lopez Del Valle analyzed the text of 13 million patents filed with the US Patent and Trademark Office between 1887 and 2024. Their National Bureau of Economic Research working paper traces relationships between older technologies and the patents associated with later ones.
The headline result, 54%, concerns the technologies identified as US-origin within this network. It is not a finding that 54% of all American patents, research spending or economic output comes from overseas.
For companies managing research and development, the paper raises a practical issue: the knowledge needed for a new product may sit outside the firm, its industry or its home country. Finding it is only the beginning of making it useful.
The researchers followed technologies rather than just patent counts
A patent document can describe several technologies, while the same technology may appear in many documents. Counting patents alone therefore does not reveal how different technical ideas fit together.
The authors used technology names drawn from Wikipedia to identify 20,786 technologies in the patent text. They assigned an origin to the country whose patenting activity for a technology first crossed a minimum threshold.
They then examined the early patents associated with each technology. An older technology qualified as an input, or “parent”, when it appeared in at least 10% of those founding patents, which were filed within five years of the newer technology’s measured birth.
This creates a genealogy: a network showing which earlier technologies appear in the development of later ones. The researchers found foreign inputs especially common among US technologies that subsequently generated many descendants.
The origin measure is a research classification based on US patent records. It should not be treated as a definitive historical ruling about who invented a technology. Countries differ in their use of the US patent system, and important knowledge can remain unpatented.
Having access to an idea differs from putting it to work
The study tracks a technology’s absorption into a US county when local inventors begin mentioning it in their own patents. That is evidence of use within the local patenting process, rather than simply being able to find a document online.
To investigate the direction of the relationship, the authors use historical migration patterns as a source of variation in counties’ ancestry and their exposure to particular countries’ knowledge.
They report that counties more exposed to ideas from a foreign country absorbed more technologies from that country and developed more US technologies building on them.
The authors interpret their estimates as causal under the study’s identification assumptions. Migration was not randomly assigned, and the analysis must separate knowledge exposure from other characteristics of local economies. The paper is a working paper that has not been peer reviewed.
The research also connects older knowledge absorption with later online job postings mentioning related technologies. Job advertisements indicate employer demand for skills; they do not count completed hires or establish a net increase in employment.
A useful search begins with the technical problem
For a business, a patent search can start with the function a product needs to perform rather than the name of a direct competitor. An engineering problem involving heat, moisture or a material’s strength may have been addressed in a different application.
The World Intellectual Property Organization’s PATENTSCOPE service provides a practical starting point for searching published international patent applications and documents from participating national and regional collections.
Keyword searches can be combined with patent classifications, which organize documents by technical field. Trying alternative technical terms and following related documents can reveal work that a search for a familiar product name misses.
A useful search record would note the problem being addressed, the proposed solution, the relevant document and what remains untested. That gives an engineer something to assess. A list of documents without an explanation of relevance is a weak basis for committing development money.
These are practical implications for research management, not a search method tested by the economists. The paper does not establish that adopting any particular foreign idea will increase a company’s profits.
Public disclosure does not mean free commercial use
Published patent information can explain how an invention works while rights still restrict its commercial use.
WIPO’s patent guidance explains that patent rights are territorial and that an owner can license an invention on agreed terms. Finding a technical description is therefore different from establishing permission to manufacture, use or sell the protected invention in a particular market.
A company considering a product needs to check the relevant rights, their legal status and the intended markets. An invention may be described in an international application without having an enforceable patent everywhere. Conversely, several rights can be relevant to one product.
A patent search can help identify technical knowledge and possible rights to investigate. It is not, by itself, a legal clearance for a product launch.
The company still needs people who can evaluate the idea
Absorbing knowledge requires engineering judgment, testing and the ability to connect a proposed solution to a real production process. A promising material may still fail a durability test or prove too expensive at the required volume.
Our reporting on AI-assisted patent drafting and professional expertise examines a separate experiment in which better current output did not translate into equal lasting skill gains for all users. It illustrates why access to a tool or document and the ability to judge its output need separate attention.
A development team can assess an outside idea through a small technical trial before committing to a full redesign. Its record should separate demonstrated performance from the inventor’s claims and identify any licensing, integration or supply requirements.
The economic study measures connections in inventive activity, not the cost of those commercial steps. It also does not test specific export controls, trade restrictions or research policies.
Its contribution is to make part of innovation’s accumulated knowledge visible. For companies planning a new product, the resulting question is concrete: which missing technical input already exists, and what would it take to understand, test and use it lawfully?
