Industry Classification of Patent Data

Patent data are usually classified by functional use instead of industry of use or industry of manufacturing. For example, a fishing reel and a bobin are both classified under U.S. Patent Class 242: Winding, Tension, or Guiding. This makes it difficult to link patent data to other data that are classified by Standard Industrial Classification (SIC).

The most correct solution to this is to manually classify the patent in question into the SIC by reading through the patent grant document that specifies the invention in details. However, this is an insurmountable task. Alternatively, (for contemporary patent data) we can utilize a patent-industry concordance system that assigns a probability associated with each patent class and each industry class, and maps a patent count for each patent class into a patent count for each industry class. For example, if a probability that a patent awarded in class A is made by industry B is 0.2 and if there are 100 patents awarded in class A, then such a concordance system will yield 20 patents made by industry B.

There are two patent-industry concordance systems that convert International Patent Classification (IPC) into SIC:

The two systems are quite similar and were developed from the Canadian patent data (that have both SIC and IPC untill the early 1990s). Nevertheless, there are a few notable differences between the two systems. The YTC is a basis for the newer OECD Technology Concordance (OTC), and was developed using the Canadian patent grants data during 1983 and 1993. As a result, the YTC generally makes large errors in predicting patents by industry after 1989. In contrast, Silverman’s Concordance utilizes Canadian patent grants during 1990 and 1993 which I think leads to smaller errors in later years. Moreover, some manufacturing sectors (for example, shoe and leather, printing, and furniture industries) are missing from the YTC output matrix.

In addition to the two patent-concordance systems discussed above, sometimes we may need to perform another conversion before being able to utilize these concordance systems. For example, patents granted by USPTO are mainly classified according to the U.S. Patent Classification (USPC). Therefore, we will need Wellesley Technology Concordance (WTC) that converts USPC to IPC before using either YTC or Silverman's Concordance.