Marco, Carley, Jackson, and Myers, 2015

• Alan C. Marco, Michael Carley, Steven Jackson, and Amanda Myers. 2015. The USPTO Historical Patent Data Files: Two Centuries of Innovation. SSRN working paper 2616724.

Author info

• Alan C. Marco, Georgia Institute of Technology - School of Public Policy
• Michael Carley, Independent
• Steven Jackson, US Department of Transportation
• Amanda Myers, USPTO
• There are 2 versions of this paper; Date Written: June 1, 2015

Abstract (compressed)

Patent classification systems are largely designed for administrative purposes, with extraordinary detail and many small changes over time. A simplified category system for research purposes was developed by Hall, Jaffe, and Trajtenberg (2001). These higher-level classification for the NBER Patent Citation Data File are aggregated from detailed USPC classes into six economically relevant technology categories and 37 sub-categories. This classification scheme has proved useful for researchers investigating granted US patents. Comparable information on pending or abandoned patent applications has not been available. We apply the NBER sub-categories to published and publicly-available unpublished patent applications and to in-force and expired patents to create the USPTO Historical Patent Data Files, four research datasets containing time series and micro-level data by NBER sub-category. These new datasets have annual data back to 1840 on patent applications, patent grants, and patents-in-force. Here we also provide data on the monthly stocks and flows of utility patent applications and grants for 1981-2014, supporting detailed studies of patent application disposal and the dynamics of new filings, pendency, abandonment, litigation, and technological change.

Notes and quotes

• "Patent classification systems were created to ensure that applications are routed to the appropriate examiners and to facilitate efficient prior art searching. Accordingly, classifications are primarily based on the technological and functional features of inventions and encompass the entire spectrum of subject matter that can be claimed."
• The USPC system has "more than 450 unique classes and 150,000 subclasses. . . [and they] are not hierarchical . . . classes 101 (printing) and 102 (ammunition and explosives) do not roll up into any common higher level class. To address this, Hall, Jaffe, and Trajtenberg (2001) developed a hierarchical classification based on aggregating USPC classes into 37 (two-digit) sub-categories, which are further aggregated into six main technology categories. . . The 1999 version of this data file represented a boon in the use of patent data by economists and other researchers." (p3)
• In the USPTO now: "Within the PTO, there are three organizational tiers associated with patent examination: technology centers (TCs), examining art units, and individual examiners.12 When an application is filed with the Office, it is classified and assigned to a TC and art unit based on its subject matter. Each TC specializes in a particular field (e.g. biotechnology and organic chemistry, computer architecture and software, etc.) and is composed of a number of examining art units. Patent examiners reside within art units, and applications are docketed to examiners by a supervisory patent examiner (SPE) within the art unit. Initial classification always precedes examination, though, . . . reclassification may occur during prosecution. The USPC system classifies applications based on subject matter in order to route them through these organizational tiers to the appropriate examiner. Each USPC subject matter division includes a class (which delineates one technology from another) and a subclass (which delineates structural and functional features of the subject matter within the scope of the class)." Subclass info is not used in mapping patents into NBER categories. "Each art unit . . . is responsible for a set of USPC subclasses and contains examiners qualified to examine the technology classified in those subclasses." (p6)
• "With more than 450 classes and 150,000 subclasses, and ever-evolving technology, there is some degree of ambiguity in classification." "Before patents are published, classification experts review the primary class [aka "original class"] and add several “cross-reference” classes to facilitate prior art searching" (p6)
• "The USPC system frequently undergoes revisions - over 1,000 since 1981" (p7)
• "Applications can take several months to receive an initial USPC classification code" as of 2015 (p10)
• The X patents: "2,633 patent numbers are preceded by “X.” In 1836, a fire destroyed many patents and applications, some of which were never recovered; the “X” patents indicate those that were granted before the fire. This means that patent 0000001 (granted 7/13/1836) is wholly separate from patent X000001 (granted 7/31/1790)" (p13)
• "examination decisions by examiners (allowances and rejections) are not technically disposals. For allowances, applicants must pay a fee in order for the patent to be granted. In the data file, we use the day that the patent is published as the disposal, or issuance, date." "Some confusion exists because grants are sometimes called “allowances,” and the grant rate (the number of issuances divided by the number of disposals) is sometimes called the allowance rate" (p14-15)
• After a rejection of an application, the applicant usually "abandons" the application, usually by ceasing to respond to PTO "office actions". Applicants have [other] ways to respond to rejections: appeals to the Patent Trial and Appeal Board or its institutional predecessors, and ultimately to the federal courts. So rejections do not always represent a terminal disposal of applications." (p15)
• "We include only “regular” utility patent applications." We do not include "provisional applications, re-issue applications, and Patent Cooperation Treaty (PCT) applications" (p15)
• "patenting associated with Semiconductor Devices (46), Drugs (31) and Computer Hardware and Software (22) was minimal until [the 1950s, then] climbed rapidly." Some computer-like inventions did exist before 1900. Some patents thus classified from that time were related to photography, e.g. Patent US-1850-7655, Patent US-1860-30850, and Patent-US-1870-104963. Some early “information storage” patents cover inventions related to phonographs, e.g., Patent US-1878-200521 and Patent US-1888-382416. The earliest patented invention classified as a “semiconductor device” (Patent US-1904-755840) covered a detector for electrical disturbances." (p20)
• "The Semiconductor Devices sub-category climbs from the bottom ten to the top three" in 20 years (p21)
• cites Strumsky, Lobo, and van der Leeuw, 2012
• Table 2 is replicated here: NBER patent classifications

Original title	The USPTO Historical Patent Data Files: Two Centuries of Innovation
Simple title	USPTO Historical Patent Data Files
Authors	Alan C. Marco, Michael Carley, Steven Jackson, Amanda Myers
Date	2015
Countries	US
Languages	en
Keywords	Patents, JEL O3
Journal	SSRN
Related to aircraft?	0
Page count
Word count
Wikidata id