Business Dynamics Statistics of Patenting Firms (BDS-PF)
Business Dynamics Statistics of Patenting Firms (BDS-PF)
The Business Dynamics Statistics of U.S. Patenting Firms (BDS-PF) is an experimental data product developed in collaboration with the United States Patent and Trademark Office (USPTO), which extends the set of statistics published by the Business Dynamics Statistics program. BDS-PF is a component of a broader initiative aimed at improving the measurement of the business dynamics of innovative firms (BDS-IF), described in greater detail in Goldschlag and Perlman (2017). BDS-PF provides annual measures of the business dynamics of firms that recently received a patent grant. For additional details about the classification of firms based upon patenting activity, see the Methodology page.
Jump to:
The BDS-PF show key economic data including the number of establishments, firms, employment, job creation and destruction, establishment openings and closings, and the number of startups and firm shutdowns for firms with and without a recent patent grant. Specifically, they track firms with a patent grant in years t, t-1, or t-2. The BDS-PF provides annual statistics for 1978 to 2022 by patenting status and a series of firm and establishment characteristics including size, age, industry, and geography. Below we summarize some of the patterns in the BDS-PF.
Patenting is a relatively rare activity for firms in the U.S. economy. About 0.24% of firms in the early 1980s had a recent utility patent grant, rising to 0.5% by 2022. This amounts to about 9,100 firms in the early 1980s and 27,800 firms in 2022. The BDS-PF tabulations also allow us to see how patenting firms differ from other firms in the economy. On average, patenting firms tend to be older and larger. In 2022, about 13.6% of patenting firms were at least 45 years old, compared to about 5.6% of all firms. Also in 2022, 2.1% of patenting firms were large, with 5,000 or more employees, while about 0.03% of all firms had that many employees.
Interestingly, the share of young firms—i.e., those less than six years old—has been much more stable among patenting firms over the past several decades than among all firms. Figure 1 shows the percentage of all firms (in red) and all patenting firms (in blue) that are young. The share of young firms in the economy declined from about 44.2% in 1990 to 32.2% in 2014, then began to rise to 36.1% in 2022. Among patenting firms, by contrast, the share of young firms mostly remained between 20% and 25% over the past several decades. Moreover, the rise in the share of young firms begins earlier for patenting firms, turning in 2011, than for all firms, which only begins to rise after 2014.
Figure 1. Percent of all Firms and Patenting Firms that are Young
Figure Notes: Figure shows the percentage of all firms that are less than six years old in a given year (red line) and the percentage of patenting firms that are less than six years old (blue line). Patenting firms are identified as those with a utility patent in years t, t-1, or t-2 (bds2022_pfg_utly_fac.csv, Data Management System (DMS) number: P-7083300, Disclosure Review Board (DRB) approval number: CBDRB‑FY25‑0097, CBDRB‑FY25‑0143).
Even though patenting firms are much larger, on average, than other firms in the economy, there has been a secular increase in the share of patenting firms that are small. Figure 2 shows the percentage of patenting firms with 1 to 19 employees. The percentage of patenting firms that are small rose from about 30.7% in the early 1980s to 50.4% in 2022. Meanwhile, among all firms, the percentage of firms that are small fell from about 89.5% in the early 1980s to 88.8% in 2022. Naturally, because they are small, firms with 1 to 19 employees account for a relatively small share of employment among patenting firms. The percentage of patenting firm employment associated with small firms is about 0.23%.
Figure 2. Percent of Patenting Firms that are Small
Figure Notes: Figure shows the percentage of patenting firms that have 1 to 19 employees. Patenting firms are identified as those with a utility patent in t, t-1, or t-2 (bds2022_pfg_utly_fzc.csv, DMS number: P-7083300, DRB approval number: CBDRB‑FY25‑0097, CBDRB‑FY25‑0143).
Patenting activity varies significantly across sectors. Using the BDS-PF tabulations, we can observe how many firms in different sectors are granted patents. Figure 3 shows the percentage of firms in a subset of sectors with a patent grant. The highest patenting firm share is in the Manufacturing sector (31-33), where about 3.7% of firms (8,200 firms) had a recent patent grant in 2022. The next three sectors with the highest percentage of patenting firms in 2022 were Information (51), Wholesale (42), and Professional, Scientific, and Technical Services (54). The Information sector saw the largest increase in patenting activity over the past several decades, rising from about 0.73% (330 firms) in the early 1980s to about 3% (2,200 firms) in 2022. The share of firms with patents in the Education sector (61) and Retail sector (44-45) was more stable over time, at roughly 0.37% (215 firms) and 0.18% (1,200 firms), respectively.
Figure 3. Percent of Patenting Firms by Sector
Figure Notes: Figure shows the percentage of firms with at least one establishment in a given sector with a recent utility patent grant (bds2022_pfg_utly_sec.csv, DMS number: P-7083300, DRB approval number: CBDRB‑FY25‑0097, CBDRB‑FY25‑0143). The “Other (exl. 55)” line includes all other sectors not shown in the figure except the Management of Companies and Enterprises sector (55). Sector 55 is excluded because the overall share of patenting by firms with at least one establishment in that sector is significantly higher (between 8 and 11%), which is driven by the fact that many patenting firms are large, multi-unit firms with headquarters establishments in sector 55. Note that firms can operate establishments in multiple industries.
One unique feature of the BDS-PF tabulations is the ability to observe what types of patents are being granted to firms in different industries. In the BDS-PF tabulations, we include classifications of firms based upon the World Intellectual Property Organization Technology Classification for Country Comparisons system (WIPO TCCC) (Schmoch, 2008). For a description of how firms are classified as having patent grants within a technology class, see the Methodology page. An example of the type of analysis made possible by the BDS-PF tabulations is measuring the share of firms, across sectors, that were granted a computer patent (a sub-grouping of the WIPO TCCC Electrical Engineering sector, adding Computer Technology, IT Methods for Management, and Semiconductors). Figure 4 shows the change (indexed to 1978) in the percentage of patenting firms in each sector that have a computer patent grant. Unsurprisingly, the percentage of computer patents in the Information sector (51) saw the largest increase over the past four decades, more than tripling over this period. However, we also see a significant rise in the share of firms with a computer patent in Manufacturing (31-33), Wholesale (42), and the Professional, Scientific, and Technical Services (54) sectors, all more than doubling over this period. Interestingly, the Retail (44-45) sector saw a marked increase in computer patents starting in the late 1990s, shortly after the rapid diffusion of barcodes (Basker & Simcoe, 2021).
Figure 4. Rise of Firms with Computer Patents Across Sectors
Figure Notes: Figure shows the change in percentage of patenting firms in each sector with at least one recent patent grant that was classified as a computer patent (bds2022_pfg_comp_sec.csv, DMS number: P-7083300, DRB approval number: CBDRB‑FY25‑0097, CBDRB‑FY25‑0143). We first compute the percentage of patenting firms in each sector with a computer patent as the count of firms with a computer patent divided by the count of firms with a utility patent, by sector. We then index changes in that fraction by taking the ratio of the share of patenting firms with a computer patent in a given year divided by the share in 1978. By construction, this ratio is equal to 1 in 1978. A value of 3 on the y-axis, for example, means the share was three time higher than it was in 1978. The “Other (exl. 55)” line includes all other sectors not shown in the figure except the Management of Companies and Enterprises sector (55), as described in Figure 3.
These changes, though quite large, occur from very different initial levels of computer patenting across sectors. Figure 5 shows the percentage of patenting firms with a computer patent in the first year of the BDS-PF tabulations (1978) and the most recent year of the BDS-PF data (2022). In the Information sector (51), about 25.4% of firms with a patent had at least one computer patent in 1978, which rose to 78.3% in 2022. Interestingly, in 1978 the Information sector and the Education sector had about the same share of patenting firms with a computer patent, but computer patenting in the Education sector rose less dramatically, from 25.8% to 55.6%. Even in the Manufacturing sector (31-33), which had the lowest share of patenting firms with a computer patent in 1978 (6.5%), saw a significant increase in computer patenting, rising to 20.7% by 2022.
Figure 5. Rise of Firms with Computer Patents Across Sectors
Figure Notes: Figure shows the percent of patenting firms in each sector with at least one recent patent grant that was classified as a computer patent in 1978 and 2022 (bds2022_pfg_comp_sec.csv, DMS number: P-7083300, DRB approval number: CBDRB‑FY25‑0097, CBDRB‑FY25‑0143). The denominator includes all firms with a utility patent grant (bds2022_pfg_utly.csv). The “Other (exl. 55)” line includes all other sectors not shown in the figure except the Management of Companies and Enterprises sector (55), as described in Figure 3.
These figures demonstrate the types of analyses made possible by the BDS-PF tabulations. These data, for the first time, allow data users to observe the stock and flow of firms, employment, and establishments among patenting firms.
Download experimental Business Dynamics Statistics of Patenting Firms (BDS-Patenting Firms) data tables below. Variable definitions can be found in the BDS-Patenting Firms Definitions page.
Data Sources
The Business Dynamics Statistics of U.S. Patenting Firms (BDS-PF) is an experimental data product developed in collaboration with the USPTO. The BDS-PF data are compiled from the Longitudinal Business Database (LBD) and PatentsView, an open data platform supported by the USPTO. The LBD is a longitudinal database of business establishments and firms with coverage starting in 1976. There are ten patenting firm classifications: utility patent grants (pfg_utly), design patent grants (pfg_dsgn), all patent grants (pfg), chemistry patent grants (pfg_chem), communications patent grants (pfg_comm), computer patent grants (pfg_comp), instrument patent grants (pfg_inst), mechanical engineering patent grants (pfg_mech), other electronics patent grants (pfg_oelc), and other patent grants (pfg_othr). For a description of how firms are classified with patents of a given technology class, please see the Methodology page. For descriptions of firm and establishment classification see the BDS Codebook and Glossary.
The Census Bureau has reviewed this data product to ensure appropriate access, use, and disclosure avoidance protection of the confidential source data used to produce this product (Data Management System (DMS) number: P-7083300, Disclosure Review Board (DRB) approval number: CBDRB‑FY25‑0097, CBDRB‑FY25‑0143).
All BDS-PF CSV files are available for download in https://www2.census.gov/ces/bds/experimental/bds_pf/. For each of the 10 patenting firm classifications list above, there is one Economy-Wide dataset, 17 One-Way datasets, and 10 Two-Way datasets. Filenames are structured as follows:
Economy wide:
https://www2.census.gov/ces/bds/experimental/bds_pf/bds2022_pfg_[TYPE].csv
One-Way:
https://www2.census.gov/ces/bds/experimental/bds_pf/bds2022_pfg_[TYPE]_[ONE-WAY].csv
Two-Way:
https://www2.census.gov/ces/bds/experimental/bds_pf/bds2022_pfg_[TYPE]_[TWO-WAY].csv
Valid values of [TYPE] are:
- pfg_utly (Utility Patent Grants)
- pfg_dsgn (Design Patent Grants)
- pfg (All Patent Grants)
- pfg_chem (Chemistry Patent Grants)
- pfg_comm (Communications Patent Grants)
- pfg_comp (Computer Patent Grants)
- pfg_inst (Instrument Patent Grants)
- pfg_mec (Mechanical Engineering Patent Grants)
- pfg_oelc (Other Electronic Patent Grants)
- pfg_othr (Other Patent Grants)
Valid values of [ONE-WAY] are:
- fa (Firm Age)
- fac (Firm Age Coarse)
- fz (Firm Size)
- fzc (Firm Size Coarse)
- ifz (Initial Firm Size)
- ifzc (Initial Firm Size Coarse)
- ea (Establishment Age)
- eac (Establishment Age Coarse)
- ez (Establishment Size)
- ezc (Establishment Size Coarse)
- iez (Initial Establishment Size)
- iezc (Initial Establishment Size Coarse)
- st (State)
- met (Metro/Non-Metro)
- sec (Sector)
- vcn3 (3-digit NAICS)
- vcn4 (4-digit NAICS)
Valid values of [TWO-WAY] are:
- fzc_fac (Firm Size Coarse by Firm Age Coarse)
- sec_fac (Sector by Firm Age Coarse)
- sec_fzc (Sector by Firm Size Coarse)
- sec_ifzc (Sector by Initial Firm Size Coarse)
- sec_eac (Sector by Establishment Age Coarse)
- sec_ezc (Sector by Establishment Size Coarse)
- sec_iezc (Sector by Initial Establishment Size Coarse)
- st_sec (State by Sector)
- st_fa (State by Firm Age)
- st_fz (State by Firm Size)
As an example, the full set of 28 Utility Patent Grants files are listed below. Other patenting firm classification files can be found at https://www2.census.gov/ces/bds/experimental/bds_pf/.
All Utility Patent Grants (pfg_utly) Datasets
Economy-Wide Datasets
One-Way Datasets
Two-Way Datasets
pfg_utly – Firm utility patent grant indicator. Utility patents are issued for the invention of a new and useful process, machine, manufacture, or composition of matter, or a new and useful improvement thereof. For our purposes, only utility patents are counted in the technology class tables. For a description of how firms are classified as having a patent grant, please see the Methodology page.
pfg_utly |
Description |
No Utility Patent Granted |
Firm has no utility patent granted in t, t-1, or t-2. |
Utility Patent Granted |
Firm has a utility patent granted in t, t-1, or t-2. |
pfg_dsgn – Firm design patent grant indicator. Design patents are issued for a new, original, and ornamental design embodied in or applied to an article of manufacture.
pfg_dsgn |
Description |
No Design Patent Granted |
Firm has no design patent granted in t, t-1, or t-2. |
Design Patent Granted |
Firm has a design patent granted in t, t-1, or t-2. |
pfg – Firm patent grant indicator. Includes utility, design, and plant patents.
pfg |
Description |
No Patent Granted |
Firm has no patent granted in t, t-1, or t-2. |
Patent Granted |
Firm has a patent granted in t, t-1, or t-2. |
pfg_chem– Firm chemistry patent grant indicator. For our purposes, only utility patents are classified into technology classes. For a description of how firms are classified with patents of a given technology class, please see the Methodology page.
pfg_chem |
Description |
No Chemistry Patent Granted |
Firm has no chemistry patent granted in t, t-1, or t-2. |
Chemistry Patent Granted |
Firm has a chemistry patent granted in t, t-1, or t-2. |
pfg_comm – Firm communications patent grant indicator.
pfg_comm |
Description |
No Communications Patent Granted |
Firm has no communications patent granted in t, t-1, or t-2. |
Communications Patent Granted |
Firm has a communications patent granted in t, t-1, or t-2. |
pfg_comp – Firm computer patent grant indicator.
pfg_comp |
Description |
No Computer Patent Granted |
Firm has no computer patent granted in t, t-1, or t-2. |
Computer Patent Granted |
Firm has a computer patent granted in t, t-1, or t-2. |
pfg_inst – Firm instrument patent grant indicator.
pfg_inst |
Description |
No Instrument Patent Granted |
Firm has no instrument patent granted in t, t-1, or t-2. |
Instrument Patent Granted |
Firm has an instrument patent granted in t, t-1, or t-2. |
pfg_mech – Firm mechanical engineering patent grant indicator.
pfg_mech |
Description |
No Mechanical Patent Granted |
Firm has no mechanical engineering patent granted in t, t-1, or t-2. |
Mechanical Patent Granted |
Firm has a mechanical engineering patent granted in t, t-1, or t-2. |
pfg_oelc – Firm other electronics patent grant indicator.
pfg_oelc |
Description |
No Electronics Patent Granted |
Firm has no other electronics patent granted in t, t-1, or t-2. |
Electronics Patent Granted |
Firm has an other electronics patent granted in t, t-1, or t-2. |
pfg_othr – Firm other patent grant indicator. Incudes patents not classified as chemistry, communications, computers, instruments, mechanical engineering, or other electronics.
pfg_othr |
Description |
No Other Patent Granted |
Firm has no other patent granted in t, t-1, or t-2. |
Other Patent Granted |
Firm has an other patent granted in t, t-1, or t-2. |
For all other variable definitions refer to the BDS Codebook and Glossary.
For a complete description of how the BDS data measures the stock and flow of establishments, firms, and employment, see the main BDS Methodology page.
The BDS-PF experimental data product classifies firms based upon a match of patent documents issued by the USPTO to the Census Bureau’s universe of non-farm employer businesses in the Business Register (BR). The BR is the source data for the Longitudinal Business Database and in turn the BDS. The USPTO’s PatentsView database provides information on the name and location of assignees and inventors, along with the dates of application and grant for each patent. Patents are matched at the firm-level. Exact and fuzzy name and geography matching was used to link patent assignees to firms in the BR. Matches were made more accurate using the disambiguated assignee identifiers found in the USPTO’s PatentsView data. PatentsView assignee identifiers link together the same business across different patent documents. Those identifiers were used to evaluate the longitudinal consistency of firm matches over time for a given assignee. Ultimately, we match roughly 92% of U.S. patent assignees and 58% of foreign assignees to firms in the Census data with a match precision of nearly 94% for U.S. patent assignees and 81% for foreign assignees. For additional details about the matching methodology see Fort et al. (2025).
Care is required when interpreting tabulations that combine establishment and firm-level characteristics. Fundamentally, the BDS tabulations are establishment-based statistics. Some characteristics, like geography and industry, are observed at the establishment-level. Others, like age and size, are measured at both the establishment- and firm-level. BDS tabulations that use firm characteristics, such as firm size, classify establishments by their associated firm’s characteristics. Patenting classifications in the BDS-PF data follow this approach, classifying establishments based upon their associated firm’s patenting activity. If a large multi-establishment firm operating in multiple states or sectors is classified as having a recent patent grant, it will contribute to the count of patenting firms in each geography and sector in which it operates, regardless of whether the patent is related to only some of those sectors or geographies. For example, if a firm has establishments in both manufacturing and finance and receives a patent grant associated with its manufacturing activities, it will nonetheless contribute to the count of patenting firms in the finance sector.
The BDS-PF experimental data product is based on patents granted since 1976, and distinguishes between different types of patents granted to each firm. In the pfg_utly tables (firm utility patent grant indicator), firms are classified based upon whether they were granted a utility patent in any of the years t, t-1, or t-2. Utility patents are issued for the invention of a new and useful process, machine, manufacture, or composition of matter, or a new and useful improvement thereof (see the USPTO documentation for details). The pfg_dsgn tables (firm design patent grant indicator) classifies firms based upon whether they were granted a recent design patent in any of the years t, t-1, or t-2. Design patent grants are issued for new, original, and ornamental designs embodied in or applied to articles of manufacture. Functional aspects of an article of manufacture are not eligible for a design patent. The overall patenting variable (pfg) combines utility, design, and plant patents. This category excludes all other types patent grants such as reissues.
A firm is classified as “patenting” if it is matched to a patent granted in years t, t-1, or t-2. The timing of this classification relies on grant year rather than application year. Conceptually, the application year is closer in time to the research and development activity that led to the invention, whereas the grant year captures when the firm received intellectual property rights over the invention. Once a firm receives a patent grant, it will continue to be considered a “patenting firm” for an additional two years after the year in which the patent was granted. Identifying a firm as patenting based upon previous-year patent grants increases sample sizes, allowing for more granular cross tabulations, and captures periods in which the firm is likely to be actively commercializing innovations. When firms receive patent grants in multiple years, they tend to be relatively close in time. According to PatentsView data, 64% of assignees with patents granted in more than one year receive their next patent within three years (the average gap among these assignees between grants is 2.9 years).
The BDS-PF tabulations also distinguish patent grants based on technology class. Each utility patent is assigned to one or more technology classes based upon WIPO TCCC (Schmoch, 2008). This system is built for high-level comparisons using the International Patent Classification (IPC) technology field codes, with the aim of splitting the world-wide patent corpus relatively evenly among categories. The USPTO classifies utility patents using the Cooperative Patent Classification (CPC) system, which was designed to be more detailed than the IPC. The PatentsVeiw database contains each patent’s current CPC classes (patents are reclassified into new technology codes as the CPC scheme develops and changes). Patents may be assigned as many CPC technology codes as needed to describe the invention, as inventions often combine and relate to several technologies. Each patent is assigned WIPO TCCC using the PatentsView current inventional CPC codes, mapped to IPC codes using the IPC-CPC crosswalk.[1] Each patent, therefore, may be classified in more than one WIPO technology class.
The classifications used in this data product are based on the June 2024 WIPO TCCC release,[2] and include both sector and field to more evenly balance the number of patents in each group. The Electrical Engineering sector is split into three groups: Communications (Telecommunications, Digital Communication, and Basic Communication Processes), Computers (Computer technology, IT Methods for Management, and Semiconductors), and Other Electronics (Electrical machinery, Apparatus, Energy and Audio-visual Technology). All other sectors remain the same. About 75% of patents belong to one WIPO technology class as defined above, 21% to two, and the rest belong to 3 or more.
Firms are classified as having a patent in a given WIPO technology class if any of the patents matched to that firm have a current inventional CPC code in that class.[3] Thus, a utility patenting firm may be classified in several (or all) WIPO technology classes. A firm may be associated with more than one WIPO technology class because a single patent is associated with more than one technology class, or because the firm holds multiple patents that are associated with different technology classes.
The BDS-PF experimental product includes ten firm-level patent classifications. The first three classify firms based upon utility patent grant (pfg_utly), design patent grant (pfg_dsgn), or any patent grant (pfg). The other seven classify firms based upon utility patent grants with specific technology classifications. Each are listed below. For additional information about the values of each variable, see the Definitions page.
- pfg_util classifies firms based upon whether they received a utility patent grant in t, t-1, or t-2.
- pfg_dsgn classifies firms based upon whether they received a design patent grant in t, t-1, or t-2.
- pfg classifies firms based upon whether they received a patent grant (utility, design, or Plant) in t, t-1, or t-2.
- pfg_chem classifies firms based upon whether they received a chemistry patent grant in t, t-1, or t-2.
- pfg_comm classifies firms based upon whether they received a communications patent grant in t, t-1, or t-2.
- pfg_comp classifies firms based upon whether they received a computer patent grant in t, t-1, or t-2.
- pfg_inst classifies firms based upon whether they received an instrument patent grant in t, t-1, or t-2.
- pfg_mech classifies firms based upon whether they received a mechanical engineering patent grant in t, t-1, or t-2.
- pfg_oelc classifies firms based upon whether they received an other electronics patent grant in t, t-1, or t-2.
- pfg_othr classifies firms based upon whether they received an other patent grant in t, t-1, or t-2.
Footnotes
- If the IPC-CPC crosswalk does not have a mapping for a CPC code, the CPC sub-class (4 digits, representing the third level of division) is used as the IPC class for mapping to WIPO TCCC. If no inventional codes are listed, the first code listed, and only the first code, is used. If PatentsView does not list a current CPC code, then the IPC inventional codes at issue are used (unless no inventional code is listed and then the first and only the first listed code is used).
- The June 2024 WIPO TCCC release can be found at https://www.wipo.int/en/web/ip-statistics under the heading “Other Intellectual Property Data resources” and sub-heading “IPC Concordance Table” as well as at https://www.wipo.int/documents/2948119/3215563/ipc_technology.xlsx
- For more details on difference between inventional and additional classifications, please see the USPTO Manual of Patent Examining Procedure (MPEP) section 905.03(a) “The CPC Database” [R-07.2022]
Below are links to selected publications related to the BDS-Patenting Firms.
Basker, E., & Simcoe, T. (2021). Upstream, downstream: Diffusion and impacts of the universal product code. Journal of Political Economy, 129 (4), 1252–1286.
Fort, T. C., Goldschlag, N., Liang, J., Schott, P. K., & Zolas, N. (2025). Growth is Getting Harder to Find, Not Ideas. CES Discussion Paper Series (CES-WP-25-21), Center for Economic Studies, U.S. Census Bureau.
Goldschlag, N., & Perlman, E. (2017). Business Dynamic Statistics of Innovative Firms. CES Discussion Paper Series (CES-WP-17-72). Center for Economic Studies, U.S. Census Bureau.
Schmoch, U. (2008). Concept of a technology classification for country comparisons. Final report to the World Intellectual Property Organization (WIPO) . World Intellectual Property Organization.
Questions? Contact us at ces.bds@census.gov