The BDS-HT experimental data product classifies industries as High Tech using the concentration of Science, Technology, Engineering, and Math (STEM) occupation employment as described in Goldschlag & Miranda (2020) (henceforth GM2020) and Hecker (2005). Here we briefly describe the details of how that methodology was applied to create the BDS-HT data products. We apply the STEM method to eleven years of industry-occupation data covering the 2007 Economic Census to the 2017 Economic Census. In each year, we identify 2017 NAICS vintage industries with a STEM employment share at least five times the national average STEM employment share across all industries. We classify industries as High Tech for the BDS-HT experimental data products if they meet this threshold in at least six of the eleven years. The set of industries classified as High Tech for the BDS-HT products are shown in Table 1.

Table 1: BDS-HT Industry List

There are several reasons for using the industry occupation data in this way. Having the window of years tied to Economic Census years, which occur every five years, facilitates future updates to the industry listing as NAICS vintages change. NAICS industry code vintages change every five years with each new Economic Census. When NAICS vintages change, we will shift the eleven-year window by five years ending with the latest Economic Census year. Having the window of years cover two inter-Economic Census periods will make changes to the industry list more gradual. Finally, using eleven years of data near the end of the time series simplifies the interpretation of the statistics. The BDS-HT data products will capture the current and historic business dynamics patterns of industries that were classified as High Tech in a decade near the end of the BDS data. This approach also limits the likelihood of the classification capturing industries that were High Tech many years ago but are no longer High Tech.

There are several notable differences between the industry list used for the BDS-HT data products and that described in GM2020. Despite the minor differences in industry listings, the patterns in business dynamism for High Tech industries described in GM2020 also appear in the BDS-HT experimental data product. For example, the rise of job creation rates in the late 1990s and the surge of firm entry through the mid and late 1990s are evident in both the BDS-HT experimental data tables and GM2020.

The differences between the industry listing presented in GM2020 and that used for the BDS-HT data products occur for several reasons. First, GM2020 used the union of industries classified as High Tech via STEM concentration in 2005, 2012, and 2014. The BDS-HT experimental product definition uses a majority rule with different and additional years of data. Second, GM2020 used 2007 NAICS industry codes and 2010 SOC occupation codes. The splitting and collapsing of industries between NAICS vintages between the 2007 and 2017 NAICS codes can change the relative STEM concentration across industries.

Nearly all industries classified as High Tech in GM2020 are also identified as High Tech in the BDS-HT experimental data product. The four exceptions are “Oil and Gas Extraction” (NAICS 2111), “Pharmaceutical and Medicine Manufacturing” (NAICS 3254), “Wired Telecommunications Carriers” (NAICS 5171), and “Other Telecommunications” (NAICS 5179).  Industry 2111 only met the STEM threshold in 2012, one year of our 11-year window and therefore, is not included in our list of High Tech industries. NAICS industry 3254 met the criteria in five of the eleven years and NAICS 5179 only met the criteria in one year. Starting in 2017, BLS aggregated 5173, 5174, and 5179 into one four-digit industry: 5170. This does not affect whether these industries are classified as High Tech. Finally, 5179 met the STEM threshold in the 2000s, but not after that.

NAICS industry 5171 is a unique case among the set of industries that were previously classified as High Tech in GM2020 but are not considered High Tech in the BDS-HT product. NAICS 5171 experienced complex industry classification changes across NAICS vintages. This industry was derived from a 2002 NAICS code that split into 3 different 2007 NAICS codes (5171, 5179, and 5191). In the 2017 NAICS vintage, 5171 is combined with another industry (5172) to form a new four-digit NAICS industry: 5173. Therefore, 5171 does not exist in the 2017 NAICS vintage. Moreover, the STEM intensity of 5173 will depend on the combined composition of the prior vintage 5171 and 5172 components.

Additionally, the BDS-HT follows the BDS Methodology to calculate the various statistics reported in the tables.

References

Goldschlag, Nathan, & Miranda, Javier. (2020). Business Dynamics Statistics of High Tech Industries. Journal of Economics & Management Strategy, 29(1), 3-30.

Hecker, Daniel E. (2005). High-Technology Employment: a NAICS-Based Update. Monthly Lab. Rev., 128, 57.

Questions?  Contact us at ces.bds@census.gov.