Interactive research figures
Explore the evidence.
Move through industries, years, and published findings. These figures use public data or estimates reported in the papers.
No confidential microdata or reconstructed individual records.
Figure 01 · Digital capabilities
How IT-intensive is each industry?
Choose an industry to trace the share of job postings mapped to computer and mathematical occupations (SOC 15) from 2010–2023, alongside the median industry.
Share of job postings mapped to SOC 15 occupations. Hover or focus a point for its value.
Public dataset ↗Figure 02 · Technology & work
Digital intensity meets remote-work feasibility.
Each circle is an industry group. Move the year slider to see how its computer and mathematical occupation share changes; circle size indicates total job postings.
Horizontal position is the public WFH-feasibility index; vertical position is the SOC 15 job-posting share. This figure uses the archived 2010–2020 public file.
Archived data ↗Figure 03 · Skill demand
Where are cloud and cyber skills most concentrated?
Rank industries by the share of observed skill clusters across job postings that are cloud- or cybersecurity-related. Toggle the measure and move through 2010–2023.
The chart shows the ten highest of 24 NAICS series in the selected year. Values are percentages of observed skill clusters across job postings.
Variable definitions ↗Figure 04 · Open AI ecosystems
A policy shock, two repository ecosystems.
The pooled shock-window estimates show a much larger increase in weekly forks for China-associated repositories.
Reported pooled estimates, shown as reader-friendly approximations; the figure preserves the paper’s repository-week unit and does not imply equal baseline activity.
Working paper ↗Figure 05 · Finding explorer
One estimate, in its own unit.
Select a study to explore a headline result. Unlike units are intentionally never placed on the same comparative axis.
What these figures show.
Figures 1 and 3 use the current public 2010–2023 industry file released with research on IT capabilities; Figure 2 preserves the archived 2010–2020 file because it includes the WFH-feasibility measure. Figure 4 and the finding explorer use values reported in the linked papers or publisher pages.
They are descriptive views of published evidence—not substitutes for the papers’ research designs, uncertainty estimates, or identification assumptions. Confidential Census and firm microdata are not exposed.
Return to research →