Table 3. FTE-to-population growth spread with Yale AI controls

Ages 22–26 minus ages 30–49 · 2023 baseline · Separate specification exercise

Coefficients with robust standard errors in parentheses. Significance stars are reproduced from the draft; occupation-clustered standard errors and p-values for the focal exposure appear in the final two rows.
Young: nondisabled, non-enrolled ages 22–26 with exactly a bachelor’s degree. Adults: nondisabled ages 30–49, all education. Payroll jobs only; main-job FTE = weekly hours / 40 (actual hours if usual hours vary). October is excluded from both comparison periods; survey weights are adjusted to pre-2026 population controls and fixed single-age shares.
WFH = any telework in the latter sample period. DN = Dingel–Neiman pre-pandemic WFH feasibility; PCA = principal component analysis; BTOS = Business Trends and Outlook Survey. AI adoption pools January–June 2026. pp = percentage points; EPOP = employment-to-population ratio; FTE = full-time equivalent.
Source(s): Current Population Survey, The Budget Lab, Census Business Trends and Outlook Survey, Dingel and Neiman (2020), Stripe analysisGet the dataDownload image