01
Migration
The only lever that moves on policy time: immigrants arrive already trained — 42% of home-health aides, ~28% of physicians — filling shortage roles in 1–3 years, vs. 2–4 to train a nurse or ~20 for a birth cohort.
Beta
First-pass policy explorer. Figures are pinned to the projection cube at a fixed scenario; switch axes on the Atlas to test alternatives.
Policy · Beta
Policy happens somewhere. Pick the place.
The national gap is real, but no one governs the nation's labor market in the abstract. Choose a scale — the country, a state, a city — and the map and the read-out below shift to that place: where its shortage sits, where AI displacement lands, and what migration and reskilling actually buy there.
Shaded by projected 2034 shortage · click a state, then a metro, to drill in
The United States
Worker shortage · 2034
—
transformative AI · 10% wage cap
AI-displaced · 2034
—
transformative AI · AI-attributable · rigid floor
Where the shortage is
Where the displacement is
Migration — the fastest lever, and AI rewrites it
—
vs. baseline immigration 2034 shortage AI-displaced
Lower (−50%) — —
No new immigration — —
AI inverts the immigration math. Today, immigrants are the fastest fill for the shortage roles — 42% of home-health aides, ~28% of physicians. But once a transformative AI shock has hollowed out demand, trimming immigration eases both the residual shortage and the displacement — at the cost of the workers, and the local demand their spending creates, that immigrants bring. NAS 2017; Hong & McLaren 2015.
Reskilling — real, but slower and only partial
—
Nationally, of the 10.6M workers a transformative AI shock displaces, about 43% have openings in their own occupation elsewhere, ~10% can move to a nearby field, and 47% would need deep retraining into a different occupation.
Sector-specific, employer-linked programs lift earnings 13–40% (Project QUEST, Per Scholas); generic public training shows no measurable effect (WIA). Reskilling takes months to years, is credential-gated for healthcare, and workers move within skill clusters ~3.8× more easily than across them. Katz et al. 2022; Escobari et al. 2021.
The toolkit
Five levers that narrow the gap.
02
Wage flexibility
Let prices clear: lifting the wage-growth cap absorbs much of the gap, but the bill shows up as consumer-price pass-through (see Trade-off).
03
Reskilling
Move displaced office, admin, and sales workers toward shortage clusters — but only ~3% of the displaced are a CPS-feasible jump away, and the highest-shortage healthcare roles are credential-gated. Works when targeted; null when generic.
04
Late-career retention
Healthcare and education carry the deepest shortages and the oldest workforces; phased retirement and licensure portability buy years per worker.
05
Targeted automation
Substitute capital where labor is hardest to find — transport, material moving, food prep — to pull the headline number off the most-binding sectors first.
Projections: Bahar & Wright (2026), BW-LaborShortages Composite Spatial Equilibrium. Immigration ladder and AI-displacement come straight from the model cube (baseline / −50% / no-new immigration). Evidence on lever magnitudes: National Academies of Sciences (2017), The Economic and Fiscal Consequences of Immigration; Hong & McLaren (2015), NBER w21123 (immigrants’ local-demand channel); BLS & MPI (2024) on foreign-born occupation shares; Katz, Roth, Hendra & Schaberg (2022), NBER w28248 (sectoral-training RCTs); Roder & Elliott / Economic Mobility Corp. (Project QUEST); Fortson et al. (2017), DOL WIA Gold-Standard evaluation (null generic training); Escobari, Seyal & Daboin Contreras (2021), Moving Up, Brookings (skill-cluster mobility). Reskilling-absorbability split is national (transformative-AI scenario); per-place reskilling routes are a planned refinement.