The United States spends roughly 0.1% of GDP on active labor-market policies — training, job-search assistance, wage subsidies, employment services — against an OECD median of about 0.5% and a Danish benchmark above 2%. That ten-to-one gap is the single most defensible explanation for why displaced American workers experience longer unemployment spells, larger wage scarring, and worse long-run earnings recovery than displaced workers in nearly every peer economy. The labor economics literature on this is unusually consistent: Daron Acemoglu and David Autor's work on labor-market polarization, the OECD's Back to Work series, and Till von Wachter's research on the persistent earnings losses of displaced workers (American Economic Journal and Journal of Economic Perspectives) all point at the same conclusion. Spending on ALMPs is one of the highest-return interventions a wealthy government can make, and the U.S. has been chronically underspending for forty years.
So the argument here is direct: the central workforce-policy reform agenda for the next decade is not new programs. It is roughly tripling ALMP spending, restructuring it around three program families that have empirical track records, and modernizing unemployment insurance into a labor-market-transition system rather than a partial wage-replacement system. Everything else — short-term credentials, individual training accounts, generic career counseling — is a marginal optimization on top of that base case.
The case for tripling ALMP spending
The benefit-cost evidence on the better ALMP designs is strong enough that the question is not whether to spend more but how to spend it well. The OECD's repeated synthesis evaluations show that sectoral training programs with employer co-investment, short-duration job-search assistance with case-management support, and well-designed wage subsidies all routinely return more than a dollar in increased earnings and tax revenue per dollar spent, often substantially more. Project Quest in San Antonio, evaluated by Economic Mobility Corporation over nine years, produced sustained annual earnings impacts of roughly $5,000 per participant. The Year Up MDRC randomized trial, published in 2019, found earnings gains of around 30% three years post-program. Apprenticeship returns documented by Mathematica and Robert Lerman's research at Urban Institute consistently show wage premiums of 30–40% over comparable non-apprentice workers, with most of the cost borne by the employer.
The U.S. policy machinery has produced none of those things at scale. WIOA Adult and Dislocated Worker programs serve roughly 0.5–1% of the unemployed in any given year, according to ETA's reporting. Trade Adjustment Assistance, before its 2022 expiration, served fewer than 100,000 workers a year against millions of trade-displaced workers in the prior two decades. The Job Corps serves about 50,000 young people. These are rounding errors against the labor force.
The three program families that should anchor the rebuild
Registered apprenticeship at European scale
Switzerland enrolls roughly 70% of upper-secondary students in dual-track apprenticeship programs. Germany enrolls roughly 50%. The U.S. has approximately 650,000 registered apprentices in any given year against a labor force of 160 million — a participation rate two orders of magnitude lower than those benchmarks. The Department of Labor's Apprenticeship USA initiative, the Biden-era Apprenticeship Building America grants, and the Industry Recognized Apprenticeship Program debates have all been small relative to the gap. A serious federal agenda would set a five-year target of 3 million registered apprentices, with public co-investment that brings employer net costs to parity with hiring inexperienced workers off the open market. For a deeper treatment, see our flagship Apprenticeship 2.0 →.
Sectoral partnerships with binding employer commitments
The sectoral-partnership model — long-duration training in a specific industry, with hiring commitments from named employers — has the best evidence base in U.S. workforce policy. Per Scholas in IT, Project Quest in healthcare, JVS Boston in advanced manufacturing — these aren't experimental anymore. The federal Strengthening Community Colleges Training Grants and the state-level National Governors Association sectoral pilots have produced replicable templates. The barrier is not knowledge. It is funding scale.
Wage subsidies for hard-to-place workers
The Work Opportunity Tax Credit, in its current form, is small and poorly targeted. The U.K.'s Restart Scheme, the German Eingliederungszuschuss, and the Danish flex-job program all use direct wage subsidies more aggressively for workers with long unemployment spells, disabilities, or other persistent barriers. The U.S. literature on wage subsidies — including work by Lawrence Katz, Jeffrey Smith, and others — supports tighter targeting and larger per-worker subsidies for workers who would otherwise face very long unemployment durations.
UI as a labor-market-transition system, not just wage replacement
The U.S. unemployment-insurance system is the most embarrassing part of the workforce-policy stack. State-administered, severely undermodernized, and replacement-rate volatile, it pays an average of roughly 38% of prior wages — below the OECD median, and well below the 70%+ replacement rates in flexicurity systems like Denmark's. The COVID expansion temporarily papered over this with Federal Pandemic Unemployment Compensation, but the structural system reverted.
The reform agenda that matters is more than benefit-level expansion. It includes:
National minimum standards for benefit duration and replacement rate, conditional on participation in active job-search and training (the Danish "rights and duties" model). Universal coverage of independent contractors and gig workers through portable benefits programs — Washington State's 2023 portable-benefits law for gig workers is the most replicable U.S. template. A modernized state UI IT stack — the Department of Labor's 2024 funding round began this, but the technical debt across 53 state systems is staggering. Wage-insurance pilots for displaced older workers, an idea Lawrence Summers and others have proposed for two decades that has never been seriously funded.
The labor-market-regulation question
Labor-market regulation in the U.S. operates on two contradictory premises. At the federal level, the National Labor Relations Act framework has been static since 1947 (excluding the brief 2024 NLRB activity now in legal uncertainty), and the employee/independent-contractor distinction has been litigated rather than legislated. At the state level, California's AB 5 and the resulting Castellanos v. State of California (2024) decision, Massachusetts' similar 2023 ballot measure, and Washington State's portable-benefits law have produced three incompatible models for gig-worker classification. The result is the worst of both worlds: substantial regulation, inconsistently applied, and no portable benefits to compensate for the precarity it does not eliminate.
The EU's Platform Work Directive, adopted in 2024 with a 2026 transposition deadline, creates a rebuttable presumption of employment for platform workers — a cleaner default rule than the U.S. patchwork. Even a partial U.S. adoption of a presumption-of-employment framework, paired with a portable-benefits financing mechanism, would simplify a regulatory landscape that currently costs both workers and employers in compliance overhead. For the full treatment of this question, see our flagship The Gig Economy Settlement →.
The political economy
The reason the U.S. underspends on ALMPs is not that the evidence is weak. The evidence is strong. The reason is that ALMP spending is structurally pro-labor in a system where pro-labor spending requires assembling specific congressional coalitions. The CHIPS and Science Act, the Bipartisan Infrastructure Law, and the IRA all included meaningful workforce provisions, and they were politically possible because they were braided into industrial policy. That is the political model. Future expansions of ALMP spending should be braided into climate, AI, and reshoring legislation, where the workforce component has a national-competitiveness frame and the constituencies extend beyond traditional labor.
The U.S. spends one-tenth what peer economies spend on active labor-market policy, and then expresses surprise when displaced workers take a decade to recover. The math is not subtle.
The workforce-policy debate has been stuck for twenty years in a fight over the right adjective in front of the word "training." Targeted training. Skills-based training. Demand-driven training. The adjective is not the constraint. The level of spending and the structural integration with industrial policy are the constraints. A serious reform agenda triples the budget, channels it through apprenticeship and sectoral partnerships, modernizes UI into a transition system, and federalizes a presumption-of-employment rule for platform work. That is what would actually change the labor-market outcomes the prior twenty years of incremental reform have failed to move.
Updated May 21, 2026. This piece was substantively rewritten as part of NWLB's 2026 editorial refresh.
