Review of Worker Retraining Programs Shows Modest Returns for AI Labor Disruption

Review of Worker Retraining Programs Shows Modest Returns for AI Labor Disruption A review coauthored by independent researcher David Roodman and Anthropic's Maxim Massenkoff examines the effectiveness of worker retraining programs as a policy response to potential labor market disruption from AI. A meta-analysis of 56 randomized US studies alongside European evidence shows retraining increases employment by two to three percentage points and earnings by roughly $1,000 per year. Retraining costs approximately $13,000 per slot offered, with governments recouping over half the expenditure via increased tax revenue and reduced benefit payouts. Sector programs tied directly to high-demand employers produce significantly larger gains, but attempts to replicate them often fail. The authors conclude existing programs would likely be insufficient for large-scale AI displacement and recommend testing and scaling promising models now. Meta-Analysis of Job Training Outcomes Drawing on 56 randomized US studies combined in a new meta-analysis and experimental data from Europe, the review finds that standard job training programs yield positive but modest results. Offering a training slot increases employment by two to three percentage points and annual earnings by around $1,000, set against an average program cost of roughly $13,000. Taking into account increased tax revenues and lower benefit outlays, governments recoup more than half of their spending, allowing programs to roughly break even overall. Nevertheless, the authors determine that these existing models would likely fall short if AI triggers large-scale worker displacement. Sector Programs and Recommendations The review highlights that a small category of sector programs—which collaborate with employers in high-demand industries to place participants straight into open roles—achieve gains several times higher than conventional training. However, efforts to replicate these sector models have frequently failed. Consequently, the authors' primary recommendation is to invest immediately in testing, evaluating, and scaling the most viable initiatives, including rapidly expanding a leading program for a targeted worker demographic under rigorous measurement.

Anthropic与独立学者回顾员工再培训项目实据：成效温和且难以应对大规模AI冲击 独立研究员与Anthropic联合发布对员工再培训项目的实据审查，通过对数十项实验研究的元分析发现，现有培训项目虽有积极效果但提升幅度温和，若AI引发大规模失业，现有项目可能难以应对。 研究基于56项美国随机对照实验及欧洲数据开展元分析，评估职业再培训应对AI冲击的可行性。 普通培训项目平均带来2至3个百分点的就业率提升和每年约1000美元的薪资增长，成本约13000美元，整体大致收支平衡。 与高需求行业雇主直接对接的“行业项目”收益更高，但复制难度极大，难以直接规模化推广。 研究建议立即对最具前景的项目进行论证、评估与规模化试验，以备未来劳动力市场剧变。 常规培训项目效果温和且成本高昂 基于56项美国随机实验及欧洲实验证据的元分析显示，职业再培训平均能产生积极但较为温和的效果。每提供一个培训名额，受训者的就业率提高2到3个百分点，年收入增加约1000美元，而人均培训成本约为13000美元。计入税收增加和福利支出减少后，政府可收回一半以上的开支，项目整体大致打平。 高成效行业项目复制困难，难以应对大规模置换 少部分与高需求行业雇主合作并直接安置就业的“行业项目”所带来的收益高出数倍，但多次复制这些项目的尝试往往以失败告终。作者据此指出，如果人工智能造成大规模劳动力置换，现有的再培训体系很可能力不从心。 政策建议：及早测试与拓展优质项目 研究的核心建议是立即展开投资，对最有前景的项目进行展示、评估与规模化推广，包括针对特定工人群体快速扩张标杆项目并严格衡量成效，为未来的政策应对提供可靠依据。

We're sharing a review of the evidence on worker retraining programs, coauthored by independent researcher David Roodman and Anthropic's Maxim Massenkoff. Retraining workers is the most popular policy option for mitigating labor market disruption from AI. In this report, the authors investigate whether these programs would work in the face of significant labor market disruption. The review is part of our Economic Research team's work on AI's effects on the economy. Our [Economic Index](https://www.anthropic.com/economic-index) tracks how AI is being used across occupations and industries. Earlier this year, we published a [framework](https://www.anthropic.com/research/labor-market-impacts) for measuring AI's effects on the labor market and identifying the jobs most likely to be affected. Our [Economic Policy Framework](https://www.anthropic.com/policy-on-the-ai-exponential/epf) sets out possible policy responses, including worker retraining, across a range of scenarios. We want to understand the evidence behind each of these policy responses. The review draws on 56 randomized US studies, combined in a new meta-analysis, along with experimental evidence from Europe. On average, job training programs produce positive but modest effects: for each person offered a training slot, employment rises by two to three percentage points and earnings by roughly $1,000 a year, against a cost of about $13,000. Counting the added tax revenue and reduced benefit payments, the government recovers more than half of what it spends, and programs roughly break even overall. A small set of “sector programs”—programs that partner with employers in a high-demand industry and place people directly into jobs in that field—produce gains several times larger. But attempts to replicate them have often failed. The authors conclude that if AI displaces workers at scale, existing retraining programs would likely fall short. Their central recommendation is to invest now in demonstrating, evaluating, and scaling the most promising programs, including through an effort to rapidly expand a leading program for a specific group of workers and rigorously measure the results. Our [Economic Futures Research Fund](https://www.anthropic.com/news/economic-futures-research-fund-agenda) is designed to fund investigations into questions like these.