When a steel mill closes, the governor shows up. There are cameras. There is a promise of retraining funds. There is a community college that will offer welding certificates and coding bootcamps. There is a press release. And then, about eighteen months later, there is a study showing that most of the displaced workers didn't complete the program, and the ones who did are earning 30% less than they were before.
This cycle has repeated itself so many times that it has become a ritual. The plant closes. The politicians arrive. The retraining is announced. The workers are not retrained. Nobody is surprised. The ritual continues.
The evidence on workforce retraining programs is remarkably consistent and remarkably ignored. A 2022 meta-analysis of 37 retraining programs found that the average participant earned $900 less per year after completing the program than they would have earned without it, after accounting for the opportunity cost of the training period. The programs that showed positive returns were narrow, employer-specific, and short — essentially on-the-job training with a different name.
The programs that don't work — the ones that dominate the policy conversation — are the broad ones. The ones that promise to transform a 52-year-old machinist into a software developer, or a 48-year-old retail manager into a data analyst. These programs fail not because the people are incapable of learning new skills, but because they are solving the wrong problem.
The problem is not that experienced workers lack skills. The problem is that the labor market has a systematic bias against hiring experienced workers regardless of their skills. You can retrain a 55-year-old for any role you want. If the hiring system filters them out before a human sees their application, the training is irrelevant.
Retraining programs serve a political function, not an economic one. They allow elected officials to respond to job displacement with something that sounds like action. They allow companies to demonstrate corporate responsibility without changing their hiring practices. They allow the workforce development industry to generate revenue. They do not, in any systematic way, solve the problem of experienced workers being excluded from labor market participation.
This is not a cynical observation. It is a structural one. Retraining programs are designed to change workers. The problem is not the workers. The problem is the infrastructure that processes workers — the hiring systems, the screening algorithms, the credential requirements, the age biases embedded in every stage of the hiring process. Changing the workers doesn't fix the infrastructure.
The interventions that show consistent positive results for displaced experienced workers are not retraining programs. They are infrastructure interventions: direct placement services that bypass the broken hiring pipeline, credentialing systems that make existing competency legible to employers who don't know how to evaluate it, and economic pathways that don't require navigating the traditional hiring process at all.
The most effective thing you can do for a 52-year-old with twenty years of supply chain experience who can't get past the ATS filter is not to teach them Python. It is to build the infrastructure that gets their actual capabilities in front of the people who need them, in a format those people can evaluate, through a process that doesn't filter them out before the conversation starts.
That infrastructure doesn't exist at scale. Building it is the work. Retraining is the comfortable story we tell while we avoid doing it.
