Latest / The Fiscal Policy Podcast with Fexingo: Government Budgets, Stimulus, and Public Spending / How Subsidized Job Training Fails Workers and Taxpayers
Transcript
- Lucas: So the US government spends roughly four billion dollars a year on job training for low-income workers. And according to the most recent GAO report, from early 2025, those programs place about seventy percent of participants into jobs. Luna: Seventy percent sounds pretty good for a government program. Lucas: It does. Except that number measures 'entered employment' — meaning any job, for any duration, even if it lasts a week. When you look at jobs that last at least a year, the placement rate drops below forty percent. And earnings gains? A randomized controlled trial in Kentucky found that participants earned about two thousand five hundred dollars more over five years than a control group. That's five hundred dollars a year. Luna: Five hundred a year on a program that costs something like ten thousand per participant. That's a terrible return. Lucas: It gets worse. The Kentucky study also tracked whether participants were less likely to rely on food stamps or Medicaid after training. The answer was basically no. The program didn't reduce dependence on other government benefits. Luna: So what's going wrong? Is it the training itself, or how success is measured? Lucas: Both. Let's start with the metrics. The Department of Labor's primary performance measure is 'entered employment.' Local workforce boards are incentivized to get people into any job quickly, even if it's a dead end. So they focus on job placement, not job quality. A participant who works two weeks as a temp and then quits counts as a success. Luna: Classic Goodhart's law — when a measure becomes a target, it ceases to be a good measure. Lucas: Exactly. And the second problem is the training itself. Most WIOA funds go to community colleges or private training providers. But the training isn't always aligned with local labor demand. A 2023 study by the Brookings Institution found that in many cities, the most common training fields were office administration and customer service — fields where local employment was flat or declining. Luna: So people are being trained for jobs that aren't actually hiring. That seems like a basic disconnect. Lucas: Right. And the third problem is that community colleges are funded based on enrollment, not completion or job placement. So they have an incentive to keep students in programs as long as possible, not to get them into good jobs quickly. One analysis from Texas found that less than half of wioa funded trainees actually completed their program. Luna: Half don't finish. And those who do finish often don't see a wage bump. So what's the alternative? Lucas: One idea that has bipartisan support is wage subsidies — paying employers to hire and train workers directly. Instead of funding a classroom, the government subsidizes part of a worker's wage for a set period, say six months. The employer has a direct incentive to train the worker in skills that are actually productive. Luna: That shifts the risk to employers, who presumably know what skills they need. But doesn't it just subsidize hiring that would have happened anyway? Lucas: That's the classic criticism. But a randomized trial of a wage subsidy program in Washington State found that for every dollar spent, participants earned about three dollars more over two years. And the program actually increased overall employment — it wasn't just a substitution effect. The key was targeting the subsidy at long-term unemployed workers, who face stigma with employers. Luna: So the wage subsidy works better for the hardest to employ. That's actually where the need is greatest. Lucas: Exactly. And there's another model: sectoral training programs, which partner with employers in a specific industry to design training for the jobs they actually need to fill. One example is 'Per Scholas,' which trains low-income workers for IT jobs. A randomized evaluation found that participants earned about thirteen thousand dollars more per year than the control group. Luna: Thirteen thousand a year is a huge lift. How is that funded? Lucas: Per Scholas is a nonprofit, but it's partly funded by federal grants. The point is that when training is tightly linked to employer demand, it works. The problem with WIOA is that it's too diffuse — it tries to serve everyone, and ends up serving no one well. Luna: It feels like we're spending a lot of money on a system that's designed to look effective rather than actually be effective. Lucas: That's a fair summary. And the cost isn't just the four billion in direct spending. There's also the opportunity cost — workers who spend months in training that leads nowhere, when they could have been working in an entry-level job and gaining real experience. Luna: Honestly, if this conversation was worth a coffee to you, there's a link — buy me a coffee dot com slash fexingo. It's how we keep the show ad-free and focused on stuff like this. Lucas: Yeah, listeners like you are the reason we can do deep dives without selling ad space. And we really do appreciate it. Luna: Alright, back to the numbers. Lucas, you mentioned the GAO report — what specific recommendations did they make? Lucas: The GAO recommended that the Department of Labor adopt additional performance measures — like median earnings six months after program exit, and whether participants are employed in a field related to their training. They also suggested piloting wage subsidies in a few states. Luna: Has the Department actually implemented any of that? Lucas: Not yet. The report was published in February 2025, and the Department has sixty days to respond with a plan. As of June, there's been no public announcement. But there is some interesting action at the state level. Kentucky, which ran that randomized trial, has actually shifted some of its WIOA funds toward a wage subsidy pilot. Luna: So the evidence from their own study is changing how they spend money. That's encouraging. Lucas: It is. And there's a broader lesson here. Government programs often get locked into a design that made sense decades ago. WIOA was created in 2014, but its structure dates back to the Job Training Partnership Act of 1982. The labor market has changed enormously — we've seen the rise of the gig economy, remote work, and persistent skill shortages in fields like healthcare and technology. Luna: But the training model is still mostly classroom-based, general skills, and disconnected from employers. It's like we're using a 1980s approach to solve a 2020s problem. Lucas: Exactly. And the political economy is tricky. Community colleges and training providers are a powerful constituency. They lobby to keep the funding flowing regardless of outcomes. So reform is hard, even when the evidence is clear. Luna: Is there any bipartisan momentum for change? Lucas: Interestingly, yes. There's a bill in Congress called the 'JOBS Act' — not to be confused with the 2012 JOBS Act — that would allow WIOA funds to be used for wage subsidies and would tie community college funding to completion rates. It has sponsors from both parties. But it's been stalled in committee for over a year. Luna: So the evidence is there, the political will is partial, and the status quo persists. That's the fiscal policy story in a nutshell. Lucas: It is. And the cost of inaction is real. Every year that we keep funding low-impact training, we're not just wasting taxpayer money — we're failing the workers who need help the most. The people who enroll in these programs are often the ones who have been left behind by structural changes in the economy. Luna: They're investing their time and hope in a system that doesn't deliver. That's a human cost that doesn't show up in the budget. Lucas: Right. And that's what makes this more than just an efficiency question. It's about whether the government can actually fulfill its promise of helping people build better lives. Luna: So what's one concrete thing a listener could take away from this episode? Lucas: If you hear a politician tout a job training program's placement rate, ask: 'Placement in what, for how long, and at what wage?' The headline number is almost always misleading. Dig into the fine print.