People sometimes talk about education as though it’s something that ends. High schools, colleges, grad schools: They commence, they conclude, and then you just work.

But in an economy in which automation and globalization are rapidly changing and even eliminating certain jobs, American workers and companies might come to see education not as a life-stage, but as a way of living.

The need for ongoing education is clear from the recent experience of America’s manufacturing sector. Between 2000 and 2008, industrial production grew by 12 percent while the number of U.S. manufacturing jobs declined by more than 3 million. Many of these lost jobs were concentrated in Rust Belt areas that swung heavily toward Donald Trump. When The Atlantic's Alana Semuels investigated ways for Democrats to help the Rust Belt economy, she reported that almost every economist she spoke with mentioned the same thing: worker training program.

But as Semuels pointed out, there’s just one problem with government-run retraining programs. They don’t really work. Workers are reluctant to attend them for fear that they’re crummy (a reasonable estimation, given the research), or else that there won't be suitable jobs once they’ve finished the program.

For this reason, the best training might be on-the-job, like the well-respected apprenticeships in Germany. U.S. companies have resisted European-style training programs so far, since they’re expensive, unproven, and risky for the firm, as Americans switch jobs often and can take their training to another company.

The U.S. government has the money to retrain workers, but not the curriculum. Companies have the expertise to teach relevant skills, but won’t spend the money. So why not bring them together to create a government-backed corporate retraining program, one in which Washington pays companies for only those curricula that raise workers’ wages? That’s the gist of a proposal, "Toward a New Capitalism,” from the Aspen Institute's Future of Work Initiative, which was co-chaired by Senator Mark Warner and former Indiana Governor Mitch Daniels.

It is called “pay for performance.” Here’s the idea in a nutshell. A young worker comes to Derek’s Factory to make freezers. After a year, I decide freezers are a terrible business, so I spend a couple thousand dollars training the young worker to make dishwashers. If she’s great at it, and her wages rise, the government will pay back Derek’s Factory a certain portion of my training cost. And what if the young worker leaves for Goldberg’s Factory? Doesn’t matter. I still get the same compensation from the government.

There’s a slightly different way to think about this idea. It’s not just a retraining policy. It’s an investment policy. By training workers, businesses are essentially buying a small equity stake in their future wages. If their wages rise, the company gets money, while the worker gives up nothing, purely benefiting from the training program. “We have done a really crummy job to help people who are dislocated by trade,” Senator Warner told me. “We’ve struggled to retrain them with government programs. This is an experiment that could work.”

What would be some of the negative implications of this idea? It’s possible that government compensation still wouldn’t be enough to encourage cash-strapped companies to train workers, who could simply automate or offshore their labor. The opposite concern is that the government might end up handing out too much money to rich companies whose “training" was going to happen anyway. But there are some ways to address this, like restricting the policy to non-highly compensated workers. Since the “recoup rate” is only a fraction of the training investment, it’s not like companies like Apple could hack a federal retraining program to become a profit-gusher at the expense of taxpayers.

The larger potential objection is that “pay for performance” and other policies in the Aspen Institute paper aren’t dramatic enough to make a different. Other proposals include raising the cap on employer-provided education assistance and reforming occupational licensing. According to Senator Warner, this micro-focus is partly by design. “I’ve talked to a lot of people about AI and automation and all those changes that are 15 years out,”he said. “But what do we do in the next five years to buy us some time, so that people feel like they have a shot?”

In the presidential election, the most famous proposals were gargantuan in their scale and dramatic appeal (build the wall!, make college free!, etc.). The Aspen paper represents an different ethos, a philosophy of tweaks that is perhaps more aligned with the reality of policy-making. Politicians promise political revolution as if it might fall from the sky, like a spaceship, fully-formed. But most policy is a boring of hard boards, an assembly of little parts built from the ground up.