Democrats have begun the presidency of Donald Trump exiled to the political wilderness. They’ve lost the White House, both houses of Congress, a shocking number of state governments, while the “blue state” vote has turned out to be really just the “blue city” vote.

The party has cast about for solutions, battling it out over identity politics, the proper opposition strategy, and more. But Democrats might consider taking a cue from Trump himself. Namely, his relentless promises to bring back good-paying American jobs.

“It’s the first and most consistent thing he discusses,” observed Mike Konczal, a fellow at the Roosevelt Institute, after reviewing Trump’s speeches. The President understands, as The New York Times’s Josh Barro noted, that most Americans think the purpose of private business is to provide good jobs, not merely turn a profit. Even Trump’s xenophobia and white nationalism are not totally separate from this: Kicking out all the immigrants and rolling foreign competitors are critical components of how he would restore jobs.

Democrats tend to treat jobs as the happy by-product of other goals like infrastructure revitalization or green energy projects. Or they treat deindustrialization and job dislocation as regrettable inevitabilities, offering training, unemployment insurance, health care, and so on to ameliorate their effects.

All these policies are worthy. But a job is not merely a delivery mechanism for income that can be replaced by an alternative source. It’s a fundamental way that people assert their dignity, stake their claim in society, and understand their mutual obligations to one another. There’s pretty clear evidence that losing this social identity matters as much as the loss of financial security. The damage done by long-term joblessness to mental and physical health is rivaled only by the death of a spouse. It wreaks havoc on marriages, families, mortality rates, alcoholism rates, and more. The 2008 crisis drove long-term unemployment into the stratosphere, and today it remains near a historic high.

Trump went right at this problem, telling Michigan in October of 2016: “I am going to bring back your jobs.” Period.

Democrats should consider making the same moon shot promise. But unlike Trump, they should back it up with a policy plan. And there’s an idea that could do the trick. It emerges naturally from progressive values. It’s big, bold, and could fit on a bumper sticker. It’s generally called the “job guarantee” or the “employer of last resort.” In a nutshell: Have the federal government guarantee employment, with benefits and a living wage, to every American willing and able to work.

Why  We Need It

Before getting into the nuts and bolts of how this would work, a word on the broader goals is in order. The job guarantee’s ultimate aim is sustained full employment: A job for every American who could conceivably desire to work, leaving the labor market unable to find enough workers for all the jobs it wants to create.

The benefits of the guarantee for the formerly jobless should be obvious. But just as crucial is how it would help all already employed Americans. When workers compete with one another over an inadequate supply of jobs, they have no power. Conversely, when employers must compete over an inadequate supply of workers, a subtle but profound shift happens. Freed of the fear that they can be cast off into the unemployed and be unable to find a different job, even the lowest-paid workers can demand higher wages and more generous benefits. They can insist on better working conditions and schedules, and job training on their employers’ dime. They can challenge discrimination, harassment, and mistreatment. Unions and labor organizations are empowered. There’s more family stability, healthier communities, more social trust, and more participation in civic life.

Practically speaking, this would require driving the labor force participation rate—the measure of people employed or actively looking for work—as high as it can naturally go, and driving the unemployment rate—that portion of the labor force not employed—to 1 or 2 percent. And then keeping them there.

That may seem a fantastical goal. But we’ve done it before: During the economic mobilization of World War II, the unemployment rate briefly fell to an eye-popping 2 percent. Then, between 1945 and 1970, the rate spent roughly one-third of its time below 4 percent. It’s no accident this period is remembered as an economic golden age, or that inequality remained low during it.

Things have changed since.

As of this writing, there are 1.4 job seekers for every job opening. At minimum, our goal should be no more than one job seeker for every job opening; better yet, less than one job seeker for every opening. Yet this is the best that metric has been after eight grinding years of recovery. More to the point, the closest the job seeker ratio ever got to parity was 1.1 seekers per opening, just for a moment at the end of the 1990s boom. The data doesn’t go back earlier. But there are other signs for when parity is close: The late ’90s was also the only time since 1970 the unemployment rate went as low as 4 percent. It’s currently at 4.6 percent—again, the best level since the Great Recession.

So even in the fondly remembered midcentury, America’s overall rate of joblessness tended to bob like a cork above the threshold that signals full employment. Occasionally, it dropped low enough to touch it, then rose back up. And for the last 40 years, the cork has basically never touched full employment at all.

This multi-decade failure has left workers chronically powerless in the face of employers’ demand for cheap, compliant, and disposable labor. It lies behind the vertiginous rise of inequality, the widespread loss of good job benefits, the expansion of financial precariousness to more households, the spread of contract work and the “1099 economy,” and more.

Meanwhile, what jobs the economy creates, for whom, and on what terms, is determined by the people who own and control the flow of capital and property. This is the nature of private markets in capitalism. Both the left and right accept this arrangement as a fait accompli, and effectively try to bribe this rarified group into creating employment. The right-wing bribe is tax cuts and deregulation. The left-wing bribe is muscular deficit stimulus, welfare state spending, and public investment. The latter works much better at juicing aggregate demand, which is what makes it possible for employers to profit and gain market share by hiring more. But in both cases, the power to shape the job itself and to hire remains with the capitalists.

Inevitably, the most privileged or attractive job seekers are hired first: those who are well-educated, who have histories of stable employment, who are free of criminal records, and (let’s be blunt) those who are white. Only after they’ve been employed does new demand for labor trickle down to everyone else.

The last to be reached are racial minorities, those without college or even high-school degrees, those with longer spells of past unemployment, veterans, and those with felony histories. And their turn is far more brief, since it arrives only at the peak of the business cycle. Then, when the downturn comes, the least privileged are the first to be let go. The result is a bitter dynamic of last-to-be-hired, first-to-be-fired. Even when they’re working, their bargaining position is perpetually precarious, resulting in lower wages and more exploitative treatment.

This is why roughly 5.5 million people are still working part-time when they’d like to work full-time. It’s a big part of why employers can get away with rampant wage theft, abusive scheduling, racial discrimination, sexual harassment, unsafe working conditions, and violations of labor rights. It’s why wages have stagnated for decades for far too many.

That full employment’s absence hits less-privileged Americans harder isn’t just problematic because it’s unfair. It also condemns their communities to repeated cycles of destruction, from which they never have time to properly recover. The unemployment rate for black Americans, for example, is perpetually twice as high as that of whites—a situation that holds within each education level. Same goes for the unemployment rate of all Americans with a college degree versus those with only a high school education.

Conversely, the beneficial effects of full employment, when it happens, are the most pronounced for the least privileged: It’s when employers are finally forced to cater to the people and communities normally exiled from the economy.

Imagine what sustained full employment could do.

So it shouldn’t surprise that African-American thinkers are longtime champions of the job guarantee. Martin Luther King Jr. repeatedly called for it. So did Bayard Rustin, another 1960s African-American civil rights activist, and Sadie T.M. Alexander, the first African-American woman to receive a PhD in economics. Today, black advocates of the job guarantee include Duke economics and public policy professor William Darity, Jr. and New School economist Darrick Hamilton.

With the post-1970 rise of deindustrialization and globalization, and the policy retreat from public investment, regulation, and antitrust enforcement, the American economy has finally begun treating white workers without college degrees as it’s always treated black Americans. Capital is fleeing the rural and small-town communities where Hillary Clinton lost to Donald Trump by three to one, and where new job growth and business creation during recoveries has all but vanished. Their jobs are falling to foreign competition and automation; their rates of unemployment and earnings are worse; and they’re being buried by their mortgages. While Clinton won less than 16 percent of all counties, the counties she did win generated 64 percent of all economic output. This great economic shift gave rise to the economic frustration that flipped key Rust Belt states, handing Trump his victory.

But the President also relentlessly exploited white Americans’ racial fears and anxieties, and the temptations of white nationalism. Combating him will require a genuinely multiethnic coalition; one that can press the anti-racist cause to its fullest, while still giving the white working-class an equal seat at the table. A job guarantee could lay the foundation for such a coalition based on a shared economic interest.

How It Would Work

So that’s the context in which the need for such a program emerges. How would the job guarantee work in its particulars? First, it would be universal: an offer to everyone willing and able to work. It would also be purely voluntary: No one would be required to work.

Every job would pay a non-poverty wage with benefits. One oft-suggested possibility is $25,000 a year—roughly $12 an hour, assuming 52 weeks a year at 40 hours per week. An additional cost-of-living adjustment could be tacked on for local markets as needed. Health coverage and retirement, respectively, could be provided through access to the already-existing Federal Employees Health Benefits program and the Federal Employees Retirement System. Paid vacation days, paid sick leave, and paid family leave would also be included.

Recipients could request part-time work, and their pay would be adjusted downward accordingly. But a minimum number of hours would have to be set for receiving the benefits package; Darity suggests 30 hours a week.

The program would be driven by the demand of Americans for work. This means that it would function as an entitlement, such as Social Security, Medicare, unemployment insurance, Temporary Assistance for Needy Families, or Obamacare subsidies: funded automatically by law, according to how many people qualify, rather than by Congress’s annual appropriations process.

This would require fully funding the job guarantee at the federal level. When a recession hits, nearly all other actors in the economy—state and local governments or private businesses and institutions—see their revenue shrink. They cannot spend at the same or higher levels lest they risk a debt default. But the U.S. government is unique: Its borrowing is backstopped by the Federal Reserve’s ability to create new money supplies. The federal government literally cannot default on its debt (unless it chooses to do so). Rather, the limit on federal deficit spending is the risk of inflation—which is always at its lowest in a downturn.

Then there’s the question of institutional design. No one’s ever instituted a job guarantee in its full glory. But some real world instances have come close, and they provide important pointers. Argentina’s Plan Jefes y Jefas de Hogares—a modest, short-lived, far stingier, but similar job guarantee program in the 2000s—divided functions between the federal level and the local and municipal level. The federal government disbursed the funds and laid out general guidelines for approving employment projects. It created one national database of all people looking for work, and their skills; and another database to track all projects, their development, and assessments. This promoted transparency and cut down on corruption. Given the history of racist exclusion in New Deal work programs in the United States, robust federal enforcement of all anti-discrimination and labor rights law would be essential.

But Argentina left the job of reviewing, approving, and administering projects to the local and municipal governments. Bard College economist Pavlina Tcherneva, who’s studied the job guarantee and Argentina’s experience with her Levy Institute colleagues, emphasizes that local nonprofits, community groups, and social organizations across America already know their local communities’ needs and are working to meet them. They’re just perpetually under-staffed and under-funded. So in our case, the first step would be for local and municipal administrators to ask these groups what projects they have already developed that just need to be scaled up.

Next, those same local administrators should determine what public works and pieces of infrastructure each community needs to have built or revitalized. The crisis of lead in the water pipes of Flint, Michigan, shows there’s no shortage of need in this arena.

Finally, local job-guarantee administrators should open the floor to proposals from the local community: Let everyone from churches to civic organizations to individuals submit ideas. That would create a ground-level dialogue, giving Americans more democratic control over the planning of their local economies. Perhaps some degree of local democracy could be introduced into picking the program’s local administrators.

People seeking jobs would come to these local offices, which would draw on the federal databases to link the potential workers up with the most appropriate projects. Crucially, workers would be matched with nearby jobs according to skills they already have. Training and child care could be provided as needed by other job-guarantee recipients, as happened in Argentina. The idea would be to “meet workers where they are,” in terms of their capabilities, their needs, and their geographic location. Contrast that with the private sector, where low-income workers can spend enormous time and energy commuting, finding child care for their children, “acquiring skills,” or even moving to a different city or state, all for employers’ convenience.

We could create an entirely new federal agency, with internal power and decision-making distributed as described among local and municipal branches. Another possibility is to rebuild and repurpose the institutions we already have. The Labor Department could handle the federal-level duties, while unemployment benefit offices could become “employment” offices—the local points of contact for job seekers. Both could liaison with local governments. Even the U.S. Postal Service could get involved: Proposals are already circulating to provide basic banking services through the agency to the 68 million Americans who lack them. Many of those citizens would also be the job guarantee’s biggest beneficiaries, and would need an account in which to receive their wages. Postal banking could solve both problems in one fell swoop.

What It Would Cost

Now let’s talk price tag. Back-of-the-envelope calculations suggest the program described above would run at least $670 billion in its first year—or about 3.6 percent of the economy—if instituted today. That’s very large, but hardly unimaginable: It’s roughly the annual cost of the federal government’s various health-care subsidies, and less than Social Security.

More importantly, the cost would rapidly and dramatically shrink once the program was up and running. Millions of formerly jobless would have new income to spend, so private businesses would expand, their wage and benefit offers would rise, and people would leave the job-guarantee rolls for greener pastures. The program would moderate itself by expanding the private sector. Eventually, it would stabilize at a much lower level of spending. Case in point: Argentina’s Jefes program began in 2001, and recipients peaked at two million in 2003. But by 2006, the number of Jefes recipients had fallen by 40 percent. Over that time, the national unemployment rate fell from 21 percent to 8 percent.

On top of all that, the rolls of people qualifying for safety net programs like Medicaid, SNAP, housing assistance, and more would shrink as well. We’d face fewer of the social ills that flow from joblessness—crime, incarceration, mental health, addiction, etc.—further reducing spending. Rough estimates by Darity, Tcherneva, and several other economists suggest at least a fourth of the job guarantee’s annual budget would be offset by these other reductions, and maybe much more.

From then on, every private sector recession would swell the ranks of job-guarantee recipients again, but never by that much. Right now, recessions feed on themselves: People lose jobs, their consumption drops, so more people lose jobs, until the recession bottoms out. But employment through the job guarantee is limited only by human imagination and ingenuity. A recession is just a collapse in the private sector’s ability to employ everyone according to the capitalists’ priorities. So recessions would simply throw workers onto the job guarantee’s alternate employment, with its attendant wage floor. The bottom for recessions would be much shallower, and recoveries much faster, shrinking the job guarantee rolls again. We would be spared the human and social wreckage that comes with spells of mass unemployment. Preventing mass joblessness would prove far cheaper than eliminating it once it’s set in.

An even more intriguing possibility is that the job guarantee would eliminate the need for the Fed to adjust interest rates.

Here’s how that would work: Currently, Fed officials operate on the assumption that some amount of unemployment is necessary to control inflation. As workers’ bargaining power rises, an arms race sets in across businesses. Workers claim more revenue as compensation, capitalists try to preserve their cut by raising prices, then workers demand higher wages for the increased cost of living. Wash, rinse, repeat. The result is the wage-price spiral, delivering higher inflation.

The Fed intervenes by raising interest rates, which squeezes the supply of credit and forces companies to stop expanding or to downscale. This stops the arms race by destroying jobs. But, once again, it’s left up to the capitalists to decide just whose jobs are to be destroyed first.

There’s a burn-the-village-to-save-it quality to this practice, since allowing inflation to continue also destroys jobs as price signals go haywire. But again, with a job guarantee, the downturn would simply move workers onto the set wage and compensation rate, which doesn’t engage in a bidding war. This would end the arms race as well, but without casting people into joblessness and all its attendant human destruction. So the job guarantee could stabilize the ups and downs of inflation as well. (This would be complicated if the job guarantee’s wage was indexed to rise with inflation, though not indexing the wage would force regular legislative fights to raise it instead.)

The Work

Now, what of the jobs themselves? What kind of work would these people be doing? Tcherneva focuses on labor that isn’t capital intensive, that rarely requires specialized skills, and that serves ongoing needs for care, restoration, conservation, etc. Some liberal policy thinkers refer to this as “human infrastructure.” Current extant examples include ecological preservation and public park revitalization in the Hudson River Valley, creating urban gardens and local markets to combat the problem of food deserts, or refurbishing historic buildings in Newburgh, New York. In Argentina, local communities used the Jefes program to set up butcher shops, bakeries, clothing and toy makers, grocery stores, homeless shelters, domestic violence shelters, mass recycling centers, urban farms, food kitchens, and more. Along with infrastructure projects, the New Deal’s work programs put on plays, built sculptures, and painted murals.

Certainly, the program could encompass traditional infrastructure and public works as well. But these are finite projects with a completion date, and the need for them may not line up with the business cycle. Meanwhile, there will always be children and the elderly to care for, sick people to minister to, citizens to educate, cities to clean, neighborhoods to beautify, and parks to tend. This work would be the backbone of the job guarantee.

The fear of “make work” has always dogged these sorts of public employment programs—the idea that people are being paid to do things that are useless for society. But this is an odd worry in a world where private employers hire people to create endless potato chips flavors and messaging apps that say nothing but “Yo.” Anthropologist David Graeber even coined the term “bullshit jobs,” referring to legal, clerical, financial, and administrative work, sometimes highly paid, that’s done by people “who are themselves convinced their jobs are meaningless.” Many of us are put to work for the capitalists’ convenience, in accordance with what they find to be profitable. Hence all the paper-pushing.

Behind worries about “make work” lies the unspoken assumption that what is profitable—i.e. what is desirable to shareholders—is the sole possible determinant of value. So job guarantee supporters should embrace this fight, and force opponents and critics to argue the reductio ad absurdum.

All usefulness is irreducibly subjective. Human beings decide for themselves what economic value is, in dialogue with one another. Capitalism’s competition, price signals, and profit motive are one way to have that conversation. And in many instances it’s a very good way. But one feature of the job guarantee is to provide an alternative way: It opens up political space for local democracy and dialogue to determine value instead, and to then use the public tool of the federal government to pay people to produce it. After all, nobody audits public libraries to see if they’d be better provided by the private markets—we just decided, as a society, that they’re worth having around.

Problems with corruption and graft will be inevitable, and the best cure will always be healthy institutional enforcement and self-policing. But by 2005, reports of Argentina’s Jefes program’s mismanagement, waste, or corruption were rare. A 1937 assessment of New Deal work programs rated the quality of 40 percent of them as “excellent,” and only 9 percent as inferior.

Of course, a job guarantee set to $25,000 a year plus benefits would wipe out a fair amount of low-pay private employment at a stroke. No employer could underbid it, since people could “vote with their feet” and take (or propose) job guarantee employment if they preferred. Yet no business is entitled to exist, regardless of the conditions in which it employs people. Like the minimum wage, the point of the job guarantee’s compensation floor is to wipe out low-pay private employment. And, unlike the minimum wage, the job guarantee carries no attendant risk of increasing unemployment.

So no doubt, the shape of employment and of our lifestyles would change. There’d probably be far less fast-food employment and low-wage service work, and fewer Uber drivers. We’d all have to cook more, drive less, and do more of our own household chores (pending the arrival of the robots). But would that really be a bad thing?

Time for a Guarantee

There is an objection on the left of a different sort: That the job guarantee goes much too far in indulging the right’s romanticization of the dignity of work. These critics often argue for the universal basic income (UBI), another ambitious policy idea that would give every American an unconditional monthly cash grant. It, and the job guarantee, have sometimes been cast as an either-or choice.

There is no reason for this; the two proposals are complementary.

The UBI does provide every worker the option of exit from the labor market, and could thus increase bargaining power as well. But this effect is passive. The job guarantee’s strength is providing Americans direct control over the social infrastructure of job creation.

In a way, the program could be the welfare state’s biggest champion. It would finally put the idea that unemployment is frivolous or voluntary to the ultimate test. No longer could people assume joblessness is chosen, based purely on the fact that joblessness exists, without ever reckoning with the supply of work. And no longer would policymakers run themselves in circles trying to “fix” the jobless—either through education, improved habits and values, or just by relocating—to better meet the demands of the capitalists.

In so doing, the job guarantee would lay bare that plenty of people have very good reason to not be in the labor force. Perhaps they are retired, or raising children, or going to school, or caring for a sick family member. While not only a delivery mechanism for income, jobs certainly serve that role. But they cannot and should not be the one mechanism for everyone. That we’ve failed to plug these holes in our society’s “income delivery system” is the driving force behind America’s intransigent poverty rates, particularly for young children.

Yes, American political rhetoric is shot through with the idea that individuals have a moral duty to work, and this is sometimes used to galling effect. But it isn’t entirely off: For the UBI to purchase anything in the first place, someone must labor to produce those goods and services. As the effects of joblessness on individuals’ physical and mental well-being shows, the vast majority of people really do feel better about themselves when they’re contributing to that social project in some way. One of the most remarkable aspects of the Jefes program was the number of women who joined up. Given the chance to participate in their local communities and economies, and not just perform domestic housework, these women reported striking changes in their sense of self-worth—like they had “grown wings.” Indeed, political panic over the rapid shift in gender norms apparently played a big part in the Argentinian government’s decision to kill the program—and to replace it (for women, at least) with a UBI-style check.

The crucial thing to remember is that it’s in the capitalists’ interests to never provide enough jobs, and to keep large amounts of the employed in a permanent state of precariousness. This is the bait and switch at the heart of the right-wing habit of “incentivizing” people to work, either by cutting their government aid or by shaming them into it. That joblessness really does wreak havoc on human well-being gives this posture a certain plausible sheen of beneficence. But the work that awaits people is only offered on the capitalists’ terms; rather than being dignified, it is often exploitative and demeaning, and there is never enough of it to go around.

As Polish economist Michael Kalecki observed, the real issue is that full employment strips business owners and the capital-owning class of power. This is what worries that the job guarantee will displace private employment, or provide mere “make work,” are really about.

With full employment, the capitalists lose their leverage to depress workers’ wages and must give up more profits. But, more than that, when it comes to running endeavors that are ostensibly “theirs,” the capitalists are forced to bargain with and bend to the will of workers “below” them. Their position as the demigods of the economy—granting employment when they are appeased, and taking it away when they are angered—is undone.

Capitalists do not want recessions, of course, since their income and wealth holdings suffer as well. But they don’t want the economy running at full steam, either. Their solution, as John Maynard Keynes put it, has been “abolishing booms and thus keeping us permanently in a quasi-slump.” Perhaps this sounds familiar.

In defiance of this economic regime, the job guarantee asserts that, if individuals bear a moral duty to work, then society and employers bear a reciprocal moral duty to provide good, dignified work for all. It would finally make real the ideal, stated in Franklin Roosevelt’s “Economic Bill of Rights,” that every American possesses a “right to a useful and remunerative job” and “to earn enough to provide adequate food and clothing and recreation.” Not a paternalistic aid, and not some tribute to aristocratic virtue, but a right to be claimed and exercised. As the history of the civil rights era and the battles for the voting franchise demonstrate, rights provide the basis for mass political organization, and for demands of inclusion and equality.

A job guarantee would not only remake the American economic order. It would arguably remake the country’s political and moral order as well.

Jeff Spross is the economics and business correspondent at TheWeek.com. He was previously a reporter at ThinkProgress.