Next to the daily acrimony of the Trump era, the 2016 Democratic primary controversies now seem almost quaint. One of the most divisive issues in the contest between Hillary Clinton and Bernie Sanders was one on which both candidates agreed: opposition to the 12-nation Trans-Pacific Partnership. Although Clinton had announced her opposition nine months prior, lingering distrust of the former State Department negotiator haunted her through July’s platform drafting and Philadelphia convention. Whenever Clinton surrogates addressed the stadium crowd, they were met with anti-TPP signs and chants. The pact was high on symbolism—a reminder of an elite-led foreign policy strategy often at odds with the party’s base.

Since the 1980s, trade has been tough for Democrats. As Obama’s trade representative Michael Froman wrote in these pages, the party has championed the global, low-tariff liberal order going back to Franklin D. Roosevelt. Yet, since Bill Clinton’s fateful advocacy for the North American Free Trade Agreement (Nafta), Democrats in Congress and the party’s activist base have become trade skeptics as concerns over offshoring of jobs and pollution have become more prominent. After decades of rushed congressional votes and low transparency on trade deals, trust is nil between the pro- and anti-trade camps.

Progressives have a stake in getting out of this rut, however—for both electoral and substantive reasons. If politically important swaths of the country see trade-related foreign entanglements as a threat, it’s but a short leap to seeing international engagement writ large through that lens. Donald Trump—by flattening the more nuanced progressive critique of globalization into blunter bromides—did exactly that. “Hillary Clinton unleashed a trade war against the American worker when she supported one terrible [trade] deal after another,” he told a crowd of metalworkers in Monessen, Pennsylvania in June 2016. “A Trump administration will end that war by getting a fair deal for the American people and the American worker. The era of economic surrender will finally be over.” This message may have seemed tough, but the promise was empty.

It was at the end of World War II that we created our set of global institutions to regulate trade. (Most) every country was at the table. For decades, this created profits and good jobs. But as these global institutions produced diminishing returns and gridlock, policymakers have turned toward bilateral and regional deals that give capital more flexibility. Meanwhile, the trade regime hasn’t nurtured our domestic institutions that support workers and their families, such as unions and the social safety net; thus, people are angry. In focus groups conducted by Democracy Corps and the Roosevelt Institute in Macomb County, Michigan after the election, the 45th President’s call to slow down trade resonated. One homemaker expressed their hope that Trump is “going to hold companies accountable for taking jobs out of our country, and charging them if they want to continue making their products in other countries.”

Can we not, in fact, find a way for the system to again be win-win? Does it have to be finance wins, workers lose? That bargain is not sustainable for progressives.

Washington abhors re-litigating the past. But in trade policy, history exerts a special pull. By signing Nafta and other trade and investment agreements in the 1980s and ’90s, policymakers unleashed powerful path dependencies. It became easier to do more-of-the-same, and harder (in the absence of catastrophe) for critics to demand abrupt course correction. Businesses have come to think of trade arrangements as locked in, buoyed by other countries signing thousands of similar pacts in the years since. But negotiators were not omniscient, and so they delegated much of the interpretation of trade and investment rules to third-party adjudicators, resulting in unintended consequences and mission drift. Thus, in approaching reform options, we begin not from a blank slate but a heavily annotated one.

A workable path forward will require a re-examination of fundamental assumptions. We need less talk about promoting efficiency and more talk about creating sustainable institutions. After all, the point of an economy is to service people—people who inhabit families, neighborhoods, and societies. The economy is not a separate game-scape, where policymakers get points for hitting certain targets for productivity or stock prices even if those gains are not widely shared. Indeed, a growing body of research shows that pursuing growth without equity is bad for both.

“Institutionalism” is, therefore, an alternative, and a more holistic policymaking framework. The phrase has diverse meanings in the academy, but a few of its insights suffice for our purposes. First, policies that look like a good idea on paper can have wildly different results across institutional settings. Second, how society views the outer limit of the possible, or desirable, is shaped by the institutions that came before. If one policy agenda (say, trade) has been associated with destruction of middle-class livelihoods in the past, no amount of reasoning will convince workers otherwise. Likewise, new policies need to be designed with an eye toward nourishing the constituencies that will support it, thereby guarding against future rollback. Below, I suggest two components for a more institutionalist trade policy: rebuilding labor domestically and embedding it in the global order.

The Domestic Lens: Rebuilding Labor

When discussing trade pacts’ economic costs and benefits, both sides focus on aggregates: Has the economy as a whole lost or gained jobs? From critics, we hear about the giant sucking sound of jobs moving to Mexico and China. Advocates tout the upside of lower prices and downplay the costs by pivoting to factors that reduce the number of manufacturing jobs more than does trade, such as automation.

Such aggregates tell us little about why trade politics have become so toxic. While 5.1 million U.S. jobs have been created in the health and education sectors since China entered the World Trade Organization (WTO), the manufacturing sector lost 1.3 million. Seems like a net gain, by these numbers alone. But not all jobs are created equal. The jobs lost were in specific places and in sectors disproportionately represented by unions, unlike the job-gaining sectors. Sixty percent of the jobs lost in American manufacturing over the past 15 years were unionized, over 750,000. These losses were concentrated in the communities that flocked to Trump. In contrast, only 9 percent of the gained jobs in health and education (472,353) were represented by unions.

Deunionization has a negative externality felt beyond the workers directly affected. A loss of union jobs has been a major factor behind the rise in economic inequality, as workers have lost institutional leverage to bargain for a fair share of productivity gains. Economists at the International Monetary Fund have concluded that a 10 percent decrease in union density is associated with a 5 percent increase in the top 10 percent’s income shares. Political scientists Sung Eun Kim and Yotam Margalit provide one reason why: Unions have an unparalleled ability and (still) wider reach than other civic institutions to communicate anti-inequality positions to their members and the public at large. As unions weaken, the political base of support for more egalitarian tax, health, and trade policies has eroded, hurting the populace as a whole.

On top of the generalized economic impact, the Democratic Party pays a particularized cost. Hillary Clinton lost Michigan, Pennsylvania, and Wisconsin by far fewer votes (77,764) than the number of union jobs lost there in the last 15 years (392,063). If those union jobs had remained and the union had been able to motivate even one in five of these members to vote for Clinton, she could have achieved an Electoral College victory of 274 to 264. In Michigan, for example, Hillary Clinton lost by 10,000 votes—almost exactly the number of workers certified for trade adjustment assistance under a narrow Nafta-only program. The academic research is clear on what these particularized losses have meant for the Democratic Party: the loss of a geographically distributed, electorally efficient institutional base.

Over the last year, we’ve become familiar with one aspect of this problem: the white working-class voter. Analysis by the Pew Research Center finds that Trump’s margin of support among whites without a college degree was the highest on record: 67 percent, a gap of 39 points over Hillary Clinton. Yet this is but a culmination of a longer-term trend that needs to be parsed carefully. According to calculations from University of Puget Sound political scientist David Sousa, in 1964, nearly 60 percent of non-union whites voted Democratic. Eight years later, this number was cut in half, and never again has it reached above 40 percent. While white union workers demonstrated a similar drop, it was from a much higher starting point (80 percent in 1964, dropping to a low of around 40 percent in 1972). In any given year, unionized white workers were 11 to 22 percent more likely to go blue—a vital buffer.

Unions also help with political turnout of non-whites, who overwhelmingly vote Democratic when they make it to the polls. From the 1960s to the 1990s, 45 to 65 percent of non-union blacks turned out, while unionized blacks were 5 to 15 percent more likely to turn out. Post-1992, political scientist Dukhong Kim finds that unions helped reverse low turnout by Asians and Latinos, as well. And despite an ugly history of racial exclusion, unions provide a forum (however imperfect) for challenging intra-working-class divisions on the basis of shared economic interests. Given the ongoing racial segregation of neighborhoods and churches, these forums matter. Even if automation swamps trade as a factor in job loss of union jobs, the latter is the lever that policymakers more readily controlled. For Democrats, there is almost no credible path to a 270-vote majority in the Electoral College that does not include the Midwestern states particularly impacted by trade.

Before going further, it’s worth making clear: There is nothing magic about manufacturing. You can have intense competition in product markets alongside strong unions, as well as autarky without them. Rather, the supply of, and demand for, union organizing has more to do with a country’s history and legal institutions. When the Supreme Court was pushed out of the anti-labor injunction business through New Deal-era labor laws, the aspirations of Depression-wracked workers could be quickly realized. In other words, when the institutions are right, change can happen fast. Only 9 percent of workers were in a union in 1936. After Roosevelt threatened to pack the Supreme Court, justices accepted, in 1937, a reduced role on labor oversight. By 1940, the unionization rate had more than doubled. While the defense industrial mobilization of World War II led to further gains thereafter, the lion’s share of the increase the United States labor movement would ever see came during that three-year period.

Unions, like any voluntary institution, are fragile. They are important to our democracy, yet require nurturing to sustain them. And that’s precisely why the decision of many Democrats to pass Nafta was so puzzling. In the best of times, labor organizing is hard work, replete with collective action problems. And the early 1990s was not the best of times—after decades of assault by Republican presidents and judges, union coverage had fallen from their post-war peak of 28 percent to 13.8 percent in 1993. When your electoral coalition tells you that your commercial policy will harm them (as labor did constantly during the early 1990s), it should merit heeding.

The remedy for political malpractice is not band-aids, but justice. The trade adjustment assistance offered in one form or another by the Department of Labor since the 1960s is difficult to qualify for, insufficient for the total impact, and carries the connotation of handouts. A more justice-oriented approach would acknowledge the wounds of the past by re-branding adjustment programs as “trade reparations.” These would target affected communities with reinvestment dollars, and the ribbon-cutting ceremonies would emphasize themes of contrition and reconciliation. An incidental benefit could be signaling to white voters that making financial amends for past wrongs is not only for non-whites but a legitimate policy tool more broadly.

To translate this into forward-looking momentum, Democrats should adopt a rule that no trade deal goes forward without substantial union support. This will lessen polarization and avoid congressional showdowns better than the bespoke labor “listening sessions” used during the early Clinton and Obama presidencies. If this seems audacious, there are precedents. Through arrangements like the U.S. Trade Representative’s industry advisory committees, business already has an effective veto. If bipartite power-sharing between business and government is possible, surely bringing labor into a tripartite framework is too. Indeed, an informal version of such an arrangement was used in the U.S.-Jordan trade deal in 2000. By that time, Bill Clinton had lost so-called “fast track” powers, which limit congressional involvement (fast track had been in place when Nafta was passed). As a result, he needed buy-in from congressional Democrats, who in turn demanded a better deal for labor. This is close to the opposite of the process during TPP negotiations, when labor was heard but not heeded.

To secure its future, the party should also adopt a goal of a net unionization stability ratio. Like cost-benefit analysis or environmental impact assessments, this means that all policies should be assessed with a goal of maintaining or increasing the number of union jobs in the economy. Thus, if trade openness and automation led to a decline of 10,000 union jobs, new jobs created elsewhere in the economy must yield the same number of union members. Andy Stern and Eli Lehrer recently provided one way of accomplishing this: waivers for states and localities of federal labor law requirements that often serve as a ceiling rather than floor for ambitious experimentation. Thus, for instance, if unionized Ohio autoworkers are displaced by rising imports, policymakers could mandate that the local health-care industry recognize unions even without a National Labor Relations Board election. Such an approach would require statutory changes and would need to be tailored to pass constitutional muster. But the basic compact would force business, as a class, to internalize the costs to K Street’s advocacy for ever more trade agreements.

Finally, industries that benefit from U.S. government overseas promotion efforts should be expected to (at least) adopt neutrality agreements to easily recognize unions supported by their workers. Financial services, for instance, have become major beneficiaries of our trade policy. There is no reason for the sector to have the lowest union density (2.3 percent) of any industry tracked by the Department of Labor.

The International Lens: A Global Order For the People

Lessening worker anxiety about the downsides of trade is necessary but insufficient to build a popularly supported international economic order. Keeping what works and fixing what doesn’t is a worthy goal. Financialization, monopolization, tax avoidance, and carbonization cry out for cross-border solutions.

In thinking how to build a lightly federated global social democracy, an institutionalist lens is instructive. The post-1980s neoliberal order was not crafted at some big constitutional convention. The skeleton was put together by a series of national, regional, and multilateral decisions over decades that reinforced one another. The meat on the bones was more often added after the fact by experts, who determined what negotiators meant when they said in Nafta (for instance) that cross-border investors should be treated “fairly” (which wasn’t precisely defined in the pact’s text.)

Just as the building up of this system was gradual and layered, so would have to be its reform. Blowing up the global order is easier said than done. As this goes to print, Trump has choked on his many threats to do something big on trade. For all the drama of Brexit, the United Kingdom seems set to basically accept the norms set by the European Union if it wants to retain market access. So takeover must be thoughtful and will probably need to be layered on top of institutions which many on the left would not have created were they in charge of policy decades ago.

But layered where? The current landscape of organizations that make and enforce our global trade rules would require serious reforms to become a viable home for progressive work. Consensus voting rules at the WTO all but block meaningful upgrades in these rules. The body is good at ferreting out small infringements on liberal norms by liberal states, but not so good when it comes to huge affronts by illiberal ones. In part because American government is relatively transparent, trade lawyers can spot slight subsidization of Boeing planes by Washington State or burdensome dolphin-safe tuna labels overseen by the Department of Commerce. In contrast, China’s pervasive statist pseudo-capitalism has made violations harder to root out, to the point that U.S. trade officials took the unusual step in January of calling China’s accession an error and WTO rules insufficient to meaningfully constraining the country’s behavior.

Partly for this reason, bilateral and regional initiatives have grown in popularity. While the World Trade Organization has adopted only around 200 rulings over its two decades in existence, the investor-state dispute settlement system codified in bilateral treaties has put out nearly double that over the same time period. According to UN statistics, more than 50 new cases have been launched every year since 2011, topping 70 in 2015 alone. The regime is an anomaly in international law, allowing multinational investors to challenge host government policies (on the environment, financial stability, and more) outside of national courts in three-arbitrator tribunals that the investors are partly responsible for appointing, and to receive an award enforceable virtually anywhere in the world.

This system has come in for clubbing by voices ranging from Elizabeth Warren to Hillary Clinton to even Trump trade representative Robert Lighthizer. Leftist Bolivian president Evo Morales declared that “the transnationals always win.” And a 2012 advocacy report deemed the system “neither fair, nor independent, but deeply flawed and business-biased.” According to a Guardian column by George Monbiot, investment rulings “overthrow the sovereignty of parliaments and the rulings of supreme courts.”

These particular critiques overreach in a few ways. It is true that investors can bring claims but states can’t. All else equal, this would suggest a pro-investor tilt in case outcomes. But this hypothesis has not been borne out in the data: Governments win more than they lose. When they do lose, the remedy is not the overruling of legislation but payment of damages. In the cases to date, public treasuries have been ordered to pay investors only pennies on the dollar of the total damages claimed.

Yet the investment law system is riddled with flaws. The argumentation (that is, the dicta) in many awards is excessively solicitous to investors’ interests and dismissive of legitimate regulatory and political considerations. And the damages remedy can be counterproductive. On the one hand, even “pennies on the dollar” can total a lot of dollars when a poor country’s budget is on the line. On the other, a dollar-denominated fix fails to give investors the policy framework they (claim to) need, instead incentivizing legal rent-seeking. For the 138 cases where we have complete data, investors are typically awarded more than 5.5 times their legal costs. The average award of $52.7 million was secured on an average legal expenditure of $6.3 million. In some years, the average return to legal costs was more than tenfold. Notably, the average holds even though investors lose most of these cases. These figures help explain why a hedge fund litigation finance industry now invests in corporations’ legal claims against sovereigns. There is no other international legal regime where the standing and remedy rules would make this a viable business model. Common-sense reforms should eliminate the possibility of damages for anything but outright expropriations.

Counterintuitive as it may seem, there are three institutional features of the investment law system that may be amenable to progressive goals.

First, compared to the WTO’s seven-person judicial appeals body, the investment system is relatively porous and open to competing ideologies and expertise. Because the arbitrators are hired on a case-by-case basis, more than 340 individuals have served in three-member tribunals over the quarter-century of investment arbitration. Arbitrators come from the public, private, and academic sectors and from numerous countries. While a smaller number of arbitrators get the bulk of the appointments, the exploding caseload and concerns about diversity have led to pushes to further enlarge the arbitral pool.

Second, the rules are left vague enough to capture violations by governments of both the letter and the spirit of their obligations to investors. Thus governments commit to not only refrain from property-destroying expropriations, but also to accord investors “fair and equitable treatment”—a loose legal standard that arbitrators have the power to expand to fit the particularities of each case. There is no binding precedent—which makes rulings unpredictable but also means bad decisions are (at least potentially) debated, contested, and updated in later arbitrations.

Finally—unlike more untethered forums in international law—arbitration decisions are ultimately supervised by national courts. Thus, when companies and arbitrators overreach, they can and have been overruled by national judges who decline to enforce international awards. National legislatures control how much scrutiny judges apply in their review, and could strengthen this at any time without cumbersome international negotiations.

If these features seem less than appealing on first glance, consider them in light of the morass in other international courts at the moment. International human rights courts have been unable to secure legitimacy with many of the governments they seek to hold accountable, with the International Criminal Court seeing a number of countries exit or threaten to exit, particularly African nations who have felt prejudiced by the court. And whether Clinton or Trump had won, the United States was set to collide with the WTO for reasons explored above.

A better balance is possible between relative effectiveness and legitimacy. First, more international rule-making on labor and environmental questions should be taken out of the hands of government—which always have an incentive to put diplomatic considerations before an unrelenting pursuit of justice for the little guy. Instead, responsibility should be delegated to a similarly porous collection of individuals as in the investment realm. Progressives should aim to flood the system with their own roster of 340 labor and environmental advocates. This would not only have the direct benefit of more progressive rule development, but would carry the indirect benefit of skilling up hundreds of individuals who could help staff international economic positions in future progressive governments.

Second, malleable legal norms should be embraced. Currently, labor violations in trade deals can be overly specific. For instance, in 2008, unions petitioned the U.S. government to sue Guatemala under a trade pact. They maintained that, far from upholding international labor rights, the country was not even upholding its own labor laws. After nearly a decade of litigation, a three-person arbitration panel (the first of its kind) agreed, but found these violations insufficiently linked to trade, a reference to a legal hurdle in the treaty text. If such hurdles were removed and violations could also be found when governments and companies act against the spirit of industrial peace, labor everywhere would be better off.

Left unchecked, such roving arbitrations would pose sovereignty concerns. Here, too, investment rules offer a vital precedent. By giving national courts tools to supervise arbitrations, we maintain a linkage between the domestic and international institutions, allowing them to balance against the other’s weaknesses. It would be possible that a government could be blame-free under domestic law and guilty under international law, or vice versa. But in a system with the right-sized damages discussed above, the monetary stakes for such inconsistencies would be limited, while the laboratory of norm development could keep cooking.

Toward Global Popular Constitutionalism

Franklin Roosevelt is often cited for his trade advocacy, but equally important was his skepticism that unelected judges would produce legitimate governance. He famously clashed with the Supreme Court on labor and regulatory issues, and his vision for the post-war order emphasized contestable diplomacy and dialogue over final-word adjudication. By buffering union organizers from assault by judicial rulings and business, he created the conditions for an impressive deepening of civil society.

Trade policy is not institution-neutral. It tears some down and builds others. It is itself sticky against efforts to pull it down. Getting beyond unproductive polarization and toward solutions will require sidelining the lawyers and economists that cloud the horizons of the possible. An institutionally robust, global popular constitutionalism—not a global government document, but a sustained practice of challenging democratic deficits and inequities through mechanisms like those outlined above—points to a better and more equitable way of managing economic integration.