Tariffs Are a Weak Climate Tool Without International Coordination

John Cochrane framed tariffs as “an answer in search of a question.” Revenue, reshoring, bargaining leverage, and geopolitical coercion had all been offered as rationales, he said, and each looked weak to him. The environmental version of the argument was at least coherent enough to examine: the United States and Europe have regulated some dirty industries so tightly that production moved abroad, improving domestic air and water while leaving global pollution unchanged or worse. If steel, coal, rare earth mining, or other pollution-intensive activity moved to China, the United States could feel cleaner while global emissions rose.

That logic creates a narrow opening for tariffs. If production brought back to the United States were subject to stricter environmental rules, domestic pollution might rise while global pollution fell. But Cochrane’s premise was that even if this story identifies a real problem, tariffs are unlikely to be the best solution. “If you identify a question to which tariffs are helping a little bit,” he told the presenters, the useful next step is to ask “what the right answer is.”

The session’s substantive disagreement turned on that point. Joseph Shapiro argued that trade policy can matter for climate because there is no global environmental regulator and because climate damages do not depend on where carbon is emitted. Arik Levinson argued that much of the popular tariff rationale rests on a mistaken picture of “outsourced pollution,” and that tariffs on clean technologies are especially hard to defend. Cochrane’s own preferred answer was closer to the standard economist package: a carbon price with border adjustment, little else, and no attempt to dress industrial policy up as climate policy.

The important distinction was between tariffs as protection and tariffs as enforcement. A tariff on imports because they are foreign is one thing. A tariff used as a threat inside a broader agreement that induces countries to adopt comparable climate policies is another. Shapiro’s strongest case for trade policy was not that import restrictions themselves clean the atmosphere. It was that countries may need a stick or carrot to coordinate climate policy when there is no higher jurisdiction to resolve cross-border externalities.

Joseph Shapiro began with the environmental side of the problem. Atmospheric carbon dioxide has continued to rise, and in the most recent year shown on his slide the annual increase was faster than the prior trend. He noted that part of the jump was due to wildfires, but his broader point was not about a single year. Press releases, cheaper renewables, city and state initiatives, and corporate innovation may all be mechanisms of progress, he said, but “there’s only one thing that matters to the atmosphere, and that’s carbon.” On the measure that matters, the rise has continued.

Shapiro contrasted climate with the kind of environmental event that helped mobilize earlier U.S. environmental law: the Cuyahoga River catching fire. Whether or not the iconic image used in public memory came from the precise event being discussed, the river fire was visceral. Climate is not like that. It is a gradual acceleration of conditions, with no single equivalent moment at which a river visibly ignites and politics reorganizes around the image. In California, he said, wildfire smoke in 2018 and 2019 felt outlandish, but the broader problem still accumulates gradually. His conclusion was explicit: he was willing to call it an environmental crisis.

The policy problem is made harder by the physical nature of greenhouse gases. Local air and water pollution largely move with the location of production. If activity shifts from the United States to China, the soot, particulates, or water contamination tend to move with it. Carbon dioxide does not work that way. A ton emitted in Ohio and a ton emitted in Shanghai have the same effect on the atmosphere. The external cost is essentially the same everywhere, so the optimal climate policy should be broadly similar across jurisdictions.

That is not what exists. Shapiro showed a World Bank map of carbon prices around the world in 2025. Many countries had no meaningful price; others had stringent prices; many regulated only some industries or some greenhouse gases. The result, in his account, is not just insufficient climate policy but expensive climate policy. When one jurisdiction regulates carbon and another does not, carbon-intensive activity can shift. Europe reduces emissions at home and imports carbon-intensive goods from China, India, Mexico, Morocco, or elsewhere. One country pushes the mole down; emissions pop up somewhere else. His undergraduates call this “whack-a-mole.” Economists call it leakage.

That leakage problem is central to why trade policy enters the discussion at all. If climate damages are global and policies are local, countries need some way to make one another’s policies stick. Within a country, conflicts over pollution moving downstream can be mediated by state or federal authority. Across countries, there is no equivalent sovereign. Trade policy therefore becomes one of the few available tools: a possible stick, a possible carrot, and a possible source of conflict.

Shapiro’s second crisis was trade policy. For many years, he said, the average import tariff in the United States and Europe, weighted by trade value, was roughly in the range of 1.5 to 2.5 percent. That was the kind of point he had long used in teaching. It now requires revision. Average U.S. tariffs, as shown in his slide, had sharply increased, with a displayed Fitch forecast from April 3, 2025 putting the full-year 2025 average effective tariff rate on U.S. imports at 22 percent.

For Shapiro, the tariff level was only part of the deterioration. Non-tariff barriers also matter: quotas, product standards, and other impediments to cross-border exchange. So does uncertainty. Whatever one believes the optimal trade policy should be, volatility imposes costs on companies, regulators, and trading partners.

The longer institutional story, as he presented it, is that before World War II countries set trade policy largely unilaterally. After World War II, multilateral institutions helped coordinate gradual reductions in protection. Countries agreed to rules and ratcheted down border taxes. That process, Shapiro said, has “pretty quickly unraveled.”

One mechanism of unraveling is the weakened World Trade Organization dispute system. Shapiro described the WTO Appellate Body as the closest thing international trade has to a court system for resolving disputes among countries. Under the Biden administration, he said, the federal government declined to reappoint a judge to the Appellate Body. That may sound technical, but in his account it “took away the teeth” of the WTO’s dispute-resolution process. Without an enforceable set of rules for adjudicating disputed tariffs and other barriers, the cooperation built after World War II becomes more fragile.

The environmental angle is sharpened by the structure of clean-energy supply chains. Shapiro showed a chart indicating that manufacturing capacity for several renewable technologies in 2023 was highly concentrated in China: solar PV modules, wind nacelles, battery cells, electrolyzers, and heat pumps. The recent progress in solar, wind, batteries, and electric vehicles, he said, has been “unprecedented and unfathomable,” with solar finally becoming cheaper than coal in parts of the world after decades of predictions that it was five to ten years away. But the supply chains for these technologies are not broadly diversified. The pandemic made the value of diversified supply chains more visible; renewable technology, in his account, has been moving in the opposite direction.

The result is a collision between two stresses. Climate policy is fragmented in a world where carbon is global. Trade policy is becoming less cooperative in a world where clean technologies depend on international supply chains. Linking the two could improve coordination, or it could accelerate the unraveling of both.

Shapiro’s most direct evidence concerned the environmental bias of existing trade policy. In work he published in 2021, he examined tariffs and non-tariff barriers across countries and industries, then asked whether dirtier industries face more or less trade protection. “Dirty” in the analysis meant higher carbon emissions per dollar of output. Trade protection included both tariffs and the tax-equivalent value of non-tariff barriers.

The pattern he described runs opposite to a climate policy. Fossil fuels, coal, cement, steel, and other carbon-intensive industries tend to face low levels of trade protection. Cleaner industries, with lower carbon emissions per dollar of output, often face high border protection. In the graph he showed, the relationship between non-tariff barriers and carbon intensity sloped downward: cleaner industries were taxed more heavily at the border, dirtier industries less heavily.

Shapiro called this “almost like a negative carbon tax at the border.” It encourages production, consumption, and trade in dirty industries because those industries face low trade barriers. It discourages production, consumption, and trade in cleaner industries because those industries face higher barriers. When he summed the effect globally, he said, it implied a $500 billion global subsidy of greenhouse gas emissions.

The mechanism is not confined to one country. Shapiro said the pattern holds in most countries. He later linked it to the political economy of upstream industries. Economists have begun using the term “upstreamness” for industries that mostly sell to other firms rather than final consumers. Bubble gum is downstream: it is mostly sold to consumers. Hot-rolled steel coils are upstream: consumers do not usually buy them at Walmart or CVS. The most upstream industries tend to be highly carbon intensive — steel, cement, refined petroleum, aluminum, rubber — and they also tend to be inputs into many other sectors. Firms that use those inputs lobby for low protection on them because low input prices reduce costs. That helps explain why many dirty upstream industries face relatively low trade protection.

This point answered, in part, an audience question about whether average tariff measures miss intermediate goods and supply-chain links. Shapiro agreed that the simple average tariff chart from the media did not capture those input-output relationships. But his pollution-intensity analysis did. It traced inputs to goods, inputs to inputs, and so on through the supply chain. The structure of trade protection is therefore not just a matter of consumer-facing goods; it is embedded in production networks.

Shapiro distinguished between two uses of trade policy for climate. One is a tariff proportional to the carbon emitted in traded goods. Europe’s carbon border adjustment mechanism fits that category. The other is a broader trade agreement in which countries commit to a minimum level of climate policy and face tariffs if they renege. William Nordhaus, Shapiro noted, called this idea “climate clubs.”

The contrast in estimated effects was stark. Shapiro cited work by Farid Farrokhi and Ahmad Lashkaripour comparing non-cooperative border taxes with cooperative carbon taxes. Border taxes on carbon embodied in trade were estimated to reduce global carbon dioxide emissions by about 0.6 percent. Cooperative carbon taxes, backed by a trade-agreement enforcement structure, were estimated to reduce global emissions by 61 percent. In the displayed table, the welfare effects also differed: non-cooperative border taxes reduced global welfare by 1.7 percent, while cooperative carbon taxes increased global welfare by 0.4 percent.

The lesson Shapiro drew was not that tariffs are a climate solution by themselves. The limited effect of border taxes suggests the opposite. But tariffs used as an enforcement threat inside a cooperative framework may have substantially greater potential. His final question was therefore deliberately unresolved: will linking trade and environmental policy make both systems better, or will it exacerbate the crisis in each?

That question matters because the same instrument can serve different political purposes. A tariff can be a climate-club enforcement mechanism. It can also be a protectionist measure with environmental language attached. Levinson’s response focused on that risk.

Arik Levinson accepted the setup and then challenged a central premise behind pollution-based import tariffs. He said there is bipartisan enthusiasm for linking trade and climate policy, and he wanted to “throw some shade at that bipartisanship.”

Environmental tariffs, in Levinson’s taxonomy, come in two broad categories. The first taxes imports based on pollution caused in producing them — dirty import tariffs. These are justified in two ways. One is as a narrow form of pollution tax on imported goods. Economists have supported pollution taxes for a century, though tariffs apply only to a slice of the economy. The second and more politically important justification is as a defense against offshoring: the fear that environmental regulation at home causes firms to relocate abroad and then export back to the regulated market.

That concern has been around for decades. Levinson pointed to a 1991 Senate bill, the International Pollution Deterrence Act, that would have charged tariffs equal to the environmental compliance costs foreign manufacturers would have faced under U.S. standards. He described it as a precursor to Europe’s carbon border adjustment mechanism, which taxes imports based on what those imports would have paid under European carbon prices. He also cited a 2025 U.S. Senate proposal, the Foreign Pollution Fee Act, associated with Senators Graham and Cassidy, that would charge tariffs based on the carbon content of imports even though the United States has no domestic carbon price.

A New York Times headline shown on Levinson’s slide asked: “You’ve Heard of Outsourced Jobs, but Outsourced Pollution? It’s Real, and Tough to Tally Up.” Levinson’s reply was blunt: “I think the New York Times is wrong on both counts.” In his view, pollution embodied in trade can be tallied, and when it is, outsourced pollution is not a major explanation for rich-country cleanup.

He used data for the 24 OECD countries as of 1993 — essentially North America, Europe, Japan, Australia, New Zealand, and other rich economies. From 1990 to 2018, their real inflation-adjusted GDP doubled. Over the same period, particulate pollution measured as PM10 declined slightly, and carbon dioxide emissions remained relatively flat. The tempting explanation is imports: perhaps rich countries simply stopped producing dirty goods and imported them instead.

Trade flows seemed to support that at first. Imports from the rest of the world into those OECD countries increased by a factor of 10. Imports from China rose sharply. But exports also increased by a factor of 10. The question then becomes whether the composition of imports became dirtier while exports became cleaner.

Levinson’s method was to take more than 300 industries, estimate the pollution emitted per dollar of production in each, multiply those intensities by the dollar value of trade in each industry, and sum the result. For PM10, pollution embodied in imports rose by a factor of four. That might seem to explain part of the OECD cleanup. But pollution embodied in exports also rose by a factor of four. For carbon dioxide, he said, the same pattern held: imports rose a lot, but exports rose even more in terms of embodied pollution.

The crucial comparison was pollution intensity per dollar of trade. When the embodied pollution measures are divided by total trade volume, PM10 intensity in imports, exports, and imports from China all declined at approximately the same rate. The same was effectively true for carbon dioxide. If rich countries were outsourcing pollution in the way the popular story suggests, Levinson said, one would expect pollution intensity in exports to fall while pollution intensity in imports rose. If China were the main channel, imports from China should rise even faster in pollution intensity. That is not what the charts showed.

His conclusion was narrower than saying trade has no environmental effects. It was that offshoring pollution is not a large enough empirical fact to carry the political weight being put on it. Pollution is not especially hard to tally, and in his data it is “not a big deal.”

Levinson then turned to the second category of environmental tariffs: tariffs on clean goods such as batteries, solar panels, and electric vehicles. Both the Biden and Trump administrations, he said, had been enthusiastic about these. His question was simple: why tax goods that help reduce pollution?

He listed five rationales commonly offered for tariffs on clean goods: the United States is large; the China shock; national security; learning by doing; and the claim that China is cheating. He focused on several.

The “United States is large” argument is a terms-of-trade claim. If the United States is a large consumer, tariffs can drive down world prices and benefit U.S. consumers. Levinson said that is fine in theory. But the United States accounts for only 8 percent of global solar panel installations and 10 percent of electric vehicles. Other countries may retaliate. And even if the tariff lowers prices for some U.S. buyers, shrinking global demand reduces global sales, which can increase carbon intensity for everyone. He called that a “cut off your nose to spite your face” strategy.

The China-shock analogy also failed, in his view. During the Biden administration, it became common to argue that tariffs could prevent a second China shock like the disruption associated with China’s accession to the World Trade Organization in the early 2000s. But before China’s accession, the industries most exposed to import competition — apparel, textiles, furniture, electronics — accounted for about 20 percent of U.S. manufacturing jobs. U.S. solar panel manufacturing today, Levinson said, accounts for 0.3 percent of U.S. manufacturing jobs. That makes it unlikely that solar-panel imports could roil the U.S. economy in the same way.

The choice between subsidies and tariffs is especially important when the protected sector has climate benefits. For an industry with no climate externality, Levinson said, protecting domestic producers through subsidies or tariffs mainly changes who pays: taxpayers pay for subsidies; consumers pay for tariffs. But for clean technologies, the environmental consequences differ. Subsidies increase global production, which helps the climate objective. Tariffs shrink global production, which works against it.

National security arguments require sharper distinctions. If the fear is that China could embed disruptive bugs, flaws, or back doors in solar panels or electric vehicles, then Levinson argued that taxes are the wrong tool. Dangerous goods should be banned, not taxed. The United States does not tax cars without brakes or seat belts; it bans them. It does not tax goods contaminated with E. coli; it prohibits them.

The second national security fear is an embargo. China could withhold solar panels the way OPEC withheld oil in the 1970s. Levinson rejected the analogy. Oil is a fuel used continuously by durable technologies to provide transport and power. If the supply stops, cars stop driving. Solar panels are themselves durable technologies that use a free fuel — sunlight. If China stops selling solar panels, already-installed panels keep generating electricity. “OPEC can’t embargo the sunshine,” he said.

Levinson’s bottom line separated the two tariff categories. Dirty import tariffs may be defensible in theory, even if offshoring is not as large a problem as claimed. They are not true carbon taxes, because they touch only a small part of the economy, but they have a carbon-tax flavor. Tariffs on clean solutions to pollution are much harder to justify. The danger, he said, is that environmental concern becomes a pretext for protectionism.

Cochrane used the moderation period to sharpen the cost-benefit critique. On carbon dioxide, he said, China and India are the central issue; U.S. consumer choices are not decisive by comparison. If trade policy is supposed to matter, then its real function would be to pressure those countries to change domestic policy, not merely to alter the carbon embodied in goods exported to the United States.

He strongly endorsed one answer: “get out of the way.” If the question is climate, he argued, the origin of a clean technology should not matter. The climate does not care whether an electric car is made in China and sold in the United States for $15,000 or made domestically with union labor, American parts, regulatory compliance, lawyers, and subsidies at far higher cost. If the policy objective is government-provided jobs, that is a different objective.

Cochrane then named the economist’s preferred instrument: a carbon tax with border adjustment and nothing else. No subsidies, no industrial policy, no mandates. In his words, “that’s the economist deal.” But he also noted that voters and politicians do not seem to want it. Neither right nor left embraces it when presented directly, he said; voters prefer boondoggles to a border-adjusted carbon tax. He did not see either the Biden or Trump administration moving toward that package.

His examples of cost-benefit failure were pointed. He cited high-speed rail as “$230 billion down a rat hole” that he believed would produce more carbon than it saves. He cited corn ethanol as a policy that, once protected by industrial policy, could not be reversed even when it proved useless. He warned that current enthusiasms for subsidies should be judged against past enthusiasms that became entrenched.

The deepest disagreement was over the word “crisis.” Cochrane pressed Shapiro on whether rising carbon dioxide justified crisis framing. He invoked Steven Koonin and noted that the annual increase Shapiro had discussed was measured in parts per million against an atmospheric level around 400 parts per million. He also disputed linking wildfires to climate in the way the public debate often does, arguing that wildfires are declining worldwide, reflect forest management failures, and do not contribute carbon dioxide on a 50- to 100-year climate timescale because forests regrow and absorb carbon.

Cochrane’s concern was that crisis language causes policymakers to forget costs and benefits. If the problem is a loss of 5 percent of GDP in 100 years, and GDP will grow two- or threefold over that time anyway, he argued, that does not justify trillions of dollars of mandates today. He also argued that climate has crowded out other environmental problems: chemical pollution, lack of clean water, indoor air pollution from burning dung, species extinction, ocean pollution, and possibly microplastics. Some of those, he said, he would be more willing to call crises.

Shapiro accepted much of the optimal-policy critique. He agreed that a carbon tax plus border adjustment is central, but added a subsidy for research and development because innovation is a public good and underprovided by markets. If every country adopted the same climate policy, he said, border adjustment would not be needed. The border adjustment exists because policies are heterogeneous.

He also described new survey work. In a representative sample of Americans using the same platform as the Associated Press, he and coauthors presented cost-effective policies such as pollution pricing and expensive, less cost-effective policies such as standards. Voters preferred standards and disliked pollution pricing. Environmental economists showed the opposite preferences. But after respondents watched three-minute videos explaining economic principles such as pass-through and allocative efficiency, their preferences moved toward the economists’ view. Shapiro’s conclusion was that policymakers may be reflecting voter preferences, and that voter preferences can change when economic principles are explained clearly.

On “crisis,” Shapiro’s response returned to political economy. The term was partly a contrast with the Clean Water Act story and the galvanizing image of a burning river. Climate lacks that focal event. The costs will occur over a century, while mitigation requires decisions over the next 5, 10, 20, or 30 years. The policy challenge is how to generate investment now for benefits that arrive later, especially when citizens respond more readily to visible environmental emergencies.

Koonin intervened from the audience as a physical scientist. His “touchstone,” he said, is data. He cited a paper from the prior year or two, in either Science or Nature, that examined more than a thousand mitigation policies across more than 60 countries and found that only 5 percent had any detectable effect on emissions. He called that sobering and asked for comment.

Levinson did not read the finding as an argument against climate policy. He pointed to two complications. First, many existing policies have been relabeled as climate policies. Fuel economy standards existed for decades before being described primarily as climate measures. The Clean Air Act was not designed for climate, nor even for international pollution; successive Democratic administrations tried to shoehorn climate policy into it. Second, overlapping policies make evaluation difficult. California’s climate policy includes cap and trade, fuel economy standards, and interactions with federal clean-power subsidies. If the cap-and-trade permit price is low, that may mean the policy failed, or it may mean other overlapping policies reduced the work the cap needed to do.

Shapiro’s response was that many policies in every domain are ineffective, which is part of the appeal of pollution pricing. A positive price in a cap-and-trade market indicates that firms are optimizing along some margin. A standard mandates behavior, but “all bets are off the table” about whether it is doing so cost-effectively. He also said his graph of rising carbon was meant to show that global emissions have not deviated much from trend over the last half-century. The trend in global carbon emissions per dollar of GDP is close to a straight line. That does not mean no policy has any effect; it means the aggregate effect has been modest relative to the scale of the problem.

Cochrane emphasized Koonin’s point more aggressively. He described current carbon policy as a disaster in cost-benefit terms. Nuclear power had been curtailed under other environmental regimes; had the United States followed something closer to France’s path, he argued, there would be much less talk about carbon. Europe, he said, had largely stopped growing after 2010 in part because of energy prices, imposing trillions in costs for little measurable climate benefit. Fuel economy standards were “jiggered” in ways that encouraged heavier vehicles such as Ford F-150s. Mass transit subsidies and high-speed rail, in his view, had consumed large sums with little carbon effect.

Koonin later added a scientific objection to how emissions are commonly discussed. Human influence on climate, he said, depends on the concentration of carbon dioxide in the atmosphere — more precisely the logarithm of the concentration — not directly on emissions or changes in emissions. He would have preferred to see the contribution of CO2 to radiative forcing, or total radiative forcing, rather than the annual increase in parts per million. Annual increases near three parts per million should be viewed against an atmospheric level around 400 parts per million, he said; then one should take the logarithm.

That exchange left an important distinction intact. Shapiro’s policy argument was driven by the need to act before the worst damages arrive and by the difficulty of international coordination. Cochrane and Koonin pushed on measurement, magnitude, and cost-benefit discipline. Levinson and Shapiro both agreed that pricing is superior to the patchwork of mandates and standards, but neither treated the poor performance of existing policy as a reason to abandon policy altogether.

Levinson raised a legitimacy concern about using tariff policy to pressure China. On human rights, he said, the United States might claim moral standing because it eventually abolished slavery and Jim Crow before lecturing others. On climate, the analogy is weaker. The United States does not have a domestic carbon tax. It is historically responsible for much of the carbon in the atmosphere, even if that stock is now a sunk cost. To tell China that it is “the real problem” and impose tariffs for insufficient carbon policy, while the United States lacks its own comparable policy, “just feels wrong,” he said. There may be a demonstration benefit to cleaning up domestically, even if the United States is no longer the sole or largest current source of emissions.

That point sits uneasily beside the economist’s ideal policy. A border-adjusted carbon tax is attractive because it treats domestic and foreign carbon symmetrically: tax carbon at home, adjust at the border, avoid leakage. But a border tariff without a domestic carbon price looks different. It can look like climate policy aimed outward and industrial policy aimed inward. That was Levinson’s warning about pretext.

Cochrane tried to reframe public acceptance by suggesting that salable carbon rights might be more politically attractive than taxes: give everyone a right to emit a certain amount of carbon and allow them to sell it. Shapiro’s survey result suggested that presentation matters, but only up to a point. Levinson noted that even after Shapiro’s educational intervention moved respondents toward market-based policies, people still resisted the idea that taxing consumers and taxing producers are economically equivalent. Voters may learn some economic principles quickly, but certain tax-incidence intuitions remain hard to dislodge.

The discussion therefore ended with a practical tension rather than a unified policy program. The economists mostly agreed on the efficient core: price pollution, adjust at the border if other countries do not price it, support R&D, and avoid mandates and protectionist industrial policy where possible. They also agreed that current politics does not naturally deliver that bundle. Voters prefer standards. Legislators prefer visible subsidies, local jobs, and sectoral protection. Agencies retrofit old statutes to new goals. Tariffs can be sold simultaneously as climate policy, industrial policy, China policy, and labor-market policy.

That multiplicity is precisely why tariffs are tempting and dangerous. If used to enforce a cooperative climate agreement, they may help solve a global coordination problem. If used to tax clean technologies, they can slow deployment of the very goods that reduce emissions. If used to punish dirty imports without a domestic carbon price, they invite charges of hypocrisy and protectionism. If defended as a response to outsourced pollution, they may rest on an empirical story that Levinson’s data do not support.