Who Gets to Set a Tariff: On Article I, Ninety Years of Delegation, and What Leverage Actually Costs

There are two separate questions buried inside every argument about American trade policy, and almost nobody separates them. The first is who has the constitutional authority to set a tariff. The second is whether any particular tariff is a good idea. These are genuinely independent questions. A tariff can be excellent policy imposed by the wrong branch of government, or terrible policy imposed with impeccable legal authority, and the fact that most people's answer to the first question tracks perfectly with their answer to the second should tell you that very little actual constitutional reasoning is happening in public. On February 20th of this year, the Supreme Court answered the first question in a way that deserves more attention than it received, and it answered it with a coalition that scrambles the usual alignments: Chief Justice Roberts writing, joined on the core holding by Sotomayor, Kagan, Gorsuch, Barrett, and Jackson, with Thomas and Kavanaugh in dissent. Six to three, with three of the Court's most textualist members joining all three of its liberals. That is not an ideological ruling. That is a structural one, and the structure it addresses is a problem that has been building since 1934.

The case was Learning Resources, Inc. v. Trump, consolidated with Trump v. V.O.S. Selections, and the question was whether the International Emergency Economic Powers Act authorizes the President to impose tariffs. The Court held that it does not. Roberts began where the analysis has to begin, with Article I, Section 8, which vests in Congress the power to lay and collect taxes, duties, imposts, and excises. Not in the President. Not in some shared arrangement. In Congress, exclusively, and the Framers put it there deliberately. The government conceded, correctly, that the President has no inherent peacetime authority to impose tariffs, which meant the entire program rested on the proposition that Congress had delegated that authority through IEEPA's language authorizing the President to "regulate importation." Roberts found that reading implausible on its face. IEEPA never uses the words tariff or duty. It contains no mechanism for determining rates, no limit on duration, no procedural requirements, no notice or comment provision. It does not read like any other tariff statute Congress has written. And when Congress has actually delegated tariff authority, Roberts noted, it has done so in explicit terms with strict limits, which is exactly what Sections 232, 301, 201, 122, and 338 all demonstrate. Those statutes name the power, bound it, and specify how it may be exercised. IEEPA does none of that, which is strong evidence that Congress was not silently handing over the taxing power in a statute about emergency asset freezes.

The opinion is worth reading in full, but one line does most of the work: "There is no exception to the major questions doctrine for emergency statutes. Nor does the fact that tariffs implicate foreign affairs render the doctrine inapplicable. The Framers gave Congress alone the power to impose tariffs during peacetime." The government had argued, and Kavanaugh's dissent echoed, that IEEPA is a statute designed to address the most major of major questions and should therefore receive more deference rather than less. Roberts rejected this directly: there is no major questions exception to the major questions doctrine. Gorsuch wrote separately to defend the doctrine as a structural safeguard against executive aggrandizement, framing it as protection for the constitutional assignment of lawmaking power rather than as a policy preference. Barrett concurred but grounded the result in ordinary textualism rather than a freestanding clear statement rule. Kagan, joined by Sotomayor and Jackson, agreed the tariffs were unauthorized but thought the major questions doctrine unnecessary, since conventional statutory interpretation gets you there: "regulate importation" does not naturally mean "tax importation," and Congress has demonstrated through other statutes that it knows how to say the latter when it means it.

What makes this ruling significant is not that it constrained a particular president. It is that it interrupted a ninety-year trend. To understand why that matters, you have to understand how Congress ended up in the position of having almost no practical role in setting tariffs despite holding the exclusive constitutional power to do so, and that story starts with the Tariff Act of 1930, better known as Smoot-Hawley.

The conventional account of Smoot-Hawley is that it caused the Great Depression, and that account is wrong. The Depression was already underway when Hoover signed the bill on June 17, 1930, eight months after the October 1929 crash, and its primary causes were monetary: the contraction of the money supply, the Federal Reserve's failure to act as lender of last resort during successive banking panics, and the deflationary constraint of the gold standard. International trade accounted for less than ten percent of the American economy at the time, which mathematically caps how much damage any trade policy could do to aggregate output. The National Bureau of Economic Research analysis concluded the tariffs had a relatively minor effect on overall U.S. GDP. Anyone still claiming Smoot-Hawley caused the Depression is repeating a simplification that economic historians abandoned decades ago, and the White House was right to push back on the comparison this week when the administration invoked Section 338 of that same act against Canada.

But the reason Smoot-Hawley matters institutionally has almost nothing to do with the Depression. It has to do with how the law was made. The bill started as a Republican effort to win the farm vote in 1928 by raising duties on agricultural imports, addressing a real problem: American farmers had been in distress since European agriculture recovered from the war and global commodity prices collapsed. That is a legitimate policy concern with a defensible remedy. But once the bill entered the legislative process, it became what Douglas Irwin, whose Peddling Protectionism remains the definitive account, describes as a logrolling free-for-all. Every member with a constituent industry wanted protection for that industry, and inclusion was the price of their vote. The final bill covered more than twenty thousand imported goods. The average duty on dutiable imports rose roughly twenty percent, reaching nearly sixty percent on some categories. Nobody designed that outcome. It was the emergent product of a thousand individual bargains, each rational for the member making it and collectively producing a statute that served no coherent national purpose.

The retaliation followed, as retaliation generally does. Within two years roughly two dozen countries had protested or imposed countervailing duties. Canada raised tariffs on American goods and reoriented trade toward the United Kingdom, a shift that persisted for decades. Between 1929 and 1932, American imports of European goods fell more than seventy percent and American exports to Europe fell more than sixty-five percent. World trade declined more than sixty percent between 1929 and 1934, with economists attributing roughly twenty to twenty-five percent of that decline directly to Hawley-Smoot and the responses it provoked. American automobile manufacturing was among the hardest hit: foreign purchases of American-made cars fell roughly forty-six percent. The farmers the bill was written to protect were devastated, because they depended on export markets and those markets closed.

The political consequences were severe and swift. Voters handed Congress to the Democrats in 1932. Both Reed Smoot and Willis Hawley lost their seats. And then Congress did something that has shaped American trade policy ever since: it decided the problem was not the tariff levels but the process that produced them. The Reciprocal Trade Agreements Act of 1934 restructured the entire mechanism. Rather than Congress setting individual rates line by line, which had proven catastrophically vulnerable to logrolling, Congress delegated to the President the authority to negotiate reciprocal reductions with trading partners, subject to periodic renewal. Roosevelt used it to conclude agreements with nineteen countries between 1934 and 1939, and the RTAA framework became the institutional foundation for GATT and eventually the WTO.

Understand what happened there, because it is the actual lesson of Smoot-Hawley and it is not the one usually drawn. Congress looked at a power the Constitution assigned it exclusively, concluded that it could not exercise that power responsibly because of its own institutional dynamics, and gave the power away. Not because the executive had seized it. Because the legislature did not want it. Every subsequent delegation followed the same logic and deepened the same pattern: Section 232 in 1962, Sections 122, 201, and 301 in 1974, each handing the President another instrument, each justified as necessary flexibility in a fast-moving world, each further hollowing out the practical meaning of Article I, Section 8. By 2025, the accumulated delegations were extensive enough that an administration could construct a tariff regime covering nearly all American trade without Congress voting on any of it. That is a legislative abdication ninety years in the making, and it is not primarily a story about any particular president.

Which is why the Court's ruling is more interesting as a constitutional development than as a trade policy development. Roberts was not saying tariffs are bad. He was saying that when Congress delegates the taxing power, it has to do so explicitly, with limits, in language that acknowledges what it is giving away. The opinion pointedly cited Sections 232, 301, 201, 122, and 338 as examples of Congress doing it correctly. Those delegations remain fully intact. Section 232, which authorizes the President to adjust imports based on national security findings, has no statutory time limit and requires a Commerce Department investigation and procedural reporting to Congress. The 2018 steel and aluminum tariffs imposed under it are still substantially in force, having survived legal challenge and a change of administration. Section 301, which addresses unfair foreign trade practices, requires a USTR investigation, public comment, and formal findings. These are durable authorities precisely because Congress wrote them clearly. The Court did not narrow presidential tariff power in general. It declined to invent a new delegation in a statute that did not contain one, which leaves the real ones untouched.

The administration's response demonstrated exactly how much authority remains. On the same day the ruling came down, a ten percent global tariff was imposed under Section 122 of the Trade Act of 1974, which permits the President to respond to large and serious balance-of-payments deficits for up to 150 days before congressional action is required. That clock expires today. USTR had already, in March, initiated multicountry Section 301 investigations covering sixty trading partners on the theory that they have failed to effectively prohibit imports of goods produced with forced labor, plus a parallel excess capacity investigation covering sixteen partners. Public comments ran through April 15th, hearings were held May 5th through 8th, and the resulting tariffs of ten to 12.5 percent take effect just after midnight tonight, covering 99.4 percent of American trade. USTR Jamieson Greer has described Sections 232 and 301 as very durable tools that can remain in place as long as needed, which is accurate. The Tax Policy Center estimates the new structure brings the average tariff rate to 10.1 percent and raises $706 billion over an eleven-year window.

So the executive retains substantial tariff authority, and the practical effect of the ruling was to route policy through statutes with investigations, comment periods, and findings requirements rather than through emergency declarations. Whether you think that is good depends on whether you think process constraints on executive action are valuable in general, which is a question worth answering the same way regardless of who currently occupies the office. There is a version of conservatism that has always answered yes, and Gorsuch's concurrence is squarely in that tradition.

Now to the second question, which the constitutional analysis leaves entirely open: are these tariffs good policy? Here I want to give the strongest version of the case, because it is a real case that gets caricatured by people who learned one supply-and-demand diagram in college and think the matter is settled.

The strategic argument for tariffs rests on several distinct foundations. The first is that free trade theory assumes adjustment costs that empirically do not materialize. The China Shock literature by Autor, Dorn, and Hanson demonstrated that specific American communities experienced durable economic devastation from import competition, and that the labor reallocation the standard models assumed would smooth things out simply did not happen. Workers did not move. Wages did not recover. Communities hollowed out and stayed hollow for twenty years. The aggregate gains from trade were real, but they accrued diffusely to consumers in the form of cheaper goods while the losses concentrated geographically and generationally. Economists spent two decades dismissing this concern before the data forced them to take it seriously, and that intellectual failure did real damage to the credibility of the profession on trade questions. Anyone arguing about tariffs today who has not internalized that history is arguing from a position that was falsified.

The second foundation is supply chain resilience in strategically critical sectors. The pandemic exposed how thoroughly American access to pharmaceuticals, medical equipment, and their chemical precursors depended on a small number of foreign suppliers, several of them in a country that is a systemic strategic competitor. Semiconductors present the same problem with higher stakes: the concentration of advanced logic fabrication in Taiwan is a genuine national vulnerability that no market mechanism will correct, because the market correctly prices efficiency and does not price the tail risk of a blockade. Rare earth processing is similarly concentrated. In these sectors, accepting economic inefficiency in exchange for redundancy is not protectionism dressed up as security. It is security, and the case is strong enough that the previous administration reached largely the same conclusion through the CHIPS Act, which was industrial policy by another name.

The third foundation is leverage. Tariffs impose costs on trading partners, which creates bargaining power to extract concessions on market access, intellectual property enforcement, currency practices, and non-tariff barriers that have historically disadvantaged American exporters. The complaint that other countries maintain higher barriers against American goods than we maintain against theirs is factually accurate in many cases. The complaint that the WTO dispute mechanism proved slow and ineffective against a determined violator is also accurate. If the existing enforcement architecture does not work, and it demonstrably did not work against Chinese forced technology transfer and intellectual property appropriation through the 2000s and 2010s, then unilateral pressure is not obviously worse than continued ineffective multilateralism. The 2018 Section 301 action against China was grounded in findings about exactly those practices, and the subsequent administration kept those tariffs in place, which suggests a bipartisan judgment that the underlying diagnosis was correct.

These are serious arguments. What they justify, though, is narrow, targeted, sustained intervention in specific sectors with clear success criteria, or time-limited pressure with defined objectives that gets withdrawn when the objectives are met. And the empirical question worth asking is whether the current structure functions that way, because leverage has a price and the price is measurable.

The mechanics of who pays are where public discussion goes wrong most consistently. A tariff is collected by Customs and Border Protection from the importer of record, which is an American company. The check is written domestically. What happens next depends on incidence: the importer can compress margins, raise prices to customers, push the cost back to the foreign supplier by demanding a lower pre-tariff price, or restructure sourcing entirely. Firms do some combination, and the proportions determine who actually bears the burden. The evidence from the current round is fairly clear. Harvard Kennedy School economists found pass-through to U.S. import prices was near one hundred percent, meaning foreign exporters did not meaningfully cut their prices to absorb the tariff. The Federal Reserve's April 2026 analysis estimated that tariffs implemented through November 2025 raised core goods PCE prices by 3.1 percent through February 2026 and contributed 0.8 percent to core PCE overall, with pass-through effectively complete. A New York Fed survey found roughly three-quarters of affected firms raised prices while about a quarter absorbed the cost through margins. The Tax Policy Center puts the average household burden at about $960 for calendar 2026.

That is the cost of the leverage. It is not an argument that the leverage is worthless, but it is the number that has to be weighed against whatever the leverage buys, and it should be stated plainly rather than denied. The distributional profile is worth noting for the same reason: because tariffs function as consumption taxes, they fall harder on households that spend a larger share of income. The average federal tax rate rose 0.8 percentage points for the bottom quintile against 0.6 points for the top. Whether that tradeoff is acceptable depends on what you get for it, but a policy defended partly on behalf of working-class Americans should be honest that working-class Americans are paying for it up front.

The input cost problem is the part I find most analytically important, and it is underdiscussed on all sides. Roughly half of American imports are not finished consumer goods. They are intermediate inputs into domestic production: steel that becomes American cars, semiconductors that become American appliances, chemicals that become American pharmaceuticals, aluminum that becomes American aircraft. Taxing those inputs raises the cost structure of every domestic manufacturer that uses them, which makes American manufactured goods less competitive both domestically and in export markets. Steel and aluminum currently carry the highest effective rate of any product category at 41.2 percent. The protected constituency is the several thousand people employed in primary steel production. The taxed constituency is the several million employed in industries that buy steel. Manufacturing employment fell 68,000 jobs over 2025 according to Stanford's Institute for Economic Policy Research, which is the outcome the policy was substantially designed to reverse. This is not evidence that protection can never work. It is evidence that broad protection of upstream inputs works against downstream manufacturing, which is a well-understood mechanism and one that careful sectoral design can avoid.

There is a second cost that does not show up in price indices, and having spent my career watching how uncertainty gets priced into decisions, I think it is the larger one. Consider a mid-sized manufacturer weighing a hundred-million-dollar plant with a twenty-year payback. That investment's viability depends on input costs, which depend on tariff rates, which have changed repeatedly over eighteen months, were retroactively invalidated by the Supreme Court in February, were replaced the same day by a different authority with a 150-day expiration, and are being replaced again tonight under a third authority. The rational response to that variance is not to build the plant. It is to wait. What suppresses investment is not the level of the tariff but the width of the distribution of possible future tariffs, and firms hold capital in short-duration instruments when the distribution is wide. Brookings identified this as the central problem in its assessment of the ruling, and it is the strongest practical argument for exactly what the Court's decision pushes toward: policy made through statutes with findings requirements and procedural steps, which is slower but produces rates that firms can underwrite against.

You can already see markets routing around the current structure. The share of Canadian and Mexican imports claiming USMCA exemption stood stable through late 2024 and then surged to 83.8 percent by May 2026 as importers aggressively worked rules of origin to secure duty-free treatment. That is not compliance. That is arbitrage, and it is what tariff regimes reliably produce when the exemption architecture is more complex than the enforcement capacity. The revenue is real: $194.8 billion in inflation-adjusted customs revenue above the 2022 to 2024 average as of January 2026, with the effective rate at 9.9 percent in December 2025 and 7.2 percent by May 2026 following the Court's ruling. Given the fiscal trajectory I wrote about in March, revenue is not nothing. But raising it through a regressive levy administered by Customs, with an exemption process that the sophisticated navigate and the unsophisticated do not, is a strange design if revenue is the actual objective.

So where does this go? Three paths seem plausible, and they are distinguishable by observable signatures rather than by rhetoric.

The first is congressional codification, which is what the Court's reasoning invites. Roberts spent considerable effort explaining that Congress has repeatedly and successfully delegated tariff authority when it intended to, which functions as a roadmap. A statutory regime would end the whiplash and let firms underwrite against known rates, capturing most of the strategic benefit while eliminating the uncertainty cost. There is already legislative movement in this direction, though currently aimed at constraining rather than codifying: bills requiring ITC impact assessments, Defense Department justification, congressional notification, and fast-tracked disapproval votes for Section 232 and 301 actions. The obstacle to codification is that it forces individual legislators to own price increases in their districts, and attribution is direct in a way it never is with executive action. That is precisely the dynamic that produced Smoot-Hawley's logrolling and precisely why Congress delegated the power away in 1934. Whether the institution has changed enough in ninety years to handle it responsibly is an open question, and I would not bet heavily on yes.

The second is negotiated de-escalation, in which the tariffs function as designed, extract concessions, and get withdrawn. This is the path the strategic case actually describes, and it is not implausible. The administration has demonstrated willingness to remove tariffs when costs become acute, as with the coffee and cocoa exclusions in November 2025. Trade threats have historically produced real concessions. The difficulty is that leverage requires a counterparty who believes your commitments will hold, and eighteen months of legal instability has degraded that belief. There is also the awkward fact that the fastest instrument for imposing broad country-level duties is gone, and the replacements are narrower, slower, and procedurally constrained, which weakens the threat that makes the leverage work. You cannot credibly threaten with an instrument that requires a nine-month investigation.

The third, and to my eye the most likely, is gradual erosion through the exemption process. This is roughly how the 2018 steel tariffs softened: not repeal but an exclusion process extensive enough to hollow out the policy from the inside. The USMCA exemption surge is that mechanism operating in real time. Add firm-specific carve-outs, politically sensitive product exclusions, continued litigation at the Court of International Trade, and the administrative burden of running multiple overlapping regimes under different statutes with different expirations and different procedural requirements, and the system tends toward entropy. Tariff regimes are rarely defeated. They decay, because the concentrated interests seeking relief are motivated and well-organized while the diffuse interest in maintaining the policy is not.

The thing I keep returning to is that the strongest critique of the current approach is not that protection never works. Sometimes it does, under specific conditions, in specific sectors, with careful design. The critique is that the current structure has no falsifiable success criteria. If manufacturing employment rises, the tariffs worked. If it falls, they need more time. If prices rise, foreign exporters are being stubborn. If the trade deficit does not move, the rates were too low. A policy with no outcome that would count as disconfirmation is not being managed as a policy. It is being held as a position, and positions without exit criteria do not get closed when the thesis fails. Anyone who has run money understands why that is the dangerous configuration: not being wrong, which is normal and survivable, but having constructed a frame in which you cannot discover that you are wrong.

The constitutional question and the policy question converge at exactly that point. What Congress does, when it works, is force the articulation of objectives, the assessment of costs, and the periodic reauthorization that makes abandonment possible. It is slow and it is ugly and it produced Smoot-Hawley when it went wrong. But the alternative, ninety years of delegation to an executive that can construct a trade regime by proclamation and reconstruct it under a different statute when a court objects, removes the mechanism by which a country notices that a policy is not working and stops. The Court did not resolve the tariff debate in February. It returned a question to the body the Constitution assigned it to, which is the body that has spent ninety years declining to answer. Whether Congress picks it up is the thing worth watching, and it will tell you considerably more about the next decade of American trade policy than any individual rate schedule will.

References

Supreme Court of the United States. Learning Resources, Inc. v. Trump, No. 24-1287, consolidated with Trump v. V.O.S. Selections, Inc. Decided February 20, 2026.

Congressional Research Service. "Congressional and Presidential Authority to Impose Import Tariffs." Report R48435, May 2026.

Congressional Research Service. "Supreme Court Rules Against Tariffs Imposed Under the International Emergency Economic Powers Act." Legal Sidebar LSB11398, 2026.

Congressional Research Service. "Legal Authority for Section 301 Tariffs to Address Forced Labor and Excess Manufacturing Capacity." Legal Sidebar LSB11460, July 2026.

Congressional Research Service. "Section 301 of the Trade Act of 1974." In Focus IF11346, updated 2026.

Arnold & Porter. "Supreme Court's Tariffs Ruling Answers Some Major Questions, Leaves Others Open." February 2026.

Holland & Knight. "Supreme Court Strikes Down IEEPA Tariffs: What Importers Need to Know Now." February 2026.

Brownstein Hyatt Farber Schreck. "The Supreme Court Held IEEPA Tariffs Are Invalid: Legal Analysis and What's Next for Businesses." February 2026.

Global Trade Alert. "After IEEPA: How Sections 232 and 301 Work." February 2026.

Board of Governors of the Federal Reserve System. "Detecting Tariff Effects on Consumer Prices in Real Time, Part II." FEDS Notes, April 8, 2026.

The Budget Lab at Yale. "State of U.S. Tariffs: April 8, 2026" and "Tracking the Economic Effects of Tariffs," updated March 2026.

Penn Wharton Budget Model. "Effective Tariff Rates and Revenues." Updated July 13, 2026.

Tax Policy Center. "Tracking Trump Tariffs." Updated July 2026.

Tax Foundation. "Tariff Tracker: 2026 Trump Tariffs and Trade War by the Numbers." 2026.

Brookings Institution. "Brookings experts on the Supreme Court's tariff decision." February 25, 2026.

Council on Foreign Relations. "Tariffs and Trade: What Comes Next." February 23, 2026.

National Taxpayers Union. "Reclaiming Trade Authority: Members of Congress Introduce Reforms to Rein in Presidential Tariffs." May 2026.

Axios. "Trump revives unused Smoot-Hawley tariff power against Canada." July 21, 2026.

CNBC. "Trump to slap sweeping new tariffs on 60 trade partners as global duties expire." July 23, 2026.

Irwin, Douglas A. Peddling Protectionism: Smoot-Hawley and the Great Depression. Princeton University Press, 2011.

Autor, David H., David Dorn, and Gordon H. Hanson. "The China Shock: Learning from Labor Market Adjustment to Large Changes in Trade." NBER, 2016.

National Bureau of Economic Research. Working Paper 28616 on the Smoot-Hawley Tariff and U.S. manufacturing.

Stanford Institute for Economic Policy Research. "The U.S. economy in 2026: What to watch for." 2026.

U.S. Department of State, Office of the Historian. "Protectionism in the Interwar Period."