The Yale Law Journal


Presidential Power to Terminate International Agreements

12 Nov 2018

abstract. Could President Trump unilaterally remove the United States tomorrow from all of the thousands of international agreements to which the United States is currently a party? Common sense would suggest no, but the conventional wisdom among legal academics has leaned the other way. This Essay argues that the conventional wisdom is wrong: the Constitution affords the President no general unilateral power to terminate or withdraw from any international agreement, without regard to its subject matter. Neither historical practice nor Supreme Court precedent dictates that conclusion, nor does the Court’s misunderstood nonjusticiability holding forty years ago in Goldwater v. Carter. Constitutional, functional, and comparative-law considerations all cut the other way. Instead of a blanket unilateral power of presidential termination, this Essay suggests that the Constitution requires a “mirror principle,” whereby the degree of legislative approval needed to exit an international agreement must parallel the degree of legislative approval originally required to enter it. Such a mirror principle makes the degree of legislative approval required to enter or exit any particular agreement “substance dependent,” turning on which branch of government has substantive constitutional prerogatives to make law in any particular area of foreign policy. The Essay concludes by suggesting better foreign policy mechanisms, more reflective of modern realities, to guide America’s process of agreement unmaking in the future.


Could Donald Trump unilaterally withdraw the United States from the United Nations, the International Monetary Fund, the World Bank, and other major longstanding treaties and international organizations?

These scenarios are neither unforeseeable nor hypothetical. Less than four decades ago, a D.C. Circuit judge warned against the risk of “an ambitious or unreasoned President disengaging the United States from crucial bilateral and multilateral treaties with the stroke of a pen.”1Since 2017, the Trump Administration has announced its withdrawal from a host of bilateral and multilateral arrangements, including the Paris Climate Agreement; the Joint Comprehensive Plan of Action (JCPOA); the U.N. Educational, Scientific, and Cultural Organization; the Global Compact on Migration; the U.N. Human Rights Council; the Trans-Pacific Partnership (TPP); the 1955 Treaty of Amity, Economic Relations and Consular Relations with Iran; the 1961 Optional Protocol to the Vienna Convention for Diplomatic Relations on Dispute Settlement; the Universal Postal Union Treaty; and the Intermediate Nuclear Forces Treaty.2 President Trump has also hinted at his desire to withdraw from the North American Free Trade Agreement (NAFTA), the Korea-U.S. Free Trade Agreement (KORUS), the World Trade Organization (WTO),and the North Atlantic Treaty Organization (NATO), the critical mutual defense alliance that the United States helped found almost forty years ago.3

If asked whether the President alone possesses a general, sweeping unilateral power to terminate every U.S. treaty in force, a layperson might well answer “no.” But among the legal academic community, the conventional wisdom seems to be “yes,” or at least “maybe.”4 On closer study, however, that conventional wisdom rests not on constitutional text, structure, or Supreme Court precedent, but on the thin reed of historical practice that followed the Court’s summary disposition nearly four decades ago in Goldwater v. Carter.5 In that case, the Court declined to review President Jimmy Carter’s unilateral termination of a bilateral treaty with Taiwan in accordance with its terms, but—like the Constitution’s text—the Justices left undecided which branch of government has the power of treaty withdrawal, and under what circumstances.6

This Essay argues that the conventional wisdom is wrong. The President does not have a general unilateral power of treaty termination. Goldwater is a splintered nonjusticiability ruling, not controlling precedent on the merits of this question. The merits have now become a live judicial question because changes in the law of justiciability would allow a court today—unlike in Goldwater—to reach the merits of this issue. And on the merits, no blanket power authorizes a unilateral presidential power to terminate international agreements. The Constitution does not directly address treaty withdrawal. Nor does the relatively recent U.S. practice of permitting unilateral withdrawal confer a historical “gloss” suggesting that, as a constitutional matter, the President possesses such blanket unilateral authority.7

Given that the sitting President now actively considers disengaging from a wide range of international agreements, the time is ripe for both the academy and the courts to explore this issue afresh. A constitutional matter this important and complex cannot be addressed by a single rule that purports to be “transsubstantive,” in the sense of governing the mechanics of withdrawal, suspension, or termination of national participation from each and every international agreement addressing every subject matter.8 On examination, the most apposite statement in Goldwater regarding the merits of treaty termination is the observation by four Justices of “the fact that different termination procedures may be appropriate for different treaties.”9 That observation does not suggest a general rule authorizing unilateral presidential termination for all agreements. If anything, it should demand consideration of a commonsense “mirror principle,” whereby absent exceptional circumstances, the degree of congressional participation constitutionally required to exit any particular agreement should mirror the degree of congressional participation that was required to enter that agreement in the first place.10

Under the mirror principle, the Executive may terminate, without congressional participation, genuinely “sole” executive agreements that have lawfully been made without congressional input.11 But the President may not entirely exclude Congress from the withdrawal or termination process regarding congressional-executive agreements or treaties that were initially concluded with considerable legislative input.12That principle would make Congress’s input necessary for disengagement even from such international agreements as the Paris Climate Agreement, which broadly implicate Congress’s commerce powers, and which—while never subjected to an up-or-down vote—were nevertheless enacted against a significant background of congressional awareness and support that implicitly authorized the presidential making, but not the unmaking, of climate change agreements.13 Congress also should participate in an attempt to withdraw the United States even from such political agreements as the Iran Nuclear Deal (also known as the JCPOA), where the President is exercising plenary foreign commerce powers that were delegated by Congress and where the U.S. termination has now triggered actionable claims of violation of international law.14

In sum, the conventional wisdom must be re-examined. The policy stakes are simply too high to allow such a fundamental question to rest on an inapposite and outmoded case decided decades ago. This Essay thus closes by suggesting better policy mechanisms—which better reflect both changing legal doctrine and shifting political realities—to guide America’s process of agreement-unmaking in the future.

I. goldwater and the “conventional wisdom” regarding unilateral withdrawal

Today’s conventional wisdom favoring unilateral presidential withdrawal from treaties is of surprisingly recent vintage. The Constitution empowers the President to “make Treaties, provided two thirds of the Senators present concur,” but it says nothing about which branch of government may unmake those treaties.15 Yet as one commentator summarized after extensive historical review, “[w]hereas it was generally understood throughout the nineteenth century that the termination of treaties required congressional involvement, the consensus on this issue disappeared in the early parts of the twentieth century.”16

But standing alone, this breakdown of consensus favoring congressional involvement in treaty termination does not explain the dramatic shift in the opposite direction, toward a conventional wisdom favoring unilateral termination. That sea-change was driven by the Supreme Court’s 1979 ruling in Goldwater v. Carter. In Goldwater, the Court dismissed a challenge by a group of Senators, led by Barry Goldwater, to President Carter’s unilateral termination of the 1954 Taiwan Mutual Defense Treaty. Since then, the Reporters of the Restatement (Fourth) of the Foreign Relations Law of the United States: Treaties note, “the United States has terminated dozens of treaties, and almost all of the terminations have been accomplished by unilateral presidential action.”17 But closer study reveals that neither Goldwater nor this recent historical practice offer sufficient legal basis to support a blanket, unilateral presidential power to terminate any and all international agreements.

A. The Court’s Decision in Goldwater v. Carter

On December 15, 1978, President Carter announced his intention to recognize and establish diplomatic relations with the People’s Republic of China and to terminate, as of January 1, 1980, the 1954 Mutual Defense Treaty between the United States and Taiwan.18 Seven U.S. Senators and eight Members of the House of Representatives sued the President and the Secretary of State in the U.S. District Court for the District of Columbia. They sought an injunction and a declaration that the President’s attempt to unilaterally terminate the treaty was “unconstitutional, illegal, null and void” unless “made by and with the full consultation of the entire Congress, and with either the advice and consent of the Senate, or the approval of both Houses of Congress.”19

When the 96th Congress opened, several Senators introduced resolutions asserting that the President had encroached on Congress’s constitutional role with respect to treaty termination generally and the Taiwan Mutual Defense Treaty in particular. In October 1979, the district court held that to be effective under the Constitution, the President’s notice of termination had to receive the approval of either two-thirds of the Senate or a majority of both houses of Congress.20 A fragmented D.C. Circuit, sitting en banc, heard the case on an expedited basis on November 13 and just seventeen days later ruled for the President.21 Declining to treat the matter as a political question, the circuit court instead held on the merits that the President had not exceeded his authority in terminating the bilateral treaty in accordance with its terms.22 Pressed to decide the case before the designated January 1, 1980 termination date, the Supreme Court issued no majority opinion. Instead, in a 6-3 per curiam decision, the Court dismissed the complaint without oral argument as nonjusticiable.23

The Justices splintered around several rationales, with only one Justice reaching the merits. Four Justices—Chief Justice Burger and Justices Rehnquist, Stewart, and Stevens—found that the case raised a political question.24 Justice Powell agreed that the case should be dismissed, but on the grounds that it was not ripe and because it was unknown “whether there ever will be an actual confrontation between the Legislative and Executive branches” constituting a “constitutional impasse.”25 Justice Brennan voted on the merits to uphold the President’s power to terminate the Taiwan treaty, based on the peculiar fact that the case involved derecognition of a foreign government, an issue over which he argued the President exercises textual plenary constitutional power.26 Justice Marshall simply concurred in the dismissal without explaining why.27 Only Justices Blackmun and White dissented, voting that the Court should “set the case for oral argument [to] give it the plenary consideration it so obviously deserves.”28

B. The Thin Goldwater Precedent

Not surprisingly, Goldwater has been consistently overread, particularly by executive branch lawyers. On its face, Goldwater is not a precedent supporting a unilateral presidential power of treaty termination. Rather, Goldwater simply supports the nonreviewability of one attempted unilateral termination, and, even then, on splintered grounds: four finding a political question and one finding nonripeness.29 Even at the time, none of the nine Justices embraced a rule favoring a general unilateral transsubstantive power of presidential termination. To the contrary, four Justices observed “that different termination procedures may be appropriate for different treaties,” which logically should have led to consideration of a more context-dependent rule such as the mirror principle discussed further below.30 Only one Justice opined that the President had the constitutional power to terminate even the particular treaty at issue in that case, and this was only because the unilateral termination of the bilateral treaty happened to have been conducted both in accordance with international law and within the scope of the President’s exclusive constitutional authority.31

Fairly read, Goldwater offers no precedent on the merits regarding a claimed general unilateral right to terminate bilateral treaties in all circumstances. It says nothing at all about three different factual scenarios: (1) terminations of or withdrawals from agreements that are not bilateral, but rather multilateral; (2) terminations or withdrawals that arguably are not implemented in accordance with the agreement’s terms or that otherwise arguably violate international law;32 or (3) terminations or withdrawals that are carried out within the scope of concurrent legislative-executive authority or Congress’s plenary authority, such as over international trade or foreign commerce, particularly when those agreements were initially adopted against a general background of congressional awareness and approval.33 At most, Goldwater suggests only that, for a variety of reasons, one particular presidential treaty termination decision should not be judicially reviewed. But in the four intervening decades, the law on the issue of justiciability has significantly changed.

II. justiciability law since goldwater

The discussion in Goldwater addressed itself almost entirely to the justiciability issues of standing, ripeness, and political question. But doctrinal developments during the four intervening decades suggest that next time around, plaintiffs should be able to reach the merits to challenge an attempted unilateral presidential treaty termination.

A. Standing and Ripeness

In Goldwater, five Justices voted to dismiss based on three threshold inquiries: first, whether Congress as a whole had challenged the President’s action; second, if not, whether the plaintiffs had standing to bring the action; and third, whether the case was genuinely ripe for decision absent what Justice Powell’s concurrence deemed a true “constitutional impasse.”34 Precisely how these issues will be answered in any particular case depend upon the particular subject matter and the agreement at issue.

For example, Congress has recently introduced two bipartisan bills that would prevent President Trump from withdrawing the United States from NATO without congressional consent.35 Should either of those bills receive a majority vote in both houses, and should the President nevertheless attempt to withdraw the United States unilaterally from NATO, there can be little doubt that Congress, as a whole, will have properly challenged the President’s action; the political branches will have reached the kind of “constitutional impasse” that Justice Powell envisioned as necessary for ripeness.

Recently, the threshold issues of standing and ripeness have arisen most immediately with respect to the Paris Climate Agreement, from which the Trump Administration has announced its pending withdrawal, to start in November 2019.36 If that attempted withdrawal were challenged in court, standing and ripeness would plainly be satisfied. Under its 2007 decision in Massachusetts v. EPA, the Supreme Court found that—in part due to the “special solicitude” granted to a sovereign state in a standing inquiry—a state had standing to sue the EPA to defend its “stake in protecting its quasi-sovereign interests” over potential damage caused to its shoreline as a result of global warming.37 Nothing in that decision should limit the standing of private citizens—such as local landowners suffering from lost property value from diminishing coastlines— to challenge a national climate change policy decision that exacerbated sea-level rise by accelerating the rate of melting of the polar icecap.38 This would constitute both injury in fact and redressable injury for the purposes of the Supreme Court’s current environmental standing test stated in Lujan v. Defenders of Wildlife.39 By enduring measurable sea-level rise, affected state and local governmental and nongovernmental plaintiffs could argue that they suffered a concrete and particularized, actual and imminent injury in fact that is fairly traceable to President Trump’s attempt to withdraw from the global collaborative efforts to address those concerns embodied in the Paris Agreement. This injury would be redressable by a favorable Court decision requiring legislative input before such a withdrawal could be finalized.40 Such a matter would also be ripe for immediate judicial consideration, since potential plaintiffs could document that measurable sea level rise is occurring in real time.

Whether Congress could establish ripeness in a lawsuit attempting to require legislative participation in disengagement from the Paris Agreement would depend on two factors: first, the state of the law regarding congressional standing, and second, the state of enacted legislation at the time of the lawsuit regarding the Paris Agreement specifically. With respect to the first factor, the Supreme Court has only twice directly confronted the issue of legislative standing, recently rejecting it in Raines v. Byrd,41while earlier accepting it in Powell v. McCormack.42 However, the D.C. Circuit—the most likely venue for a congressional suit—has historically been more inclined to recognize congressional standing on a discretionary basis.43 The D.C. Circuit’s inclinations could draw support from a recent Supreme Court decision, Arizona State Legislature v. Arizona Independent Redistricting Commission, which found state legislative standing to sue—with reasoning arguably authorizing congressional standing as well—when “an institutional plaintiff asserting an institutional injury” commences an “action after authorizing votes in both of its chambers.”44 Thus, sometime before November 4, 2019—the earliest date the United States could give notice of its withdrawal from the Paris Agreement—a similar bill could be introduced and voted on in one or both houses of Congress with respect to the Paris Agreement, and if enacted, the ripeness threshold would be met.

B. The Political Question Doctrine

In the wake of Goldwater, several lower courts followed the lead of the Goldwater plurality by finding nonjusticiable suits challenging executive power to unilaterally withdraw from treaties.45 But in an important recent case, Zivotofsky v. Clinton (Zivotofsky I), the Supreme Court declined to apply the political question doctrine to bar review of the President’s power to recognize foreign states in the face of a contrary congressional statute.46 Later, in Zivotofsky v. Kerry (Zivotofsky II), the Court held on the merits that the same statute was unenforceable because it was an unconstitutional intrusion on the President’s exclusive power of recognition.47

Chief Justice Roberts’s opinion for the Zivotofsky I Court called the political question doctrine a “narrow” exception to the general rule that the judiciary has the “responsibility to decide cases properly before it.”48 The opinion notably omitted mention of the six-factor political question test originally introduced in Baker v. Carr.49 Instead, the Court narrowed that test to its first two “textual” elements, explaining that a political question exists only “[1] where there is ‘a textually demonstrable constitutional commitment of the issue to a coordinate political department; or [2] a lack of judicially discoverable and manageable standards for resolving it.’”50

Under Zivotofsky I’s narrowed two-pronged political question test, treaty termination is not a political question. First, it is not a decision “textually committed” by the Constitution to a branch other than the judiciary. Article II, Section 2 of the Constitution authorizes the President to “make” treaties with the advice and consent of two-thirds of the Senators present,51 but no constitutional text expressly authorizes any branch to unmake such treaties, whether through suspension, termination, or withdrawal. Second, there is no “lack of judicially discoverable and manageable standards for resolving” the question. A court need only decide whether the President’s action—standing alone—is legally sufficient to terminate an international treaty obligation. As Justice Powell noted in Goldwater, in such a case, “the Court would interpret the Constitution to decide whether congressional approval is necessary to give a Presidential decision on the validity of a treaty the force of law[,] an inquiry [that] demands no special competence or information beyond the reach of the Judiciary.”52

Whether the President may enter or withdraw from any agreement against the will of Congress may not be an easy case; surely, it would be a political case. But that does not make the contested issue a political question. To decide it, a court need only apply entirely familiar principles of constitutional interpretation—text, structure, and historical evidence about the nature of law and the constitutional powers at issue. As proof, when the Zivotofsky case returned to the Supreme Court a few years after the justiciability ruling, following full examination before the D.C. Circuit, the Zivotofsky II majority used precisely these “judicially manageable tools” to hold that the passport statute in question violated the President’s exclusive recognition power.53

In Zivotofsky II, the Court affirmatively answered the question whether a statute enacted by Congress had unconstitutionally encroached on the President’s foreign affairs power. The parallel question here would be whether the President’s unilateral termination of the agreement at issue unconstitutionally encroached on Congress’s foreign affairs power. That question should now fall outside the Court’s newly limited political question doctrine. Indeed, if the Court can decide on the merits such thorny separation-of-powers issues as were presented in Marbury v. Madison,54 Youngstown Sheet & Tube Co. v. Sawyer,55 Myers v. United States,56 INS v. Chadha,57 Morrison v. Olson,58and Zivotofsky, it should be equally competent to decide the merits of the fundamental constitutional question of whether the President’s unitary action is legally sufficient to bind other branches of government and the world.59 After all, as Chief Justice Roberts reminded us in Zivotofsky I, enforcing the separation of powers “is what courts do.”60

In short, today, a contested presidential effort to terminate an important international agreement would not be insulated from judicial review. Fairly read, Goldwater itself offers slim precedent on the merits to support the legality of such a unilateral termination. Whatever constitutional case exists for the President’s power to unilaterally terminate any agreement cannot rest on Goldwater or any other recent Supreme Court decision. Instead, it must rest exclusively on recent historical practice.

iii. may the executive unilaterally withdraw from any international agreement?

A. Historical Practice and “Functional Considerations”

Historical practice offers the strongest argument to support expansive application of a unilateral termination “rule.” Since the 1930s, Presidents have unilaterally terminated several international agreements, including “a few dozen” since Goldwater.61 But as the Reporters to the recent Restatement (Fourth) of the Foreign Relations Law of the United States correctly noted, “[m]ost of these terminations have not generated controversy in Congress.”62

The mere fact that the President may have unilaterally terminated agreements that Congress did not care about tells us little about what would happen if Congress were to actively contest a withdrawal, as in the NATO example discussed above. In fact, the appellate briefs in Goldwater debated whether there were genuinely more than three contested treaty withdrawals in all of American history.63 Admittedly, the Supreme Court has at times recognized “historical practice” as a basis for normative reinterpretation of structural constitutional provisions.64 But as Justice Frankfurter’s famous discussion of historical practice in Youngstown made plain, “systematic, unbroken, executive practice” places a historical “gloss on the ‘executive power’” in Article II, only if—like adverse possession in property law65—that practice has been “long pursued to the knowledge of the Congress and never before questioned . . . .66 Even four decades of consistent executive practice would not rise to the level of historical “gloss,” unless the other affected branch of government “acquiesced” in the legality of the executive acts by some unmistakable affirmative act. This plainly would not be the case with respect to important multilateral agreements that enjoy bipartisan congressional support, such as the NATO agreement.

Even more glaring, no statutory provisions currently require the President to notify Congress of an executive decision to terminate or withdraw from any treaty or international agreement.67 Nor is there any easily available listing of agreements or treaties that may have been unilaterally terminated.68 Although the State Department Legal Adviser’s annual Digests of United States Practice in International Law voluntarily report on some terminations, there is no legal requirement that the Office of the Legal Adviser or the State Department do so, and the Digests are not published until after the incidents they describe.69 Finally, the 1972 Case-Zablocki Act,70 which ostensibly requires the executive branch to notify Congress of all foreign agreements, has prominently failed to ensure the complete reporting of all agreements and political commitments that Presidents may actually make.71 Thus, following Trump’s recent “one-on-one” Helsinki meeting with Russian President Vladimir Putin, no one other than President Trump, Putin, or their interpreter knew for sure whether Trump had made or attempted to make, suspend, terminate, abrogate, or withdraw from any treaties or agreements with Russia.72 It would be very odd indeed to treat as accepted “customary constitutional practice” executive branch activity in which Congress could not possibly have “acquiesced” because it was entirely unaware.

Accordingly, the text of Section 113 of the recent draft of the Restatement (Fourth) of the Foreign Relations Law regarding “Authority to Suspend, Terminate, or Withdraw from Treaties” carefully states that “[a]ccording to established practice, the President has the authority to act on behalf of the United States in suspending or terminating U.S. treaty commitments and in withdrawing the United States from treaties . . . .”73 The accompanying Comment makes clear that [t]he Supreme Court has not resolved the constitutional authority to terminate a treaty.”74Thus, the Restatement (Fourth) rests its acknowledgement of presidential authority exclusively on historical practice. The question is how “established” that practice really is.

Like Goldwater itself, the Restatement’s black letter comes with significant limitations. First, as discussed further below with respect to the Iran Nuclear Deal, whatever unilateral termination power may be recognized by historical practice, the Restatement does not recognize it as extending to unilateral acts by the President to suspend, terminate, or withdraw from treaties not in accordance with their terms or otherwise not in accordance with international law.75Second, the Restatement nowhere “suggest[s] that Congress or the Senate lack the ability to limit suspension, termination, or withdrawal,” for example, by the No NATO Withdrawal Act described above.76 Third, the accompanying Reporters’ Notes reaffirm that “[a]lthough historical practice supports a unilateral presidential power to suspend, terminate, or withdraw the United States from treaties, it does not establish that this is an exclusive presidential power.”77 Thus, if the President were to attempt a treaty termination within zones of either concurrent congressional-executive foreign affairs authority or exclusive legislative power—such as the foreign commerce power78—the Senate could presumably limit executive discretion pre-emptively in its advice and consent to a particular treaty, as could Congress, by enacting a “no unilateral exit” statute with respect to that international agreement.

Apart from historical practice, the only other support offered for the Restatement (Fourth)’s black letter is that “structural and functional considerations are consistent with the general ability of the President to act unilaterally, as it has been exercised in practice.”79 But looked at from a dynamic perspective, this statement no longer seems true. If anything, structural and functional considerations now cut the other way. As a matter of constitutional structure, the foreign affairs power is generally a power shared: unilateral powers are the exception, not the rule, so whenever the Constitution’s text does not explicitly assign a plenary power to one branch, the multiple, overlapping grants of foreign affairs authority should presumptively dictate that powers be shared between Congress and the Executive.80

As a functional matter, an overbroad unilateral executive withdrawal power would not only risk overly hasty, partisan, or parochial withdrawals by Presidents, but would also tend to weaken systemic stability and the negotiating credibility and leverage of all Presidents. The most prominent recent example is President Trump’s abrupt withdrawal from the Iran Nuclear Deal at the precise moment that he is attempting to negotiate a similar denuclearization deal with North Korea.81 Whatever functional sense a cross-cutting Goldwater approach—strong unilateral presidential termination rights coupled with minimal judicial review—might have made when first articulated, the world has plainly changed in ways that call into question its normative logic. Back when “politics stopped at the water’s edge”—and a presumption of basic foreign policy continuity dominated whenever the White House changed hands—academics and justices might well have believed that their approach would minimize foreign policy conflict and make the United States more compliant overall with international norms. But the rise of a post-Cold War political era marked by radical foreign policy discontinuities from Presidents Clinton to Bush to Obama, and now to Trump, has dramatically undermined this assumption.82 As the current moment illustrates well, unless blessed by Congress in some fashion, unilateral presidential withdrawals based on individual presidential caprice are highly disruptive of both foreign (and increasingly domestic) policy. That very disruptiveness and unpredictability will, in turn, make it harder for future Presidents and Congresses to negotiate valuable international agreements.

Perhaps most important, in separation-of-powers disputes, such functional considerations have not historically proven determinative in dictating structural answers to questions of constitutional interpretation. In INS v. Chadha, for example, the Court famously invalidated the legislative veto, even in the face of longstanding congressional practice in using the device.83 Although 295 legislative vetoes had been inserted into nearly 200 statutes since 1932,84 Chief Justice Burger held not only that the matter was justiciable,85but also that whether

a given law or procedure is efficient, convenient, and useful in facilitating functions of government, standing alone, will not save it if it is contrary to the Constitution. . . . [P]olicy arguments supporting even useful political inventions” are subject to the demands of the Constitution which defines powers and . . . sets out just how those powers are to be exercised.86

In sum, the historical practice is far from “established” in constitutional law. The best reading of that practice is that Goldwater has functioned as a piece of “quasi-constitutional custom” that is “perennially subject to revision.87 Much of the more recent historical practice of unilateral presidential termination since Goldwater can be attributed to path-dependence and conventional wisdom, rather than to serious substantive review of the constitutional arguments on the merits. Any rule suggested by the most recent historical practice could be swiftly revised either by judicial decision or by altered interbranch practice going forward.88 In the modern era, an agreement-specific mirror principle—requiring parity of constitutional authority for entry and exit from an international agreement—represents a far better functional reading of the Constitution than a claimed general unilateral right of the president to terminate any and all international agreements.

B. The Mirror Principle in Comparative Constitutional Practice

Of course, the constitutional issue of treaty exit arises not just in the United States, but in every other country in the world. As Justice Breyer noted nearly two decades ago, the Supreme Court “has long considered as relevant and informative the way in which foreign courts have applied standards roughly comparable to our own constitutional standards in roughly comparable circumstances.”89 In Printz v. United States, he elaborated that the “experience [of other nations] may . . . cast an empirical light on the consequences of different solutions to a common legal problem.”90

The recent foreign decision casting the most relevant empirical light on the termination of multilateral agreements whose provisions are deeply intertwined with domestic law is the United Kingdom’s famous Brexit litigation, R v. Secretary of State for Exiting the European Union.91 There, the Supreme Court of the United Kingdom held that the U.K. government may not use its executive prerogative powers but rather must seek parliamentary approval to trigger Article 50, the withdrawal provision of the Treaty of the European Union.92 The court reasoned that prior parliamentary approval was required because Brexit would require fundamental constitutional changes, including the repeal of the 1972 European Communities Act, which expressly allowed for E.U. treaties to take effect within U.K. domestic law.93 This is particularly the case, the court noted, because fundamental rights would undeniably be affected within the country once the treaty withdrawal was completed. As one commentator put it, “the years of [treaty] membership, the weaving of the [intertwined fabric of domestic and international legal rules], have constructed a reality that is hard to change.94 Since the British executive would not have the power to effect the removal of the Act unilaterally as a source of U.K. domestic law, the court reasoned, neither should the Prime Minister have the power to unilaterally withdraw from the treaty without legislative participation.95

Similarly, in Democratic Alliance v. Minister of International Relations and Cooperation,96 the High Court of South Africa for the Gauteng Division recently held that the executive branch could not unilaterally withdraw from the Rome Statute of the International Criminal Court without parliamentary approval.97 The court stated:

[W]here a constitutional or statutory provision confers a power to do something, that provision necessarily confers the power to undo it as well. In the context of this case, the power to bind the country to the Rome Statute is expressly conferred on parliament. It must therefore, perforce, be parliament which has the power to decide whether an international agreement ceases to bind the country.98

The court’s decision suggests a mirror principle: the commonsense notion that the degree of legislative participation necessary to exit an international agreement should mirror the degree of legislative participation required to enter it in the first place. Relying on this mirror principle, the court held that South Africa could withdraw from the Rome Statute only on approval of parliament and after the repeal of the statute implementing the treaty. In response, the South African government complied with the court’s order and revoked the instrument of withdrawal.99

The reasoning of these and other foreign precedents100 cast “empirical light” on—and cut strongly against—the view that functional considerations should sanction a general unilateral termination rule. Instead, these cases suggest that U.S. withdrawal from a long-standing treaty or international organization —such as the United Nations or the World Trade Organization, whose rules have also been deeply internalized into U.S. law—should not become effective without congressional involvement.101 Such a withdrawal or termination would similarly necessitate unwinding many domestic law statutes that the executive could not repeal alone.

Under the mirror principle, there should be parity of authority for entry and exit from an international agreement. Absent exceptional circumstances, a treaty entered into with substantial legislative participation cannot be lawfully terminated by the President alone. The same degree of legislative participation is legally required to exit from as to enter an international commitment. By embodying that principle, the U.K. Supreme Court’s Brexit ruling recognized an important functional truth: that in the modern era, international agreements are not just transactional, but relational. These agreements do not authorize sequential, one-off transactions so much as they create organic, evolving relationships that generate a deeply interconnected set of rights, duties, expectations, and reliance for all parties. Momentary lapses may be forgiven if a fundamental commitment to the enduring relationship has been demonstrated. But the disruptive act of termination engages the interests of all parties, both participating and represented, who accordingly should be as much involved in the decision to exit as they are in the initial decision to enter.

As the British polity has learned since its June 2016 Brexit referendum vote, breaking up is hard to do. Particularly in legal systems where treaties are the supreme law of the land, they confer legal rights that have direct effect on domestic actors. Those treaties lay a foundation upon which, over time, sedimentary layers of legal acts, executing legislation, and court decisions build a deeply internalized framework of transnational law that embeds treaty membership strongly into the domestic fabric of political life.102 Withdrawing abruptly from such an organic treaty framework becomes akin to trying to pull out only the red threads from a multi-colored tapestry. Withdrawal rips the fabric of domestic law and disrupts all manner of domestic rights and expectations. As proof, even as President Trump daily invokes a transactional approach to international affairs, condemning all manner of international agreements as “bad deals,” his Administration recently made the exact opposite relational argument before the Supreme Court. In Jam v. International Finance Corp., Solicitor General Noel Francisco responded to the Court’s invitation to participate not by rejecting agreements as “bad deals,” but by arguing that the “United States’ participation in international organizations is a critical component of the Nation’s foreign relations [that] reflects an understanding that robust multilateral engagement is a crucial tool in advancing national interests.”103

Should the unilateral termination issue soon arise in the U.S. context, the reasoning from this comparative precedent should prove instructive. Many Supreme Court Justices, including current Justices Breyer104 and Ginsburg,105 as well as seven departed Justices—Kennedy,106 Stevens,107 Souter,108 O’Connor,109 Blackmun,110 Scalia,111 and Rehnquist112—have all famously referenced constitutional practice from other democratic countries.

The constitutional issue raised by Goldwater may well be relitigated in the context of an attempted executive withdrawal from a multilateral treaty whose provisions are deeply internalized within our domestic law. If so, we could expect the reasoning of these and other comparative law precedents—particularly the notion of a mirror principle requiring parallel legislative participation for entry and exit—to be prominently cited by both the litigants and Justices of the Supreme Court.

C. Constitutional and Functional Considerations

By suggesting that withdrawals from international agreements whose provisions have been deeply internalized into domestic law should be dictated by a “mirror principle,” the foreign precedents embrace a view long acknowledged by U.S. constitutional jurisprudence.113 As a constitutional matter, the mirror principle is simply a variant of the famous “last-in-time rule,” first stated in the Head Money Cases114 and Whitney v. Robertson.115 Those cases long ago settled that, however made, a binding international agreement can only be superseded by an agreement or statute that is adopted with a comparable degree of legislative input. Because such a superseding legal enactment effectively exits the United States from a prior international commitment, for all practical purposes, the mirror principle was the law long before Goldwater and remains so today. After all, self-executing treaties are the law of the land, no less than congressionally enacted statutes.116 If the President cannot enact or repeal a statute alone, why should he be able to repeal the duly enacted law of the land—and its accompanying framework of deeply internalized domestic law—just because the initiating juridical act happened to be in treaty form?117

Executive branch lawyers have tried to escape this reasoning by arguing—by analogy to executive appointments—that the President alone should be able to terminate a treaty that was made with the advice and consent of the Senate, because the President alone can fire a cabinet member appointed with the advice and consent of the Senate.118 But the analogy to appointment of subordinates is inapposite, not least because it is a largely internal matter regarding control of the executive branch.119 The removal power also rests partially on the Article II Vesting Clause120 and the Take Care Clause,121 insofar as the President cannot properly discharge executive power or take care that the law be faithfully executed if he cannot use the threat of unilateral removal to hold subordinates accountable.122 But unilateral treaty abrogation, by contrast, runs afoul of the Take Care Clause, because the President is unilaterally undoing, not enforcing, the law that has been made.

Other commentators have made a functional “quick divorce” argument: that foreign affairs exigencies may require the United States to exit quickly entangling alliances that were entered deliberately, and with extensive congressional awareness and participation, over a much longer period of time. Under this functional theory, the President alone is best positioned to decide whether and when a quick divorce is necessary.123 This theory argues that a constitutional rule that makes it harder to exit agreements will make future executives less keen to enter into them in the first place. But if anything, today, these structural and functional considerations should support the mirror principle, not a general rule of unilateral executive termination. Even if the short term, first-order “functional” effect of the mirror proposal would be to constrain the executive, the second- and third-order functional effects would strengthen the Executive’s hand in future international negotiations. As recent denuclearization talks with North Korea and Iran show, the negotiating credibility of an executive who wishes to enter into a new agreement will be strengthened, not weakened, if such agreements are made harder to break by their successors, thereby giving their negotiating partners greater assurance regarding the continuity and stability of the proposed arrangement.

In any event, as noted above with respect to INS v. Chadha, which invalidated the legislative veto, such fundamental issues of constitutional interpretation cannot be resolved by functional concerns alone. But even setting Chadha aside, it would be wildly overbroad to allow the exception to swallow the rule. There is no reason to extrapolate from the narrow notion that, in a genuine emergency, a President may be best positioned to decide whether an exit from a treaty is urgently required, to the blanket proposition that the Constitution authorizes any President, on impulse, to withdraw the United States from any and all bilateral and multilateral arrangements with which it has engaged over the centuries. As Judge MacKinnon presciently warned in Goldwater: “In future years, a voracious President and Department of State may easily use this grant of absolute power [of unilateral termination] to the President to develop other excuses to feed upon congressional prerogatives that a Congress lacking in vigilance allows to lapse into desuetude.”124

IV. a typology of withdrawals

By now it should be clear that the question that opened this Essay—whether President Trump could unilaterally withdraw the United States from any and all international treaties and organizations—is readily contestable. On the merits, only one U.S. court has ever endorsed an unnuanced rule of unilateral presidential termination—the en banc D.C. Circuit decision in Goldwater125—and that ruling was vacated less than a month later by the U.S. Supreme Court.126

Given the United States’ admitted interests in international organizations and multilateral treaties, what constitutional test should govern the President’s power to withdraw unilaterally from such arrangements? While this issue has arisen most recently with respect to the Paris Agreement and the Iran Nuclear Deal, this constitutional question should also be asked with respect to potential withdrawal from a broad range of global arrangements, including international organizations, international tribunals, mutual security organizations, and trade agreements. Applying a general unilateral rule of termination, across these varied cases, would potentially allow one man to disengage from most of our international commitments, devastating the post-World War II international order that many administration and treaties helped to construct.

If and when the Supreme Court finally considers this issue on the merits, it should conclude that no single “transsubstantive” rule governs whether and how each and every international agreement may be terminated or withdrawn from as a matter of U.S. domestic law.127 As Laurence Tribe reasoned at the time of Goldwater, “the very fact that the Constitution does not prescribe a mode of treaty termination suggests that the framers did not think any one mode appropriate in all cases, and therefore left the matter to be resolved in light of the particular circumstances of each situation.”128 In Goldwater itself, the four Justices who voted to dismiss on political question grounds expressly “decline[d] the invitation to set in concrete a particular constitutionally acceptable arrangement by which the President and Congress are to share treaty termination.”129 In contrast to an overbroad unilateral presidential termination “rule,” the mirror principle does not mandate a “one-size-fits-all” mode of agreement termination. Instead, depending on the substance and entry process, the mirror principle requires varying degrees of congressional and executive participation to exit from various kinds of international agreements. Such an approach reflects the fundamental constitutional wisdom in Justice Jackson’s famous concurrence in Youngstown Sheet & Tube Co.: that “[p]residential powers”—in this case, to terminate or withdraw from binding international agreements of the United States—”are not fixed but fluctuate, depending upon their disjunction or conjunction with those of Congress.”130

Last year in these pages, I made the case against simplistic application of the traditional transsubstantive “triptych” by which scholars have artificially divided the constitutional spectrum of U.S. international agreements into three categories: Article II treaties, congressional-executive agreements, and so-called “sole” executive agreements, made solely pursuant to the President’s constitutional powers.131 If an agreement entails new, legally binding obligations, I argued, we should examine first, “the degree of congressional approval for the executive lawmaking,” and second, “the constitutional allocation of institutional authority over the subject matter area at issue.”132 The first factor—the degree of congressional approval legally required to enter an agreement—roughly maps onto the three Youngstown categories.133 But whether that degree of congressional approval is constitutionally mandated depends on a second factor as well: which branch of government has substantive constitutional prerogatives regarding that area of foreign policy. In the area of recognition, for example, the President’s plenary power eliminates the need for congressional approval to enter an agreement; but in the area of foreign commerce, his discretion is limited and could even be barred by Congress’s foreign commerce powers.134 Under the mirror principle, the degree of congressional and executive participation required to exit from an international agreement should turn on these same two factors: the subject matter of an agreement at issue and the degree of congressional approval involved in both the entry into and exit from it. Thus, as with entry into any particular international agreement, the degree of legislative participation required to exit that same agreement is substance-dependent. Because the mirror principle requires for exit only the degree of legislative participation required for entry, it is flexible enough to vary according to the subject matter.

To be workable, of course, any constitutional approach must be able to accommodate all political exigencies.135 To accommodate bona fide emergency situations, the U.S. Supreme Court could recognize the President’s unilateral authority to suspend U.S. treaty obligations for a limited period, long enough for the President to make the case to Congress as to why withdrawal is warranted “for cause.”136But recognizing a very limited exception to the mirror principle for genuinely exceptional circumstances is to acknowledge a decidedly narrower—and functionally, far more appropriate—authority than according a single person a blanket unilateral power to terminate, withdraw, or suspend all agreements in all situations, emergency or otherwise.

The balance of this Essay reviews how the mirror principle would apply in four situations: (1) agreements lawfully concluded with no legislative inputgenuine “sole executive agreements”; (2) agreements initially concluded with considerable legislative input, such as congressional-executive agreements and treaties; (3) agreements initiated by the executive with general congressional awareness and approval in a zone of congressional subject matter authority, such as the Paris Climate Agreement; and (4) agreements arguably terminated in violation of international law in an area of congressional subject-matter authority, such as the Iran Nuclear Deal. I conclude by suggesting policy mechanisms to guide the conduct of agreement-unmaking as the future unfolds.

A. Agreements Concluded with No Legislative Input: Sole Executive Agreements

There are undeniably subject-matter areas where the President has plenary authority to make agreements acting alone. In those subject-matter areas, the mirror principle would suggest that the President alone has the power to unmake those same agreements so long as Congress has been silent on the matter.137 That said, historically, true “sole executive agreements”138—agreements based solely on the President’s plenary constitutional authorities—have been extremely rare. Consider, for example, Franklin Delano Roosevelt’s 1941 Destroyer-for-Bases deal with the United Kingdom139 or the early twentieth-century agreements recognizing the new Soviet Union,140 affirmed in United States v. Belmont141 and United States v. Pink.142

In practice, however, few modern presidents ever claim to be making a controversial agreement based solely on their own plenary constitutional authority, particularly in cases where Congress has legislated extensively elsewhere regarding the same subject. As elaborated below, recent accords like the Paris Climate Agreement and the Iran Nuclear Deal are not sole executive agreements in this narrow constitutional sense because they were not concluded solely within the zone of the President’s exclusive constitutional authorities.143 Thus, just because the President may appear to “make” an agreement alone does not necessarily mean that he is constitutionally empowered to break it alone. As noted below, those agreements are not sole executive agreements, but were premised on a broader history of legislative authorization in a particular direction. During that same period, Congress expressed no general approval for presidential actions disrupting climate change negotiations or re-imposing economic sanctions absent a breach by Iran. Thus, for the President, acting alone, to terminate such agreements would flout and disrupt congressional expectations and approbation in a way that a prior President’s initial entry in the agreement did not.

B. Agreements Concluded with Considerable Legislative Input: Treaties and Congressional-Executive Agreements

Since World War II, the number of Article II treaties approved by two-thirds of the Senate has declined dramatically, with a sharp attendant rise in congressional-executive agreements as “interchangeable instruments of national policy” approved by a majority of both houses of Congress.144 Both kinds of legal instruments require substantial legislative input for entry. The mirror principle would accordingly suggest that, ordinarily, comparable legislative input should be required to exit the same agreements.145

Applying Justice Jackson’s Youngstown test to the termination of congressional-executive agreements could reflect three distinct political realities: congressional approval, disapproval, or ambivalence towards executive action. If Congress were to approve the President’s termination, explicitly or implicitly, then the President’s power would obviously be at its zenith, and the President could terminate either a congressional-executive agreement or ratified treaty without controversy. But if Congress were expressly to disapprove of the President’s attempted termination, the legal conclusiveness of the termination would become substance-dependent. That executive termination could only stand if the President were operating within a zone of exclusive presidential authority, such as state recognition, as illustrated by Zivotofsky II146 or Justice Brennan’s dissent in Goldwater.147 Finally, if Congress were silent, or could not organize itself to make a collective statement, the President would be operating within what Justice Jackson famously referred to in Youngstown as the “twilight zone,” where the President’s power to terminate agreements unilaterally would depend on proof that Congress did more to approve that action than merely acquiesce.148

In Goldwater itself, for example, the President—in the face of collective congressional inaction—terminated the Taiwan Treaty, which as Justice Brennan noted, fell into the heart of the President’s exclusive power of recognition.149 Such a unilateral termination would invoke the President’s exclusive and plenary constitutional power respecting recognition. Indeed, Zivotofsky II later similarly invoked the exclusive recognition power to negate a contrary congressional statute. But it would be quite another matter if the President tried to act alone to terminate an agreement that was both made by and fell within the heart of Congress’s concurrent or exclusive constitutional authority over a subject matter (for example, foreign commerce). For that reason, commentators have persuasively argued that President Trump’s threat to terminate or withdraw from NAFTA would be barred by Congress’s Commerce Clause authority.150 The same could be said for a treaty that affected Congress’s exclusive power of the purse.151 Again, the mirror principle should apply. If the President lacked the constitutional authority to make an agreement in a subject-matter area without congressional approval, the President would lack authority to unmake unilaterally an agreement that Congress helped substantially to create in the same subject-matter area.

C. “Executive Agreements Plus,” Adopted Against a Background of Legislative Approval in a Zone of Congressional Power: The Paris Climate Agreement

The 2015 Paris Climate Change Agreement was not a sole executive agreement but what some have called an “executive agreement plus,” a bundle of commitments made by executive initiative, but with general congressional awareness and approval in a zone of significant congressional power.152 While there is considerable debate regarding the Paris Agreement’s status under U.S. law, no one doubts that the Agreement constitutes a “treaty” under international law, for purposes of the Vienna Convention on the Law of Treaties (VCLT).153 Some provisions of the Paris Agreement are legally binding under international law.154 But they “are largely procedural in nature and in many instances are duplicative of existing U.S. obligations under the [UNFCCC], and, therefore, could be fully implemented based on existing U.S. law.”155 This feature allowed President Obama to enter the Paris Agreement without submitting it for Senate approval, because he had sufficient domestic authority to implement any legal obligations assumed under the agreement merely by carrying out preexisting domestic legal obligations.

But exiting the same agreement is not so straightforward. International law makes clear that U.S. presidents cannot simply delete prior signatures from treaties.156 Nor under the VCLT can a single state party invalidate an entire multilateral agreement. Nevertheless, on June 1, 2017, President Trump announced his intent to withdraw from the Paris Agreement in more than two years’ time.157

Curiously, a recent Harvard Law Review article discussing presidential control over international law stated that “[t]here was significant controversy about the policy wisdom of this decision, but no one questioned the President’s legal authority to terminate in this context.”158 Putting to one side the factual inaccuracy of this statement,159 even if true, it tells us little, for the simple reason that President Trump’s announcement in June 2017 was thirty months premature. No one questioned the President’s legal authority in that context because in June 2017, Trump neither terminated nor withdrew U.S. acceptance of the Paris Agreement, nor has he done so today. His announcement simply stated a future intent; it did not in any sense legally disengage the United States from the Paris Agreement. The Paris Agreement only recognizes withdrawal under the terms specified in its text, which plainly declares that a party cannot give notice of withdrawal to the U.N. Secretary General until “three years from the date on which this Agreement has entered into force.”160 Since the Paris Agreement entered into force on November 4, 2016, the earliest date that the United States could even give such legal notice would be November 4, 2019. That notification would then take another year to take legal effect. Thus, the earliest that the United States could legally withdraw from the Paris Agreement is November 4, 2020, the day after the next U.S. presidential election—and even that assumes that such an attempted unilateral withdrawal would be upheld following constitutional litigation. The obvious reason why no one questioned the President’s legal authority “in this context” or launched any legal challenge against it simply because such a challenge was not yet remotely ripe for judicial examination.

Until November 2019, President Trump’s announced intent to withdraw in the future has no more legal force than an employee’s empty threat to leave his job in three years’ time.161 The United States has not “virtually” or “preemptively" withdrawn or otherwise formally disengaged in any way as a party from the Paris Agreement. At this writing, the U.N. Treaty Depositary page on the Paris Agreement still lists the United States as a party.162 While the State Department website indicates that it has notified the U.N. Depositary of its “intent to withdraw,” a “[m]edia [n]ote” makes clear that all the United States has done is communicate “the U.S. intent to withdraw from the Paris Agreement as soon as it is eligible to do so, consistent with the terms of the Agreement.”163 In August 2017, then-U.S. Ambassador to the United Nations Nikki Haley informed the U.N. Secretary-General of the United States’ intent “to exercise its right to withdraw from the Agreement . . . . [u]nless the United States identifies suitable terms for reengagement,” a condition subsequent that plainly leaves wiggle room to call off the proposed withdrawal.164 So in the words of a coalition of Paris Agreement supporters, “We Are Still In.”165 So long as the United States continues to participate in the Paris process,166 it will remain a state party for the balance of Trump’s first term; it has not yet withdrawn from the Paris Agreement. The United States has only “resign[ed] without leaving,” thereby prematurely identifying itself as a lame duck.167

If, in November 2019, the Administration should finally carry through on its stated intent to give notice of its desire unilaterally to withdraw from the Paris Agreement in one year’s time, new litigation brought by states, leading environmental groups, and others would almost certainly ensue. Plaintiffs would then surely argue that the notice is legally ineffective because the President lacks constitutional power to withdraw from the Paris Agreement without congressional participation.168 Only then would the broader constitutional question addressed in this Essay finally be joined.

At that time, the Trump Administration would likely reply that the President possesses unilateral authority to withdraw from the Paris Agreement because—as fourteen Senators argued in a 2016 letter to then-Secretary of State John Kerry—the Paris Agreement could be dissolved by the next President alone as a “‘sole executive agreement[]’ [which is] . . . one of the lowest forms of commitment the United States can make and still be considered a party to an [international] agreement.”169 But the Senators’ statement is legally inaccurate. As others have correctly noted, “the dividing line between sole [executive agreements] and ex ante [congressional-executive agreements], in particular, can be difficult to divine, which necessarily has important ramifications for termination.”170 As noted above, the Paris Agreement is not a “sole executive agreement,” adopted under the President’s plenary constitutional powers, but rather was expressly preauthorized171 by a duly ratified Article II treaty, one negotiated within the scope of the Senate’s original advice and consent to the 1992 U.N. Framework Convention on Climate Change. In addition, the Paris Agreement followed a general statutory landscape that envisioned future efforts at international lawmaking to regulate climate change. Congress had expressed its support for climate change negotiations in: (1) the Global Climate Protection Act of 1987, which asserted the need for “international cooperation aimed at minimizing and responding to adverse climate change;”172 (2) the Clean Air Act (which the Supreme Court held in Massachusetts v. EPA173 authorized the EPA to regulate carbon dioxide emissions from motor vehicles as a pollutant, thereby allowing the President to argue that he could negotiate international agreements as a necessary adjunct to regulating domestic emissions); and (3) Section 115 of the Clean Air Act, which authorizes federal action reciprocally with other nations to address “international air pollution,” namely, transboundary pollution causing international damage.174

Thus, the Paris Agreement was not a sole executive agreement in the traditional sense, but rather, an agreement approved by Congress through both ex ante treaty and general legislative preauthorization. Under the Supreme Court’s decision in Dames & Moore v. Regan,175the Paris Agreement represents a form of congressional-executive agreement made within the scope of Youngstown Category One: because “the President acts pursuant to an express or implied authorization of Congress, his authority is at its maximum, for it includes all that he possesses in his own right plus all that Congress can delegate.”176 Allowing the president alone to break an agreement supported by such legislative approval at the time of entry would violate the mirror principle by undermining Congress’s long-stated intent and approval for agreement-making to mitigate the effects of climate change.

Under the mirror principle, the degree of congressional and executive participation required to exit from an international agreement turns on two factors: the subject matter of an agreement at issue and the degree of congressional approval involved in both the entry and exit from it. Because the Paris Agreement strongly implicates Congress’s powers over commerce—here, the authority to regulate greenhouse gases—under the mirror principle, significant legislative input should also be required before the United States may constitutionally withdraw from it. Over time, the degree of legislative input into the making of the Paris Agreement was significantly higher than the degree of congressional input into a sole executive agreement. Moreover, Congress expressed no general approval for unilateral presidential actions disrupting future climate change negotiations. When an agreement is premised on such a broad history of legislative authorization in a particular direction (here, toward climate change negotiation), unilateral presidential termination of such an agreement would flout congressional will and expectations in a way that the making of the agreement would not.

D. The Iran Nuclear Deal: Terminating Agreements Regarding Economic Sanctions in Violation of International Law

Finally, it should not casually be assumed that the President has unilateral authority to withdraw from the JCPOA, commonly known as the Iran Nuclear Deal.177 To be sure, under domestic and international law, the JCPOA is a politically, not legally, binding arrangement, that was made by the President acting alone.178 But for two reasons, that fact alone does not dictate that the President should be able to terminate it unilaterally, without congressional participation. First, under the Iran Nuclear Deal, the President exercises not his own constitutional powers, but Congress’s delegated power to regulate foreign commerce by imposing or waiving economic sanctions. Congress is constitutionally entitled to review whether he is exercising its delegated sanctions power appropriately. Second, it has now been argued by Iran that the President terminated the Iran Nuclear Deal in violation of international law without congressional approval, potentially subjecting the United States to international law liability. A decision in violation of international law plainly falls outside the scope of Goldwater v. Carter, will directly affect many constituents, and thus should engage Congress’s institutional responsibility.

The JCPOA envisioned actions by Iran, the International Atomic Energy Agency (IAEA), and a group of states known as the P5+1.179 After extended negotiations, Iran agreed to specified limits on its nuclear development program in exchange for the P5+1’s joint undertaking to lift some domestic and international sanctions that had been imposed through the United Nations.180 The JCPOA has been implemented on the U.S. side largely through executive orders that suspended nuclear-related sanctions, in exchange for Iran’s dismantling of key elements of its nuclear program under the watchful eye of the IAEA.181 Under the JCPOA, the United States and the European Union committed to removing their nuclear-related domestic sanctions, and to proposing and voting for a new U.N. Security Council Resolution, which terminated and replaced past resolutions,182 thereby changing the nature of the other countries’ legal obligations under Chapter VII of the U.N. Charter to provide sanctions relief to Iran.

Like the Paris Agreement, the Iran Nuclear Deal derives from far more legislative input and support than is generally recognized. The Constitution plainly grants Congress subject matter authority over foreign commerce, and hence, economic sanctions. In turn, Congress delegated implementation of these authorities to the President. Congress also granted the President specific statutory authority to waive existing domestic law sanctions against Iran if he determines that it is in the national interest. Those laws gave President Obama ample statutory authority to enter a deal to waive the sanctions in question, as well as constitutional authority to make the nonbinding political commitment that the United States would not re-impose such sanctions under the terms of the JCPOA, so long as Iran kept its part of the bargain.183In addition to preexisting statutory sanctions authority, while the JCPOA was under consideration, Congress enacted the Iran Nuclear Agreement Review Act (“Corker-Cardin” bill), which did not undermine the President’s legal authorities, and arguably added to them.184 Nothing in the Corker-Cardin bill authorized a future President to violate the Iran Nuclear Deal in a manner inconsistent with international law.

But on May 8, 2018, President Trump announced that the United States would withdraw from the JCPOA. In essence, Trump announced that he would no longer exercise his statutory option to waive U.S. statutory sanctions, “instituting the highest level of economic sanction” on Iran and “[a]ny nation that helps Iran in its quest for nuclear weapons.”185 But his withdrawal alone has not killed the deal, nor has it eliminated the reasons for Congress to review the withdrawal.

At the international level, the other partners to the JCPOA—the Europeans, the Russians, and the Chinese—did not rush to default on their own obligations just because President Trump did. Nor did they return unilaterally to re-imposing sanctions on Iran. To the contrary, these other parties initially all responded by saying that despite the Trump Administration’s action, they still intended to comply with the Iran Deal. Thus, at this writing, the JCPOA remains fragile, but functioning. Trump acknowledges that his withdrawal will take until at least the end of 2018 to fully implement. Thus, the United States has again resigned from an international agreement, but not yet fully left.

At least until the end of 2018, Iran will likely keep fulfilling its JCPOA nuclear commitments to benefit from the continued sanctions relief from the United Nations and the European Union. And significantly, Trump’s May announcement of withdrawal did not state that the United States would unilaterally trigger the so-called “snapback” mechanism of U.N. Security Council Resolution 2231. That mechanism allows any permanent member of the Security Council—without a vote by the full Council—to cancel U.N. sanctions relief provided under that resolution within thirty days by claiming a violation. What probably constrained President Trump from doing so was that the other stakeholders would not likely agree that an Iranian violation has occurred. The IAEA has repeatedly reported that Iran is in compliance with the JCPOA.186 Trump’s lawyers plainly advised him that if the United States exercised the snapback provision, Iran could plausibly argue that the United States lacked a good-faith basis for accusing Iran of a breach, and that any diminution in Iran’s performance came in response to the U.S. unilateral withdrawal announcement.187

The irony, of course, is that the entire structure of the JCPOA rested on the once-reasonable assumption that Iran, not the United States, would be the likely violator. But instead of exiting the deal, Iran sued the Trump Administration before the International Court of Justice (ICJ), which recently issued a provisional measures order that could lead to a final judgment that the United States violated international law by unilaterally abrogating the JCPOA.188Iran asked the court to issue provisional measures on the ground that renewing sanctions would violate multiple provisions of a 1955 bilateral Treaty of Amity between the United States and Iran.189 The court ordered the parties to commit to “the non-aggravation of their dispute” and directed the United States to remove any impediments, resulting from re-imposed sanctions, previously lifted within the framework of the JCPOA, to the free export to Iran of humanitarian goods, equipment, and services to ensure civil aviation safety. Shortly after the court issued its order, the executive branch announced that it would terminate the 1955 Treaty of Amity with Iran in one year’s time,190 under the terms of Article XXIII(3) of the Treaty, and soon thereafter terminated the Optional Protocol on Dispute Settlement to the Vienna Convention on Diplomatic Relations.191

This legal background is doubly significant. First, historical practice does not recognize any right of the President unilaterally to breach an agreement in violation of international law. Even those who rely on recent historical practice only argue that it condones a President’s power to unilaterally terminate an agreement in accordance with its terms. Congress has undeniable constitutional authority to implement and monitor the lawful implementation of treaty commitments, as illegal implementation or termination inevitably affects many constituents and many facets of the national interest. To be sure, Iran’s legal claims before the ICJ against the United States have been challenged as weak, and may well not ultimately prevail.192 But as this episode illustrates, termination of a deal in a manner inconsistent with its terms can lead to cascading claims of international law violation and the termination of other international agreements. One of the principal aims of the U.S. Constitution was undeniably to give the federal government authority to comply with its international legal commitments.193 Thus, even in cases where the President alone may have the power to make an agreement that binds the United States, the Constitution still requires that an attempted termination of that same agreement be subject to congressional review under circumstances where that termination would render the nation as a whole externally liable for an international law violation that could implicate many areas of congressional responsibility.

Under this reasoning, Congress would have a strong claim to challenge the President’s authority to breach JCPOA unilaterally on the grounds that Iran, the other partner, had engaged in no such breach. Congress acquiesced in the adoption of the Iran Nuclear Deal through the delegated use of sanctions relief, but in so doing, anticipated that the President would grant or withhold the sanctions relief in accordance with the deal’s terms. Even when a President makes an international agreement without congressional participation, if a subsequent President’s disengagement from that same agreement would expose the United States as a nation to claims of international law violation, that disengagement should not be finalized without congressional participation and oversight.

Second, in the case of the JCPOA, the need for congressional participation is particularly strong, because the President is not exercising his own plenary constitutional powers, but rather, exercising delegated congressional foreign commerce powers to adjust sanctions in a case where the agreement partner has apparently complied with its international commitments. Congress could undoubtedly revoke its delegation and use its foreign commerce power to decline to re-impose economic sanctions on Iran despite Trump’s decision to withdraw, on the ground that it sees no basis for an executive claim of retaliatory breach.194 Or if, as Trump claims, he wants a better deal with Iran, Congress could entirely withdraw his statutory rights to waive sanctions in this setting and require—as a condition of giving consent to any new agreement with respect to Iran—a presidential undertaking to terminate or withdraw from that same agreement only in accordance with congressionally prescribed procedures, which could include specified modes of congressional notification, consultation, and participation. Now that the House of Representatives is under opposition control after the November 2018 midterm elections, one or both houses of Congress could now assert a constitutional right, under the mirror principle, to review and participate in the decision to withdraw from the Iran Nuclear Deal before it becomes final.195

V. regulating withdrawals going forward

This brings us to a final question: how best to regulate America’s withdrawals from international agreements going forward? In INS v. Chadha, the Court made clear that Congress cannot affect the rights of persons outside the legislative branch unless it acts by a joint resolution, which is enacted by both houses and presented to the President for either her signature or veto.196 But that formal constitutional rule does not prevent Congress from developing new and better ways to direct how presidential attempts at agreement termination or withdrawal should be managed in the future.

Congress could, for example, enact general laws prospectively limiting the President’s discretion to unilaterally terminate all Article II treaties and congressional agreements. Or, it could adopt specific statutes limiting that executive discretion with respect to particularly important named treaties. Either kind of legislative enactment would place any such unilateral termination into Youngstown Category Three, where the President’s “power is at his lowest ebb, for then he can rely only upon his own constitutional powers minus any constitutional powers of Congress over the matter.”197 Thus for example, Congress could pass a law ex ante opposing the President’s effort to unilaterally withdraw from a mutual defense treaty, like NATO, which does not involve the President’s plenary power.198 Alternatively, the Senate could impose a reservation, understanding, or declaration on new or existing treaties, limiting future efforts at unilateral presidential terminations unless the termination is plainly “for cause.”199

If Congress wished to address this issue more broadly, it could adopt a “framework statute” to govern all requests for agreement withdrawal or termination.200In such a framework statute, Congress could specify that the President may not terminate an important treaty or congressional-executive agreement without meaningful consultation with or notification of Congress, including the relevant congressional committees or congressional leaders. If the President claimed that an emergency situation required immediate disengagement from a particular treaty or agreement, Congress could revise its internal rules to enable him to submit the matter of withdrawal or termination to Congress for speedy congressional action, under a “fast-track” procedure of the kind used in international trade laws.201 If the President proposed to withdraw or terminate U.S. participation from a particular treaty regime, he could ask to have a bill introduced in Congress under these specialized, expedited “fast track” legislative procedures, which would require Congress to vote on a joint resolution of approval or disapproval of the action within a specified number of legislative days, without amendment or extended debate in committee. This procedure would, in effect, force Congress to approve or reject the proposed withdrawal or termination by affirmative expedited action, not merely by silence.

Executive branch lawyers should recognize that creating such a protocol to govern terminations or withdrawals, operating under such generic procedures, would serve the executive branch’s long-term interests. A legislated “agreement withdrawal” protocol would lift particular presidential decisions to terminate or withdraw out of Youngstown’s Categories Two and Three and into Category One, where they would be “supported by the strongest of presumptions and the widest latitude of judicial interpretation, and the burden of persuasion would rest heavily upon any who might attack it.”202 Such presidential actions would thus become endowed with the highest available degree of constitutional legitimacy, as opposed to the uncertain legal status that would inevitably accompany a series of ad hoc executive efforts to disengage from longstanding treaty arrangements through less formal means.


In sum, the conventional wisdom that the President possesses a general unilateral power to terminate or withdraw from international agreements is mistaken. A claimed transsubstantive rule of withdrawal and termination cannot explain why key multilateral treaties, such as the WTO, NAFTA, and IMF Articles of Agreement—which lie squarely within the zone of Congress’s exclusive foreign commerce power—should be terminable by the President acting alone, when a high degree of congressional input was constitutionally required to create those obligations in the first place. This Essay has argued that this issue is in fact governed by an agreement-specific “mirror” principle—reflected in early U.S. constitutional jurisprudence and the functional reasoning of parallel foreign decisions—that requires parity of constitutional authority for entry and exit. Absent exceptional circumstances, that mirror principle requires that the degree of congressional participation legally necessary to exit an agreement should mirror the degree of congressional and executive participation that was required to enter that agreement in the first place.

If nothing else, the foregoing analysis should make plain that Goldwater v. Carter cannot be considered controlling with respect to most of the termination or withdrawal scenarios that may lie ahead.203 The President possesses no general unilateral power of treaty termination. In future cases, the constitutional requirements for termination should be decided based on the type of agreement in question, the degree of congressional approval and subject matter in question, and Congress’s effort to guide the termination and withdrawal process by framework legislation. It would thus be grievous error for the President or his lawyers to assume that our Constitution confides in him alone the power entirely to disengage the United States from the post-World War II legal order.

Sterling Professor of International Law, Yale Law School; Legal Adviser to the U.S. Department of State, 2009-13; Assistant Secretary of State for Democracy, Human Rights and Labor, 1998-2001. I am grateful to Tess Bridgeman, Peter Cane, Sarah Cleveland, Bill Dodge, Kristen Eichensehr, Gene Fidell, Shikha Garg, Joel Goldstein, Adam Goldenberg, Dieter Grimm, Avril Haines, Frank Iacobucci, Steve Herz, Rebecca Ingber, Steven Koh, Maxine Lipeles, Hope Metcalf, Patrick Pearsall, David Pozen, Ashika Singh, Kate Stith, Phil Spector, Gib van Ert, Mike Wishnie, and the members of the 2018 Yale-Duke Foreign Relations Roundtable and the Yale Law School Faculty Workshop for their thoughtful comments and to Dylan Kolhoff, Wajdi Mallat, Catherine McCarthy, Lucie Olejnikova, Mitzi Steiner, and Danielle Zucker for their outstanding research assistance. I especially thank Stephen J. Pollak and Geoffrey Klineberg of the Historical Society of the D.C. Circuit, for giving me the chance to think through these issues by inviting me to reargue Goldwater v. Carter in a moot court before Judges Harry T. Edwards and Stephen F. Williams.204

Parts of this Essay derive from thoughts first expressed in my most recent book, The Trump Administration and International Law. This Essay completes a trilogy on Twenty-First Century international lawmaking and breaking that also includes Harold Hongju Koh, Twenty-First Century International Lawmaking, 101 Geo. L.J. 742 (2013), and Harold Hongju Koh, Triptych’s End: A Better Framework to Evaluate 21st Century International Lawmaking, 126 Yale L.J.F. 337 (2017).