Treaty rigidity and domestic democracy: Functions of and constitutional limits to democratic self-binding

Abstract International treaties create a layer of law that is particularly hard to change: While modern treaties entail significant constraints on domestic law-making, amendments require a new consensus of state parties. This creates a conflict with a core aspect of democratic constitutions: the power of new majorities to revise the laws made by their predecessors. In particular, several obligations of international economic law, namely regulatory stability promises in investment protection treaties, but also criminalization requirements under the UN drug control conventions face critique for overly diminishing domestic policy space. How can democratic constitutions deal with those tensions? One approach that is common in some countries accepts a power to legislate in violation of treaty obligations. But denying treaties the quality as real law in the domestic sphere risks to undermine the benefits of the instrument without truly solving the democracy concerns. Instead, democratic constitutions can be read to entail certain standards for engaging in treaties, namely (i) procedure-related duties to include denunciation clauses and political reform mechanisms in the treaty, and (ii) the need to establish a substantive justification for treaty constraints on future domestic law-making by legitimate aims of international constitutionalism or co-operation.


Introduction
Law-making in national democratic processes is now embedded in a dense network of international treaty obligations. The binding promises made to other state parties with the conclusion of a treaty constrain the options for future domestic politics. Several modern treaties are not just instruments that enable co-ordination and co-operation of states in a neutral way. While the consensual form of law-making has driven scholars to explain treaties by an analogy to private contracts, 1 the modern treaty regimes concluded among large groups of states rather function as a form of international legislation. Human rights treaties, for instance, do not follow the logic of do ut des, but function as 'supplementary constitutions'. 2 They were designed to ensure that individual rights guarantees cannot be abolished if an illiberal government gains power. 3 Moreover, several treaties establish regulatory regimes. According to Joseph Weiler the norms of this new 'layer' of international law have 'a far greater "direct" and "indirect" effect on individuals, markets, and more directly : : : come into conflict with social values'. 4 Those treaties oblige states to follow certain policy approaches which favour certain interests over others and may therefore be controversial in political debates. 5 It is no surprise, then, that some treaties face critique for undermining domestic democracy. This is particularly true for supra-and international economic law: The market freedoms in the EU treaties and their far-reaching interpretation by the ECJ, but also the GATT/WTO treaties and regional free trade agreements have been criticized for entrenching a neoliberal economic model of open markets that precludes democratic debates on economic policy. 6 In particular, the rights of private investors under free trade agreements like the EU-Canada CETA to claim compensation for certain regulations that affect their investments face strong academic critique as well as public protest for reducing domestic policy space. 7 In recent years, worries that treaties amount to excessive constraints on domestic democratic politics extends to another field: drug policy. The 1961 UN Single Convention on Narcotic Drugs 8 and the 1988 UN Convention against Illicit Traffic in Narcotic Drugs and Psychotropic Substances, 9 requiring the prohibition and criminalization of nonmedical and non-scientific uses of narcotic drugs including cannabis, 10 entrench a repressive approach to drug policy. In many countries, there is growing support for more liberal approaches. In Uruguay (2013) and Canada (2018), the political will to establish a regulated market for cannabis was so strong that they chose to enact new laws that openly disregarded the conventions. 11 Tensions between treaties and domestic democracy have been articulated in constitutional adjudication, too. They are at the heart of constitutional law doctrines that try to reduce the domestic impact of treaties. The German Federal Constitutional Court (FCC) recently argued that democracy implied that parliament must have the power to 'override' treaties in domestic law. 12 Such doctrines are worrying from the perspective of international law. They risk to undermine the very idea of the instrument of treaties, to enable effective co-operation of states by a self-binding of their institutions to the agreed norms. At the same time, the problem of excessive constraints on domestic policy-making from treaties can hardly be neglected by simply stating that it is the necessary consequence of concluding treaties.
The aim of this article is to go beyond pointing to a tension between treaties and domestic democracy. Self-binding of democratic institutions to treaties has important benefits which should not be sacrificed by diminishing their effect in domestic law. But under a democratic constitution, there may be standards for engaging in treaties that avoid overly tying the hands of future legislation. The article investigates how such standards can look like and how they can play a role in judicial review of treaties by constitutional courts, but also in treaty interpretation by national and international courts.
The article proceeds as follows: Section 2 explains why in an account of democratic politics that emphasizes the power of new majorities to revise decisions of their predecessors, the instrument of treaties creates a challenge. Section 3 critically evaluates the approach to solve democracy concerns by assuming a power of parliament to derogate treaties in domestic law. In Section 4, I outline the features of an alternative model how democratic constitutions can conceptualize the relation between domestic law and treaties. I argue that we can understand the treaty-making power as subjected to certain limits relating to the procedural dimension of sufficient denunciation rights and treaty reform mechanisms as well as to the substantive contents of obligations. Section 5 distinguishes two groups of treaties, for which these aspects are relevant in a different way: treaties that fulfil legitimacy-related constitutional functions, and treaties that facilitate international co-operation. Section 6 gets back to particular problems of treaty constraints on democratic legislation in certain fields: criminalization requirements namely in the drug control conventions, transnational economic rights and regulatory stability guarantees in investment protection treaties.

Treaty rigidity and dynamic democracy
The mechanism of treaties raises a challenge for democracy: It creates a layer of hard law which deviates from the core democratic idea that new majorities can revise the law.

Initial consent and democratic change
At first glance, democracy concerns about treaties can easily be dismissed. In contrast to customary international law, states are bound due to the prior consent of their institutions. Most democratic constitutions provide that treaties in legislative matters may not just be concluded by governments but require approval by parliament. 13 We can therefore speak of a self-binding or 'precommitment' of democratic institutions. 14 Consequently, there are calls for an originalist treaty interpretation, giving decisive weight to the travaux préparatoires. 15 However, the originalist approach does not only face doubts whether courts can always find clear answers to questions of interpretation from the text and the travaux préparatoires without any productive element, 16 and fits uneasily with the rules of the 1969 Vienna Convention of the Law of Treaties (VCLT) which in Article 32 only attributes the travaux the role of a supplementary means of interpretation. The legitimacy idea on which the originalist approach is based, that constraints on future democratic decisions can be justified by mere reference to a past democratic decision, neglects a core element of democracy: Law should not only be created in a democratic procedure but also be subject to revisions in new democratic procedures. Jefferson famously argued that every generation must be free to make its own laws: The living should not be bound by laws enacted by the dead. 17 But opportunities to revise the existing laws are equally important among the same citizen. One central feature of democracy is public deliberation which norms should be in place. Theorists like Jürgen Habermas emphasize the permanent character of deliberation. Decisions are necessary to settle practical problems, but they cannot be more than a 'caesura in an ongoing discussion'. 18 Another core element of democracy is majority rule. Since democracy is based on the equality of citizens, the standard must be decision by simple majority. Super-majority and unanimity rules give an advantage to the opponents of a proposed decision. Kelsen has pointed out that there is an important temporal aspect of this: 19 Deciding by majority rule implies that the minority can become the majority later on and revise previous decisions. If change was only possible by super-majorities or unanimity, there would be an unfair advantage to the status quo. From this perspective, there is a need to justify any 'higher' law that constrains majority decisions in the ordinary democratic process, while it is not in the same way subject to democratic change.

The difficulty of amending treaties
It is, of course, possible to amend treaties by a new consent of parties. But the unanimity principle gives a strong advantage to the status quo. Certainly, some regimes allow greater flexibility, e.g., when a majority of states in the plenary body can adopt revised provisions which will get valid for all states except those opting out. 20 However, changes made only among some states ('inter se') make the regime less coherent. States might refrain from proposing treaty reforms when there is no prospect for a new consensus. For instance, states favouring a less restrictive drug policy did not propose amendments to the UN conventions at recent General Assembly meetings on those conventions. 21 A more successful approach can be that like-minded states establish 'counterconventions'. The UNESCO Convention on the Diversity of Cultural Expressions has been established to challenge the treatment of cultural goods under the WTO regime. 22 Finally, some regimes establish secondary law-making mechanisms by majority. Beyond the EU, this holds particularly for technical matters like changing the list of substances in environmental agreements on chemicals and the drug control conventions. However, this only addresses the epistemic need for adapting the law to new factual developments or knowledge whereas changes of the political concepts enshrined in the treaties can only be achieved by unanimity. In drug policy, cannabis cannot be simply 'delisted' as it is included in the UN Single Convention itself (Article 28).
Even when political change of international regimes is possible, it is another question whether there can be a democratic process on the international level. One does not need to support a communitarian conception of democracy to acknowledge that it is difficult to establish participation in a public sphere beyond states. Certainly, a lot can be done to democratize international political processes. Instead of attributing governments a monopoly to represent diverse populations, it is possible, e.g., to give NGOs a voice. 23 Even so, important differences to domestic democracy remain. The most promising approach to institutionalize supranational democracy is the EU secondary law-making process, involving a parliament that represents the internal political diversity of member states. Supranational democratic fora may at least merge political discourses in the member states ('demoi-cracy'). 24 Yet, the democratization of EU law is limited as long as changing the treaties requires a consent of all member states. The dense substantive commitments in the treaties considerably diminish policy space for secondary law-making. 25

Dynamic flexibility without formal amendments
Despite the difficulties of formal amendments, treaties might leave some flexibility. International courts have often interpreted treaty provisions in an evolutionary way. 26  international organ competent for authoritative interpretation, domestic political actors may have broad leeway to argue that a new law is consistent with treaties. Subsequent state practice has to be taken into account for treaty interpretation under Article 31(3)(b) VCLT. Some treaties include specific flexibility clauses. For instance, the obligation under Article 3(2) of the UN Convention against Illicit Traffic in Narcotic Drugs and Psychotropic Substances to establish as criminal offence the possession, purchase or cultivation of narcotic drugs for personal consumption is 'subject to the constitutional principles and basic legal concepts' of their legal system. Some states argued that their new drug policy of establishing consumption rooms or tolerating 'Coffee Shops' was compatible to the convention due to that clause. 27 The International Narcotics Control Board contested that view; however, its interpretation is not authoritative. Flexibility can be even greater when other treaties, e.g., on human rights, are taken into account, as suggested by Article 31(3)(c) VCLT. 28 However, there are limits to interpretation. Under Article 31(1) VCLT any interpretation has to proceed in good faith in accordance with the ordinary meaning of treaty terms, their context and the object and purpose of the treaty. When Uruguay and Canada established a regulated market for cannabis, they had to acknowledge that this was a violation of the UN conventions which clearly aim at preventing any recreational drug use. 29 The argument of Uruguay's government that the new policy was realizing human rights obligations 30 does not help, either. When a regime conflict cannot be prevented by harmonizing interpretation, it can hardly be solved by hierarchy, which has a very limited scope in international law. 31 When international courts are competent for authoritative interpretation, the democratic benefits of an evolutionary interpretation are ambivalent. Certainly, international courts provide new opportunities to contest domestic policies, which might facilitate domestic political discussion in consequence. 32 But when they adopt interpretations that extend treaty obligations, options for majoritarian domestic policy-making decrease. Whereas in constitutional law, from time-to-time court interpretations have been countered by amendments, 33 the difficulty of amending treaties results in little prospects to politically overrule interpretations of international courts.
When subsequent state practice oversteps the limits of interpretation, it usually qualifies as a violation. True, there is some discussion on the possibility of tacit amendments to written treaty norms when states do not object incompatible actions by others ('acquiescence'). 34 However, it is controversial whether this overstretches the private law analogy and fails to take into account the objective law-making function of modern treaties, particularly multilateral treaties with mandatory dispute settlement. Even if we accept the possibility of tacit amendments, this does not provide a solution in most cases since there will usually be parties that object incompatible actions.

Unilateral withdrawal
States can regain freedom from treaty constraints when withdrawal is permitted. Most international regimes include denunciation clauses, but some do not, e.g., the UN Human Rights Covenants. Many investment treaties stipulate that withdrawal will only be effective after 10-15 years. 35 When withdrawing is not explicitly allowed, there is a presumption for pacta sunt servanda in the VCLT. Although one might argue that treaties with significant effects on domestic policy-making imply a withdrawal right due to their nature under Article 56(1)(b) VCLT, 36 international practice has handled this provision with great care. Notably, the UN Secretary General and the Human Rights Committee have denied North Korea a withdrawal right for the Human Rights Covenants. 37 The rebus sic stantibus exception under Article 62 VCLT may only be invoked in exceptional circumstances, not because of political change. 38 While withdrawal rights certainly mitigate the democratic concerns, the problem remains that exiting a treaty often faces practical difficulties. It can cause severe costs and a loss of reputation. 39 States may usually only leave a treaty all together. If a new majority opposes some specific obligations, this might be a price too high to pay, particularly taking into account that the state would lose a forum for dialogue. Still, there are sometimes options for post-ratification reservations or to combine withdrawal and re-accession with reservations. 40 In this way, Bolivia achieved an exception for the traditional use of the coca plant under the drug control treaties. 41 By contrast, Uruguay and Canada did not choose to evade a violation by withdrawing, probably because they feared to be placed in the category of pariah 'narcostates'. 42 A more promising approach might be collective withdrawal. In any event, it depends on a number of factors beyond the existence of a denunciation clause whether there is a real political option to exit treaty obligations.

A false promise: Powers to derogate treaties in domestic law
In light of the difficulties of amending treaties and withdrawing, political actors might resort to another strategy: disregarding treaties. Certainly, on the international level, states face consequences when their parliaments enact laws that violate treaties. The fact that a breach results from democratic law-making does not change anything about this. Beyond countermeasures in general international law, many regimes establish courts that can declare laws to be incompatible with the treaty and impose fines or compensation for individuals. But international law cannot prevent the domestic application of laws that violate a treaty. Reputation loss and economic costs might sometimes be taken for the expected benefits. Thus, it depends on domestic legal rules whether legislation has the power to enact laws in breach of treaties. In countries following the monist doctrine, courts usually deny to apply laws that violate treaties. 43 But even under the premise that the domestic legal order is based on its own validity criterion (dualism), this order can stipulate the primacy of treaties over legislation. Many constitutions contain provisions in that sense. 44 In consequence, constitutional courts can extend judicial review of legislation to treaty conformity. The picture is different in countries where a 'hard' form of dualism prevails and legislation is considered valid in spite of violating treaties. Since Uruguay 45 and Canada 46 figure among them, their cannabis legislation could be enacted even though it violated international law.

Primacy of treaties as a question of constitutional interpretation
In legal systems following hard dualism, this is often presented as a necessity. On the Triepelian premise of a strict separation of international and domestic law, 47 international norms can only be valid in domestic law on the basis of a domestic norm. If the constitution does not order the domestic application of treaties, they have to be implemented by specific legislation. In consequence, this legislation can be repealed later on independent of international obligations. The UK and most Commonwealth countries, where parliament is not involved in treaty conclusion, follow this model most clearly. 48 By contrast, under the US Constitution treaties concluded 'with the advice and consent' of the Senate are part of the 'supreme law of the land', along with the constitution and federal laws. 49 The Supreme Court holds that they have the same legal force as federal laws. If considered self-executing, they are applicable without implementing legislation; but Congress can repeal them by a new law. 50 In 1998, the court found that the Antiterrorism and Effective Death-Penalty Act had repealed potential rights of death row inmates under the Vienna Convention on Consular Relations. 51 In Germany, Article 59(2) Basic Law provides that parliament has to give prior consent to treaties in legislative matters 'in the form of a federal law'. According to the FCC, this entails the consequence that treaties have the status of a federal law in the domestic legal order. Treaties can be applied without further implementing legislation, but parliament is free to enact laws which contradict a ratified treaty, the lex posterior rule applies. 52 However, even if there is no explicit constitutional rule on the primacy of treaties over legislation, a capacity of parliament to repeal the domestic application of treaties is not self-evident. The text of Article 59(2) Basic Law only relates to the treaty-making process. In Belgium, Luxembourg and Switzerland, the constitution contains a similar procedural rule, 53 but courts accept that treaties enjoy primacy over legislation. 54 True, these countries follow monism. But also on a dualist premise, the primacy of treaties might, in the absence of explicit provisions, follow from an implicit constitutional rule. Kelsen argued that since constitutions transfer powers to make treaties, they take up the concept of treaties, i.e., the idea of consensual law-making. A law violating a treaty therefore had to be found illegal from a constitutional perspective, too. 55 It is indeed somewhat contradictory to assume that a constitution empowers the executive and parliament to engage in treaties and thus to make binding promises on the international level while it allows parliament to break these promises in subsequent legislation. And while international law might in principle be indifferent about the rules of domestic foreign relations law, it arguably does require that there are rules in place which ensure effective compliance. 56 Hard dualism is particularly unconvincing on constitutional grounds when the constitution entails commitments to international integration. In the German treaty override case, the Federal Fiscal Court argued that the unconstitutionality of the law that violated a treaty followed from the principle of the Basic Law's 'friendliness towards international law' ('Völkerrechtsfreundlichkeit'; an argument which the FCC rejected not because the principle did not exist but due to its reading of Article 59(2)). 57 Even in dualist states, a power of parliament to legislate in violation of treaties is the product of a certain constitutional interpretation for which there must be reasons beyond the plain text.

Sovereignty and democracy
A crucial reason why domestic institutions should be free to repeal treaties in domestic law has been seen in sovereignty. The concept traditionally implies that states areand remainfree to decide on their internal legal order, even if engaging in international treaties might also be seen as an expression of 'external' sovereignty. 58 In the UK, the idea of parliament's 'continuous' sovereignty means that it can neither be obliged to implement a treaty nor to keep the legislation: No parliament can bind its successors. 59 The U.S. Supreme Court, too, has based a power of Congress to repeal treaties in sovereignty: The last 'expression of the sovereign will' must control. 60 In Germany, sovereignty arguments have played a major role in relation to the primacy and direct effect of EU law. According to the FCC, sovereignty is preserved since the application of EU law has a basis in national law, namely the Act approving the treaties 61 (in conjunction with the constitutional rules on EU integration), which would be repealed if Germany should leave the EU.
The formal concept of sovereignty as states disposing over the valid law in the country does not bear very far, though: If the primacy of treaties over legislation follows from the constitution, states can end the primacy of all or some treaties by constitutional amendment. Legal systems which accept the primacy of treaties over legislation in their constitution are, therefore, as sovereign as systems which allow treaty derogation by legislation.
Still, sovereignty arguments are often not about this formal point but rather serve as a placeholder for the domestic constitutional value of democratic self-determination. 62 Since constitutional amendments are difficult to carry out in most countries, they are no easy solution for the problem of binding treaties. By contrast, when legislation in breach of a treaty is permitted, a simple majority can change the rules in the country. In this sense, the 'last in time rule' in the US has been welcomed for preventing 'an inflexible law which cannot be adjusted or modified through the political process'. 63 When the German FCC reaffirmed that legislation has the power to derogate treaties, it did not merely reiterate the argument from Article 59(2) Basic Law, but also highlighted the dynamic aspect of constitutional democracy: Power in democracy is always temporary in nature : : : This implies that, in accordance with the will of the people as expressed through elections, subsequent legislatures must be able to revise, within the limits set by the Basic Law, legislative acts undertaken by earlier legislatures : : : It would be incompatible with this concept if a parliament could bind subsequent legislatures during later parliamentary terms and limit their ability to rescind or correct past legislative decisions. Otherwise, political views would be set in stone. When parliament disposes over a power to legislate in violation of treaties, a decision to use it will involve considerations on possible international consequences which will be balanced against the benefits of a violation. Such balancing might be rationalized in political theory. A comment on the new Canadian cannabis policy argued that breaking the conventions was justified as the gain in democratic responsiveness (public opinion clearly favoured liberalization) outweighed the damage to the international rule of law. 65 German FCC Judge König even argued in her dissenting opinion to the recent treaty override decision that the legislator, when making use of a power to override treaties, had a constitutional obligation to strike a fair balance between the importance of the policy goals and fidelity to treaties; this should involve considerations about options to withdraw from obligations. 66 Nevertheless, this approach accepts the argument that democracy can trump international legality.
It is an open question whether assuming a power of parliament to legislate in violation of treaties can tackle the real problems of a shrinking space for domestic policy-making. Despite such power, treaties will still be relevant for law-making processes since political actors mostly try to avoid economic costs and reputation loss due to a breach. 67 In international investment law, states have strong reasons to avoid legislation for which tribunals will award compensation, as awards can be executed domestically, and arbitration and awards have negative impacts on the ability to attract investment. 68 Moreover, as legal norms, treaties make a claim to be obeyed not just for fear of sanctions, but for their authority. 69 When a treaty rules out a certain policy, the question is not just whether the policy is preferable but whether the country is willing to take the position of 'an international lawbreaker'. 70 The deeper reason for assuming a power of parliament to enact laws in breach of treaties seems to lie in a discomfort with accepting legally binding obligations on states in legislative matters at all. Doctrines of a power to legislate in breach of treaties may thus be seen as part of broader 'sovereigntist' tendencies in domestic foreign relations law (particularly in the United States) that discount the legal quality of international law. 71 4. Constitutional democracy as a limit on treaty-making powers How can constitutional law deal with democratic concerns about treaty rigidity in a more constructive way? If those concerns are about the very existence of obligations for domestic law-making, a constitutional doctrine can start here. While constitutional provisions on treaties clearly allow to incur international commitments that will have consequences on future lawmaking, fundamental democratic principles enshrined in a constitution may imply that these powers are subject to certain limits. Such limits are already widely accepted for substantive constitutional values like individual rights.
But isn't it paradoxical to assume a constitutional standard of democracy that limits democratic decisions to engage in treaties? Importantly, this would not oblige parliament to pursue certain policies or prohibit it from doing so. The standard only relates to the question whether the state may incur obligations to keep policies. Constitutional constraints for democratic majority decisions in the name of substantive aspects of democracy are common in other fields. For instance, laws that regulate elections and campaigns have to ensure fair chances for all political groups. 72 Similarly, the constitution may prevent present majorities from incurring excessive treaty obligations which undermine fair chances to change the law in the future.
In the following, I explore two key aspects of a constitutional democracy standard for concluding treaties: options for denunciation and treaty reform, and the contents of obligations. I will then discuss how courts can enforce the standard.

Denunciation clauses and reform mechanisms
As explained above, withdrawal rights mitigate democracy concerns. In consequence, constitutional democracy could require to include denunciation clauses. As to EU law, where derogation in domestic law is not possible, the German FCC argues that the exit clause included in the treaties since the Lisbon amendment clarifies that membership is reversible democratic self-binding rather than definitively giving up competences. 73 However, if the EU's secondary law-making process has to be qualified as a democratic process, too, the decision is concerned with sovereignty rather than with democracy. A 2005 decision of the French Conseil constitutionnel 74 that highlighted the importance of denunciation clauses for the constitutionality of treaties does not seem helpful, either. While the decision was concerned with substantive constraints on national law-making from treaties, the protocols to the ECHR and the ICCPR abolishing the death penalty, the short reasoning suggests that the problem was not the irrevocability as such, but that France would be bound 'even in the event of an exceptional danger for the existence of the nation', 75 i.e., that the exercise of national sovereignty could require capital punishment in emergency situations. 76 The Bavarian state constitutional court has made more specific pronouncements on the role withdrawal rights play for democracy when it assessed an agreement among all 16 German Länder to regulate gambling in a certain way. 77 The matter of gambling is a Länder competence; still, a uniform regulation was considered necessary in order to avoid regulatory competition. The court argued that the democracy principle of the Land constitution implied that denunciation of such agreements must regularly be possible after the duration of a legislative period. As a general rule, a newly elected parliament should at a certain point be able to legislate in a different manner than required by the agreement.
This approach is a valuable contribution as it does not only stress the relevance of denunciation clauses, but also suggests after what time denunciation should be permitted. A right to withdraw immediately after conclusion would neglect that treaties aim to achieve a certain stability. But enabling a newly elected parliament to depart from treaty provisions at a certain point strengthens the democratic input of elections. Even if elections are not the only way for citizens to influence politics, the duration of a legislative period can be taken as a reference for defining after what time options for change become more important.
An objection might be that denunciation is usually a power of the executive. Even where the constitution requires parliamentary approval of denunciation, 78 parliament cannot exit treaties on its own. The German FCC argued, therefore, that a power of parliament to derogate treaties was necessary irrespective of denunciation clauses. 79  problem should not be overstated. The Bavarian constitutional court simply argued that parliament can regain freedom from treaty obligations following a denunciation by the head of state.
Still, denunciation clauses might not always be required. The need for exit options will depend (i) on the contents of obligations, and (ii) on the existence of reform mechanisms in the treaty that may function as an alternative way to reflect political change in the law. The better the chances for treaty amendments in international political processes (e.g., majority decisions instead of unanimity) and the more those processes are democratized by involving various stakeholders, the more will it be acceptable for constitutional democracy that a treaty permits denunciation only after a longer period than one parliamentary term, or even not at all.

Substantive limits to the treaty-making power
In light of practical difficulties to exit treaties, a denunciation clause might not be enough. The constitutional democracy standard has to involve the contents of obligations, too. There are two possible ways to do so. Some constitutional courts have taken a negative approach: Certain subject matters must be regulated in domestic legislation and may not be the object of treaties. I argue that it is more promising to take a positive approach. Constitutional democracy could require a justification for binding domestic legislation to international obligations, which will depend on the aims of treaties and whether specific norms can be linked to them.

Negative approaches: Excluding treaties in certain policy fields
The negative approach has famously been developed by the FCC in the Lisbon judgment. According to the FCC, the democracy principle of the Basic Law prohibits to transfer to the EU certain policy fields that are particularly sensitive for domestic democracy. The list includes criminal and police law, tax law, social security law and 'decisions of particular cultural importance' as in family law, the school education system and dealing with religious communities. 80 This approach might be extended beyond EU law: safeguarding domestic democracy could imply a prohibition to incur treaty obligations in sensitive policy fields.
This would be a problematic approach, though. Apart from doubts whether it is possible to agree on a list of sensitive fields as a matter of principle, 81 we have to bear in mind that the FCC developed the approach for competence transfers to a supranational institution with its own policy-making process. In this context, it is somewhat plausible to distinguish fields that should in principle be regulated in the national democratic process from others that can be dealt with in the EU secondary law-making process, which might suffer from democratic deficits but still qualifies as a democratic process. It is a different thing when treaty norms do not transfer decision-making authority, but fix substantive decisions. Here, the FCC's approach would be too lax for non-sensitive fields: There would not be any limits to de-politization by treaties. And it would be overly strict to bar any treaty obligation in sensitive fields. Even the FCC does not object limits to policy-making in criminal law and the school education system (two of the sensitive fields) by human rights to fair trial and to inclusion of persons with disabilities. In general, a constitutional principle that rules out any international law in certain policy fields would weaken international co-operation. When sensitive fields are delineated as a matter of constitutional identity, they might differ from country to country, which would make it very hard to achieve international agreements. Reservations in order to avoid that a treaty touches sensitive fields might not be accepted by other states and could be prohibited under Article 19 VCLT for being incompatible with the object and purpose of the treaty. In the United States, debates on limits on the powers to make treaties and pass implementing legislation are primarily concerned with federalism. In the recent Bond case concerning the conviction of a microbiologist for attempting to poison her neighbour under a federal law implementing the Chemical Weapons Convention (CWC), the question arose whether Congress had unconstitutionally intruded upon state criminal jurisdiction. While the majority found that the statute did not reach the woman's conduct, Justice Scalia argued that Congress can implement a treaty only under its otherwise enumerated powers. 82 A constitution may of course leave implementing legislation to federal states when their competences are affected. But those laws have to respect the treaty. Not even Scalia argued that the constitution prohibits the federal government from concluding treaties in competence fields of the states. Such restriction would weaken international co-operation, at least when the constitution does not empower states to conclude their own treaties. To mitigate the problem of subnational entities being bound by treaties concluded by the federal government, it is possible to involve them in the treaty-conclusion process, as it is the practice, e.g., in Switzerland 83 and Germany. 84

Towards a positive approach
Instead of prohibiting treaties on certain subjects, an alternative approach could ask for legitimate aims of concluding treaties. In this respect, Justice Thomas's concurring opinion in Bond is more interesting than Scalia's. He argued that treaty-making under the US constitution was limited to 'intercourse with other nations (including their people and property)' and did not extend to 'purely domestic affairs'. 85 In the past, such limits were widely assumed in U.S. constitutional doctrine. The American Law Institute's Second Restatement of the Law on U.S. Foreign Relations of 1965 required an 'international concern'. However, current doctrine is mostly sceptical. The Third Restatement of 1987 did not include the subject limit anymore. This might be due to concerns for an overly inflexible rule. In particular, it seems difficult to consider human rights treaties to be covered by a rule excluding treaties on purely domestic affairs. 86 To counter arguments that opposed adhering to human rights covenants was one major aspect of Louis Henkin's critique of subject matter limits in the late 1960s. 87 Still, it is promising to assume a constitutional requirement that decisions to incur treaty obligations which constrain future domestic law-making must serve some legitimate aim. Constitutional doctrine may accept a great variety of legitimate international concerns, including human rights. Section 5 explores them in detail. What is problematic, by contrast, is when governments and parliamentary majorities agree on concluding treaties in order to immunize laws from change. 88 Of course, many treaties (e.g., on free trade or environmental protection) may be more in line with political ideas of the majority that concluded it than with those of the opposition. But as long as a claim for international concerns can be made, strategic benefits are merely a side-effect. Where no such claim can be made, constitutions should not allow dominant political actors in various states who share domestic policy goals to abuse treaties as a pretext for entrenching policies. 89 A constitutional prohibition of abusing treaties would not be to the detriment of international co-operation. To require a common aim of state parties is not concerned with specific constitutional rules of a single country, but with the general question what treatymaking is about, which can be asked under any constitution. 82 Bond v. U.S., 572 U.S. 844 (2014). 83 Art. 55 of the Constitution.

Judicial enforcement of the standards
Beyond providing guidance for political actors in the treaty-concluding process, constitutional standards may be used for judicial review of treaties. Constitutional review during the treaty conclusion process (ex ante) is explicitly provided for in several constitutions; in many other countries it is established in practice. 90 When the court finds that acceding to a treaty would be unconstitutional, the government may only conclude it after renegotiation or with a reservation (or after a constitutional amendment 91 ). Ex post constitutional review of treaties is established in many countries, too. 92 However, declaring a treaty void in domestic law for violating constitutional standards will usually not affect the international obligations under Article 46 VCLT. Still, constitutional courts have sometimes argued that a break of international law as interpreted by international courts may be necessary to preserve fundamental constitutional values. 93 Beyond open conflict, it is possible to issue a declaration of incompatibility which would oblige the government to withdraw or renegotiate the treaty. 94 When the constitutional problem of a treaty is not that it violates substantive guarantees like individual rights but that it goes beyond the permissible self-binding, the situation is somewhat special. Here, ex ante or ex post constitutional review of the treaty or the law that consents to it can only have a preventive function to clarify that future legislation is not constrained by the obligations. The standard might also be relevant when courts review a law that violates a treaty. As explained in Section 3, there are good reasons to assume primacy of treaties over legislation even in countries which lack an explicit constitutional rule, but this could be conditional on whether the conclusion of the treaty satisfied the constitutional democracy standard. Similarly, explicit constitutional rules on primacy of treaties could be subject to certain exceptions for democracy reasons.
It is of course not the task of judges to second-guess foreign policy. Judges reviewing treaties can defer to a considerable degree to political actors, granting them a margin of appreciation as to what obligations are necessary to deal with international concerns. But relaxing judicial review does not mean that political actors are not bound by constitutional standards. They are invited to discuss in the treaty-making process whether dealing with international concerns requires to adopt certain constraints on future law-making. Courts may intervene when there is no plausible connection to any international concern.
Moreover, the standard may be relevant for determining the scope of substantive obligations in treaty interpretation. National courts can choose a reading that avoids overstepping constitutional limits. In Bond, the majority opinion argued that the CWC and the federal law implementing the convention were not concerned with the abuse of chemicals for personal revenge; the notion of chemical weapons could be restricted to contexts of chemical warfare. The case is interesting beyond federalism problems in the US: To read the obligation to enact penal legislation under Article VII(1)(a) CWC in a way that narrowed legislative choices in dealing with common assaults that involve chemicals would be delicate for harmonizing criminal law without a justifying international concern. 95 Interpreting treaties in the light of constitutional democracy standards does not necessarily amount to the reading which entails the least obligations, the old 'in dubio mitius' approach which is inconsistent with the VCLT interpretation rules. 96 But these rules do not 90 For an overview see Mendez, supra note 13, at 95. 91 This has frequently been the case in France, which tends to undermine the normativity of the constitution, see N. Aloupi, 'The Conseil Constitutionnel's Jurisprudence on "Limitations of Sovereignty"', in Aust and Kleinlein, supra note 56, 161 As to treaty interpretation by international courts, an obligation to take into account fundamental principles of domestic constitutional law is not recognized so far (apart from the ECJ under Article 4(2) TEU). However, when treaties are among states that have taken international obligations to ensure democratic processes (namely human rights to political participation) it is possible to refer to them under Article 31(3)(c) VCLT. Even when it is not necessary to do so under the VCLT, it may at least be wise to choose interpretations that avoid conflicts with democratic principles that all or most contracting states share in their constitutions.

Legitimate functions of treaty constraints on legislation
A central aspect of the proposed constitutional standard for incurring treaty obligations that constrain future legislation is whether they contribute to achieve legitimate aims. Only occasionally, constitutions specify certain aims of treaty-making explicitly, particularly in the field of human rights. Constitutional doctrine must, therefore, rather build on the aims that are implicitly taken into account in rules empowering the political branches to conclude treaties. Against the suspicion that determining such aims as a legal criterion 'risks becoming arbitrary and inconsistent', 97 I would like to propose a rule that covers the varieties of modern international law, but sorts out problematic cases. There are two groups of reasons that justify constraints on domestic policy-making from treaties: International constitutionalism and co-operation to achieve common political aims.

Treaties fulfilling constitutional functions
In domestic law, the higher rigidity of constitutional norms compared to ordinary laws is connected to their function to organize the state and to ensure legitimacy. The basic idea of constitutional constraints on legislation is to compensate for potential legitimacy deficits of majoritarian politics. International norms can serve constitutional functions in that sense, too. 98 Admittedly, modern international law has been criticized for constraining domestic policy-making like constitutions while it lacks the legitimacy from constitutional politics. 99 The narrative that constitutions are the result of a decision of 'the people themselves', exercising constituent power, can hardly be transferred to the international level. 100 However, explaining constitutions in purely decisionist terms is a misconception. The claim of a popular decision is not only empirically not true for many constitutions. It is also hard to see that a democratic decision in the past is enough to justify constraints for present majorities. Rather, the legitimacy of constitutions is inevitably connected with the content of the norms to which ordinary law-making is bound. A similar point can be made for several treaties.

Types of 'constitutional' treaties
The most obvious example are human rights treaties. Many modern constitutions explicitly refer to international human rights and require to attribute them primacy over legislation. 101 Even where this is not the case, adhering to an international system of human rights is in line with the protection of such rights in the substantive part of the constitution. As Habermas has pointed out, rights that protect citizens' private autonomy should not be seen as a contradiction to, but rather as a necessary precondition of citizens' public autonomy in democratic processes. 102 International mechanisms add an important further layer of protection. An external control can subject societal views on marginalized minorities to an unbiased examination. 103 Human rights usually only establish a framework for politics, ruling out certain options, while it is not possible to derive specific laws from them. This even holds for most positive human rights obligations that invite states to protect individuals from harm by others, or to provide basic goods (decent working conditions, education, healthcare etc.), but do not specify the measures by which these aims are to be achieved.
Many human rights treaties are not limited to minimal guarantees on which there can be no reasonable disagreement. Some treaties include quite specific obligations, e.g., to segregate juvenile offenders from adults, 104 or to educate children with disabilities in the general school system. 105 And while many international rights consolidate rights that are already established in national constitutions, 106 others do not have a counterpart on the national level at least in some contracting states. Nevertheless, constitutional democracy does not require a rule that would restrict the range of permissible human rights obligations to abstract commitments or rights that are already recognized in the constitution. Certainly, the specific content of human rights is not to be derived from moral philosophy but results from political construction. 107 Still, there are good reasons to keep the process of developing legitimacy conceptions separate from the process of pragmatic policymaking to which those conceptions are addressed. International human rights treaties are typically drafted in thorough international deliberation by governments and stakeholder groups. 108 Another constitutional function of treaties is to safeguard democracy. International human rights to vote, to free speech and to assemble can enhance political participation. Moreover, treaties may prevent threats to democratic policy-making in other states. Sovereignty is only effective when states have control over internal socioeconomic dynamics and reasonable freedom from external interference. 109 The prohibition of the threat or use of force under the UN Charter is thus a cornerstone of a 'constitutionalized' international law. 110 But treaties also fulfil a constitutional function when they counter more subtle means of states to undermine democratic policy-making in others. While it is in principle a legitimate choice of democratic politics to incentivize multinational enterprises to make business in the country by granting them certain benefits, some forms of tax legislation have to be framed as harmful competition which undermine the effective fiscal sovereignty of other states. 111 This holds particularly for laws that enable companies to evade taxation in other jurisdictions by aggressive tax planning that artificially shifts profits. The measures of the OECD/G20 Inclusive Framework on Base Erosion and Profit 102 Shifting (BEPS), including the 2017 Multilateral Convention to Implement Tax Treaty Related Measures to Prevent BEPS, are therefore an important tool to protect democratic tax policies from harmful practices of other states.

Reforms to 'constitutional' treaties and the problem of unilateral exit
Even if there are good reasons for a certain stability of 'constitutional' treaties, the solutions reached at some point should not be beyond contestation. Moreover, certain interpretations by international courts face criticism for overly constraining domestic policy-making. 112 Domestic constitutions establish amendment procedures to enable reforms of the conception. What equivalents are there for 'constitutional' treaties? Denunciation clauses enable states to end their participation in the constitutional project. Yet, to allow states to give up their international commitments to human rights, democracy and prohibitions to harm other states contradicts the idea of international constitutionalism. The more promising alternative is to achieve reforms in international political processes. For instance, the ECHR is overseen by the organs of the Council of Europe, the Committee of ministers and the Parliamentary Assembly; NGOs is granted an advisory role. 113 Allowing amendments by a majority of state parties with effect for all is helpful, too. 114 Beyond formal amendment, political actors and social groups can engage in interpretation discourses and challenge the reading of certain norms by international courts. When there are realistic opportunities to reform 'constitutional' treaties on the international level, a denunciation clause is not essential. On the contrary, when a constitution refers to the participation in human rights treaties, it is plausible to qualify withdrawal as a constitutional amendment that must be passed in the respective special procedure. 115 A democratic argument might only be made for a constitutional rule requiring subsidiary denunciation clauses that may be used as a last resort when reforms on the international level fail. 116

Treaties facilitating international co-operation
The content-related, 'constitutional' justification of establishing a framework for domestic political decisions that enhances their legitimacy only holds for a minority of treaties. Other treaties include policy decisions themselves. Here, the treaty instrument enables states to engage in active co-operation in order to achieve certain aims. While international co-operation may sometimes be necessary for moral reasons or even as a duty under general international law or domestic constitutional law (this is recognized particularly in the context of climate change 117 ), in most cases, the aims for which states co-operate and the specific treaty commitments they take are a matter of political choice. Democratic constitutions certainly embrace such democratic decisions. But they might include certain limits which safeguard that future majorities are not excessively bound by some specific form of co-operation.

Reasons for establishing co-operative regimes
The idea of 'constitutional' treaties, preventing legitimacy deficits of national policy-making, even holds when treaty rules mainly benefit the states' own populations. When treaties do not fulfil 112 Tomuschat, supra note 2, at 1057. constitutional functions, a shared will of political actors to place rules out of the disposition of future majorities is not enough to justify constraints on domestic law-making. Rather, there must be some collective action problem that requires common rules. With this criterion, constitutional law doctrine can take up the idea of subsidiarity. 118 There are indeed many ways in which the regulations made in certain jurisdiction can be in the interest of people abroad, even when they do not directly subject them.
First, co-operative treaties can deal with transnational effects. While a 'constitutional' argument for treaties that rule out certain national laws for their negative transnational effects in terms of safeguarding democratic self-government in other jurisdictions can only be made in cases of clearly harmful practices, transnational effects are, in any event, a reason to negotiate international standards. Lax environmental policies can cause pollution across borders and adversely affect global ecosystems. Even beyond cases of tax laws that directly harm other states, regulatory competition to attract multinational enterprises can undermine the implementation of effective environmental and social standards. Habermas has put it like this: 'As states are becoming more and more entangled in functional interdependencies that reach across national borders, above all globalized markets and digital connections, there is a need for steering capacities that single nation-states are increasingly unable to meet.' 119 International co-operation is not confined to this. The rules of international regimes can also create transnational public goods which national regulation alone would not be able to achieve. The wish to support economic prosperity by creating common markets as in the EU or establishing global trade rules in the WTO is a typical example. 120 People abroad may also have immaterial interests in regulations. The protection of biodiversity under the Convention of Biological Diversity predominantly holds for the contracting states' own territories. The preamble of the convention refers to the preservation of biodiversity as a 'common concern of humankind'. Although there is a similar language used in the UN Framework Convention on Climate Change, the function is different: Transboundary effects of biodiversity loss in one country are not as obvious as climate change due to greenhouse gas emissions abroad. Rather, biodiversity is attributed an intrinsic value. 121 While states are in principle sovereign over their national resources, they owe protection of the global value of biodiversity to the international community.
Importantly, not all of those problems will make treaty constraints on national policy-making necessary. Co-operation can also be achieved by discussion forums, information exchange and technical and financial assistance for states that lack the resources to participate in joint projects. And the depth of treaty obligations may vary from establishing a framework of standards for national law-making to a full harmonization on the international level.

Procedural consequences
Co-operation treaties are more contingent than 'constitutional' ones. Here, political decisions in treaty-making are not merely about which obligations are to be derived from legitimacy ideas but concern the basic question what co-operative projects are to be pursued and which means of co-operation will be adopted. When there are serious negative externalities, particularly in the field of climate change, there may be a legal duty to engage in some form of co-operation. But in other fields, states should be (continuously) free to decide whether they want to take part in the common project of a treaty regime, e.g., on free trade. Here, it seems particularly important for reasons of constitutional democracy to include provisions on political reform mechanisms in a treaty as well as denunciation rights at least after a certain period of time.

Problematic fields for regulation by international treaties
The constitutional standard I propose does not exclude treaties in certain sensitive policy fields per se. Nevertheless, there are some types of treaties for which it is particularly difficult to see that constraining domestic democracy serves a legitimate aim. I will now take a closer look at the contested treaty obligations that I have mentioned in the Introduction.

Criminalization requirements and the drug control conventions
Criticism of the criminalization requirements in the drug control conventions must be seen in the general context of international criminal law. What legitimate aims are there to establish treaties that constrain domestic decisions about the criminal law?
Such treaties fulfil a constitutional function only in rare cases. The 'core crimes' under the jurisdiction of the International Criminal Court (genocide, crimes against humanity, war crimes and aggression) 122 are solely concerned with collective actions that threaten the peaceful coexistence of states. When human rights include positive obligations to effectively protect individuals from harm by others, this does not always require criminal law provisions. The European Court of Human Rights (ECtHR) only assumes a need for criminal sanctions when it comes to intentional attacks on life, physical and sexual integrity. 123 By contrast, positive obligations in respect to deaths or serious injuries caused by negligence, 124 or in respect to other rights, e.g., property, 125 can be met by civil and administrative remedies, too. Proposals to decriminalize petty thefts therefore do not raise human rights concerns. The ECtHR did not require states to maintain the controversial criminalization of abortion, either. Even if the right to life extended to foetuses, the states had discretion in striking the balance between the women's interests and protection of the foetus. 126 More extensive international obligations to establish criminal sanctions may be justified on international co-operation grounds. The 2000 UN Convention Against Transnational Organized Crime contains criminalization requirements relating to organized crime, money laundering, corruption and the obstruction of justice (Articles 5, 6, 8 and 23). 127 But it does so only for offences that are transnational in nature (Article 3). A crucial objective of the convention is to prevent safe havens for organized crime. The harmonization of national criminal law is important for the provisions on practical co-operation by extradition and legal assistance, since obligations to do so depend on double criminality (Articles 16.1 and 18.9).
By contrast, the penal provisions of the drug control conventions are not restricted to international drug trafficking. The obligation to establish as criminal offence any recreational drug use also extends to persons cultivating cannabis for their own consumption. The preamble of the 1961 Single Convention does not point to specific transnational problems, but declares that 'addiction to narcotic drugs constitutes a serious evil for the individual and is fraught with social and economic danger to mankind', which is why states have 'the duty to prevent and combat this evil'. The regime reflects a repressive approach to drug policy which was dominant at the time of its adoption, particularly in the United States. 128 Of course, there are arguments in favour of this approach. But it is problematic to entrench it in treaties, which can in consequence be used to end discussions whether a more liberal approach should be adopted. 129 This critique could be a basis for a constitutional argument that the criminal law obligations went beyond legitimate aims of binding the hands of future legislation.
How could this problem be tackled in international law? The best solution would be to change the treaties in order to allow states more flexibility for their drug policy. 130 However, a new consensus is difficult to reach. Some authors have argued that instead, like-minded states could agree on an inter-se modification. 131 In this respect, constitutional democracy arguments can support a reading of the conventions not as absolute prohibitions, but as a conjunction of reciprocal obligations. Inter-se modifications are, then, permissible under Article 41(1)(b)(i) VCLT as long as they respect the rights of other states under the original treaty. States that want to establish a regulated market for cannabis could fulfil this condition when they try to isolate that market from countries maintaining the repressive approach.

Transnational economic rights
Transnational economic rights have been subject to much discussion as to their effect of economic liberalization, in contrast to constitutions that are often conceptualized as neutral in socioeconomic policy. 132 The most important transnational economic rights are probably the guarantees of free movement of goods, services and capital under EU law. While EU law differs from classical international law in many ways, this holds particularly for secondary law-making by EU institutions. The market freedoms are part of the primary law of the treaties. They thus amount to a 'negative integration' that eliminates certain options for domestic policies even when no agreement on a 'positive integration' by common regulatory standards is achieved in the EU's political law-making process. 133 The ECJ interprets the market freedoms in a broad way, targeting not only discriminations of foreign goods and services, but any measures likely to prohibit, impede or render less attractive their exercise. 134 Can those rights be explained as serving a constitutional function by protecting human rights, as some scholars 135 and the ECJ 136 have suggested? A human right to transnational business activity might be defended on a Lockean account of rights as protecting 'liberty and property' from state interventions. The history of the U.S. Supreme Court's Lochner era has demonstrated the danger of rights that result to be one-sided privileges of economic power-holders against democratic policies aiming at social justice. 137 The Kantian conception on which human rights declarations have been based starts from a different idea, respect for human dignity. 138 On that basis, it is hard to see how rights to free movement of goods, services and capital could qualify as human rights. To point to the fact that foreigners whose activities are affected by economic policies have not been represented in 128 Bewley-Taylor, Blickman and Jelsma, supra note 27, at 17. 129 Ibid., at 27. the democratic process that led to their adoption 139 does not provide a justification for transnational rights, either. Enhancing the 'virtual representation' of foreigners cannot make up losses of actual inclusive democratic policy-making on the domestic level. 140 Rather than serving constitutional functions, transnational economic rights are an instrument for creating a common market, which serves the collective aim of enhancing economic welfare by international co-operation. 141 A common market implies that goods and services can be offered under equal conditions in all member states. This rules out protectionist policies. A common market does not require, though, that companies can use the market freedoms to combat any national regulation that also applies to transnational business activities as long as common standards have not been adopted. The interpretation of transnational economic rights can bear this in mind. The ECJ has developed some approaches to avoid an over-reaching protection, carving out non-discriminatory sales modalities 142 and accepting unwritten legitimate policy aims that justify restricting the market freedoms, if pursued in a proportional manner. 143 Nevertheless, some decisions have shown a bias for business interests, namely the Viking and Laval rulings, 144 which are criticized as 'the EU's Lochner moment'. 145 The most consequent approach would be to understand the market freedoms merely as prohibitions of direct and indirect discriminations. 146 The ECJ could base an interpretation of the fundamental freedoms that allows sufficient national policy space for business regulation on a systematic argument, the EU's obligation to respect the constitutional identity and essential functions of member states under Article 4(2) TEU. The court has already referred to that provision in the context of the application of substantive treaty norms. 147

Obligations to regulatory stability in investment treaties
Treaty constraints on legislation raise particular democracy problems when they entail duties to regulatory stability, i.e., restrict options to change existing laws. Numerous investment protection treaties include obligations to 'fair and equitable treatment' (FET) and to compensate 'indirect expropriations'. Some arbitration tribunals competent for investor state dispute settlement (ISDS) have derived far-reaching obligations to compensate regulatory change from those provisions. This has given rise to concerns about a 'chilling effect on regulation'. 148 Philip Morris's arbitration against new tobacco legislation in Uruguay and Australia, 149 and Vattenfall's against the German exit from nuclear energy 150 have caused particular worries.
Against this background, the ECJ accepts democracy as a standard to review whether the investment protection chapter in the CETA agreement with Canada was compatible with EU primary law. The court argued that the autonomy of the EU legal order would be violated if arbitration tribunals, when evaluating restrictions on the freedom to conduct business, could call into 139 question the level of protection of a public interest that led to the introduction of such restrictions by the Union in the democratic secondary law-making process. 151 Nevertheless, to require a certain degree of regulatory stability can serve a constitutional function. Beyond prohibitions of ex post facto law, the ECtHR has also established constraints on changing laws with effect for the future. For instance, the court argued that although the Convention did not grant a right to social security in the first place, modifications to existing entitlements based on compulsory contributions had to be consistent with the right to property, which meant that states had to take into account the grievance the change in the law caused to the affected persons. 152 The ECtHR found that frustrating legitimate expectations created under the old rules without any transitional measure was disproportionate. This approach leaves room for democratic change as it (i) only establishes limits for extending reforms to existing legal rights, and (ii) legislation does not need to uphold them forever, but only has to consider the situation of those affected by change as a temporal aspect of proportionality.
When arbitration tribunals interpret investment treaties in a similar way, they can avoid excessive constraints on states' ability to change regulations which would get into conflict with democratic principles of domestic constitutions. 153 As to the FET standard, recent decisions tend to reject claims that it entails an obligation to strict regulatory stability. Compensation is not granted for any regulatory change but only when it is considered unreasonable. 154 When in Spain subsidies for renewable energies were drastically cut for existing plants, companies that had been incentivized to invest by a law guaranteeing a fixed feed-in tariff for 20 years suffered dramatic losses. The arbitration tribunal emphasized that Spain had the right to adapt its legislation, but found that the companies could not expect such dramatic change. 155 What is particular about protecting investors from drastic change is that often stability promises are made in the first place. To a certain degree, incentivizing investment by such promises might be justified on democratic terms. When policy goals like changing the energy supply system from fossil to renewable can only be achieved in the long term and need the involvement of private actors, policy-makers should dispose over instruments to make credible promises that investments in line with the policy aims will not be frustrated. Generally, the political decision to seek economic benefits by attracting foreign investment will be more effective when credible promises can be made. That political actors should dispose over that option does not mean, though, that they should be able to completely tie the hands of legislation. Therefore, we need to specify what legitimate expectations investors have for fairness reasons. 156 It is a different thing to expect that new regulations will not completely frustrate an investment incentivized by the state than to expect that they do not affect pay-off at all.
The obligation to compensate 'indirect expropriation' may also be interpreted to be concerned with fairness rather than requiring compensation for any loss in market value suffered by investors due to a new regulation. The 'sole effects doctrine' employed by some tribunals, under which compensation had to be granted irrespective of the policy aims except in rare cases of traditional 'police powers' of states, 157 fails to acknowledge the fundamental difference between regulations of 151 property use and genuine expropriations. Whereas owners are affected by regulations due to social effects of their property use, it is by accident that someone's land is needed for a public project. 158 A different approach increasingly used by arbitration tribunals 159 requires compensation only when regulations would otherwise place burdens on affected companies that would be disproportionate even in the light of the policy aims. While it is in principle the task of political processes to balance public and private concerns, international adjudication can sort out cases where the result is obviously unfair, e.g., when a new regulation makes any economically viable use impossible without providing compensation. The ECtHR uses a similar test. 160 In the light of strong criticism, international investment protection law has evolved. 161 While there are still reasons to be sceptical of the ISDS system due to the institutional feature that companies select part of the arbitrators, 162 the substantive standards in new treaties take into account concerns about excessive constraints on domestic democratic policies. They define more specifically what is required under the FET and indirect expropriation standards and highlight the states' right to regulate. In particular, CETA clarifies that 'the mere fact that a Party regulates, including through a modification to its laws, in a manner which negatively affects an investment or interferes with an investor's expectations, including its expectations of profits, does not amount to a breach of an obligation' (Section D Article 8.9). A FET violation may only be assumed when one of the cases of a list applies, including frustration of legitimate expectations created by a specific representation made to an investor (Article 8.10). Non-discriminatory regulatory measures do not constitute indirect expropriations except in rare cases of manifestly excessive impacts (Annex 8A no. 3). This was sufficient for the ECJ to conclude that CETA did not undermine the EU's democratic process. 163

Conclusion
Modern treaties entail important constraints on domestic law-making. For constitutional democracies, this raises concerns: The core element of democracy that new majorities can change the law is not effective when the democratic decision to be bound by a treaty can only be revised by a new consensus of all state parties. Principles of constitutional democracy are at the core of the common approach to assume a power of parliament to 'override' treaties in domestic law. But 'hard dualism' undermines the treaty instrument which constitutions recognize in their text, while it ignores the political effects of international obligations even when they are not formally binding in domestic law. The alternative approach that I have proposed in this article is based on the idea that democratic constitutions include certain limits for incurring treaty obligations in order to avoid excessive constraints on future majorities. This standard may imply a procedure-related duty to include a denunciation clause or adequate mechanism for treaty reforms. As to the substance of obligations, it is hardly plausible to assume a constitutional prohibition of treaty obligations in certain fields. But constitutions may require that obligations help to achieve legitimate aims, either preventing legitimacy deficits of domestic policies ('constitutional treaties'), or facilitating international co-operation, inter alia by providing a way to deal with transnational effects and 'common concerns of humankind' or by creating international public goods. By contrast, democratic constitutions can be understood to prevent political actors from abusing the form 158 of treaties to merely entrench policies. The examples of obligations to maintain a repressive drug policy, but also overreaching interpretations of transnational economic rights in EU primary law and of regulatory stability requirements in investment protection treaties demonstrate that treaties can pose dangers to the openness of the democratic process over time. It is the task of political actors concluding treaties and judges applying them to consider whether specific obligations can be based on reasons that are compatible with the democratic constitutional structure.