Guarantor Institutions

Abstract The last few decades have seen a proliferation of constitutional institutions, especially in the Global South, that do not neatly fit within any of three traditional branches of the state. These supposedly ‘fourth branch’ institutions may include electoral commissions, human rights commissions, central banks, probity bodies such as anti-corruption watchdogs, knowledge institutions such as statistics bureaus and census boards, information commissioners, auditors general, attorneys general and so on. In this paper, I will argue that some of these new institutions are best understood as “guarantor institutions”. I will show that in a given political context, a guarantor institution is a tailor-made constitutional institution, vested with material as well as expressive capacities, whose function is to provide a credible and enduring guarantee to a specific non-self-enforcing constitutional norm (or any aspect thereof). Section I explains why polities need credible and enduring guarantees for specific norms, and claims that the expertise, independence, and accountability of guarantor institutions are likely to be key ingredients that determine their effectiveness in serving that purpose. It also argues that constitutional entrenchment of the guarantor institution is entailed in the independence requirement. Section II shows that in order to credibly and enduringly guarantee a norm, certain primary and secondary duties need to be discharged by relevant actors in relation to the norm's content as well as its impact. It further argues that while some of these duties may be performed by institutions that possess expressive capacity alone (roughly, the capacity to speak, express, communicate), others require material capacity (i.e. the physical capacity to effect material changes in the world). Guarantor institutions, unlike integrity institutions, can shoulder primary as well as secondary duties. Furthermore, they are typically vested with expressive as well as material capacities, which is key to their classification-defying hybridity. Section III argues that guarantor institutions are constitutionalised in two respects: the norm they seek to guarantee is constitutional, and the institution itself has constitutional status. What matters for a norm or institution to be constitutional is that it is entrenched, i.e. protected from change from the ordinary political and legal processes of the polity to some extent. It is their doubly constitutional character that distinguishes guarantor institutions from ordinary regulators. Section IV explains how some constitutional norms are non-self-enforcing, in the sense that powerful actors are likely to have the will as well as the capacity to frustrate or erase them. It also shows that the three traditional branches, whether acting severally or jointly, cannot provide a credible and enduring guarantee to all non-self-enforcing constitutional norms. Hence the need for constitutional guarantor institutions. Section V highlights that guarantor institutions are typically tailor-made to guarantee specific constitutional norms. Their specificity has important consequences for their internal design and their mode of functioning, which distinguish them from key institutions in the three traditional generalist branches. Section VI concludes. Attention to guarantor institutions by constitutional scholars may help the discipline escape its blinkered worldview, which sees judicial review as the only game in constitution-town.

than elect, the personnel of these institutionsdespite their frequently overt political functionsbut do so through mechanisms that often give a significant, if not equal, say to opposition political parties (usually with some judicial and civil society input too). 11 Unlike other, sub-constitutional regulatory agencies, the goal appears to be to not so much depoliticise their functioning as to 'de-partisanise' them. To the extent that they are self-consciously 'political' actors and typically subject to political oversight, guarantor institutions don't exactly constitute a 'technocracy'. This postpartisan character of guarantor institutions (at least in democratic constitutions) may well be important for their legitimacy-they are likely to be regarded as legitimate only when they have the confidence not only of the current winner of the electoral race, but also of at least some of the 'losing' parties in the opposition (who nonetheless represent a considerable section of the electorate).
In this paper, I will argue that some of these new 'fourth branch' institutions are best understood as 'guarantor institutions'. I will show that in a given political context, a guarantor institution is a tailor-made constitutional institution, vested with material as well as expressive capacities, whose function is to provide a credible and enduring guarantee to a specific non-self-enforcing constitutional norm (or any aspect thereof). This is a conceptual paper seeking to explain how we should understand guarantor institutions. Whether having a particular guarantor institution is desirable or not is another matter, and will probably depend on several contextual factors. To make normative claims, we would need to know whether creating an effective guarantor institution will result in a net reduction of risk, whether it would be a good thing all things considered, and whether there are better alternatives. But first, we need to outline the contours of these recently constitutionalised institutions conceptually. Another related issue that I am bracketing off for another paper is whether guarantor institutions together constitute a guarantor branch of the state. Tentatively, I believe they may do so in some jurisdictions, but not necessarily in every polity they exist in. Details of the argument will have to await future work. This paper is structured thus: Section I will explain why polities need credible and enduring guarantees for specific norms, and claim that the expertise, independence, and accountability of guarantor institutions are likely to be key ingredients that determine their effectiveness in serving that purpose. It will also claim that independence usually entails the constitutional entrenchment of the institution concerned. Section II will show that in order to credibly and enduringly guarantee a norm, certain primary and secondary duties need to be discharged by relevant actors in relation to the norm's content as well as its impact. It will further argue that while some of these duties may be performed by institutions that possess expressive capacity alone (roughly, the capacity to speak, express, communicate), others require material capacity (i.e. the physical capacity to effect material changes in the world). Guarantor institutions, unlike integrity institutions, can shoulder primary as well as secondary duties. Furthermore, they are typically vested with expressive as well as material capacities, which is key to their classification-defying hybridity. Section III will argue that guarantor institutions are constitutionalised in two respects: the norm they seek to guarantee is constitutional, and the institution itself has constitutional status. The term 'constitution' includes both big-C constitutional codes and small-c constitutional statutes, conventions and practices. What matters for a norm or institution to be constitutional is that it is entrenched, i.e. protected from change from the ordinary political and legal processes of the polity to some extent. It is their doubly constitutional character that distinguishes guarantor institutions from ordinary regulators.
Section IV will explain how some constitutional norms are non-self-enforcing, in the sense that powerful actors are likely to have the will as well as the capacity to frustrate or erase them. It will also 11 See, eg, Constitution of South Africa, s 193 (5) show that the three traditional branches, whether acting severally or jointly, cannot provide a credible and enduring guarantee to all non-self-enforcing constitutional norms. Hence the need for constitutional guarantor institutions. In Section V, we will see that guarantor institutions are tailor-made to guarantee specific constitutional norms. Their specificity has important consequences for their internal design and their mode of functioning, which distinguish them from key institutions in the three traditional generalist branches. Section VI concludes.
Whether guarantor institutions are fit for purpose remains to be seen. But constitutional studies have been hampered by its lawyerly blinkers that tend to ignore all constitutional actors other than the judiciary, with non-judicial actors at best being considered in the context of judicial review of their actions. It is time scholars of constitutional studies begin attending to all dimensions of constitutional practice, especially practice in the oft-ignored Southern jurisdictions, to inform theoretical frameworks. This paper is a step towards such disciplinary expansion.

Credible and enduring guarantees
The central claim of this paper is that we should understand at least some of these emerging institutions in the so-called 'fourth branch' in some constitutional states as guarantor institutions. In a given political context, a guarantor institution is a tailor-made constitutional institution, vested with material as well as expressive capacities, whose function is to provide a credible and enduring guarantee to a specific non-self-enforcing constitutional norm (or any aspect thereof). Why might a polity need credible and enduring guarantees though?
As beings blessedor cursed, depending on your point of viewwith the ability to imagine our possible future selves, humans have always sought to contain the risks inherent in the future's ultimate, radical, unknowability. Thus, even though we cannot ensure that our children will survive us, we try to ensure they have access to good health care to make that more likely. Similarly, by putting money away in a pension pot, we make it more likely than it would be otherwise that we will be provided for when we retire. Such strategies of risk reduction operate as a guarantee understood in its colloquial sense. When a person claims that "I can guarantee that I will be adequately provided for when I retire", asking 'why' or 'how' is not meaningless. A reasonable response would entail the outlines of some risk mitigation strategy, pointing to savings, wealth, social security, likely inheritance or some such source of funds outside employment. This colloquial sense of a guarantee eschews the technical, private law, sense where a guarantor conditionally promises the lender to pay off the loan in case the borrower defaults. In ordinary language, a guarantee is meaningful even without a promise made to another person, nor does it require some 'breach' or 'default' by a third party before the relevant obligations kick in.
Constitutionalizing a state, like any other human activity, is temporally risky; framers do not know at the time of making a constitution whether the norms they seek to commit the polity to through the constitution will be respected over time. Most sincere framers of non-interim constitutions, however, intend their creation to endure, and are therefore invested in risk-management. Furthermore, the nature of constitutional politics is such that participants take the constitution's aspiration to endure into account when determining their strategic attitudes towards it. 12 The higher their expectation that the constitution will endure, the greater are the chances that the participants will structure their decisions in terms of constitutional rather than ordinary politics. 13 The shape of any resultant constitutional code is likely to be remarkably different, based on whether it was an outcome of the former or the latter. In ordinary democratic politics, for example, the winner (i.e. the majority group) is often allowed to take all. The winner is less likely to take all in consensus-oriented constitutional politics because minority groups are more likely to have a veto over the final agreement. Guarantees (in the colloquial sense) for constitutional norms against future risks are therefore not merely things that framers prefer to have-the character of the process of constitution-making depends very much on whether participants in the process see these guarantees to be sufficiently credible over time. Credible and enduring guarantees are therefore important in ensuring that constitution makers adopt longer time horizons.
A constitutional guaranteeas a risk-management toolhas to be credible. Although a guarantee does not (and usually cannot) completely eliminate all risks, it is sensitive to the level of risk mitigation. Not every risk reduction strategy would qualify as a 'guarantee' in its ordinary usage: we would expect a sufficient reduction in risk, so that at least ordinary or reasonably foreseeable risks are eliminated or significantly reduced. Thus, the scale of risk reduction is subject to a threshold requirement of effectiveness to qualify as a credible guarantee. Fake, insincere, or incompetent guarantors are no guarantors at all.
Institutional credibility is not static-it can be enhanced or diminished over time. Nor is it a function of the institution's design alone. Political cultures matter a great deal. Providing institutional guarantees in a relatively stable, peaceful, egalitarian, affluent society with a high degree of political trust and low levels of corruption is likely to be a lot easier than in a post-conflict transitional context marred by poverty, inequality, and mistrust. If anything, effective design matters more rather than less in the latter context. In either case, it is not enough that the initial design of the institution is credible-it must remain credible over time. A guarantor institution can, therefore, become an ordinary regulatory agency by losing credibility, 14 and a formerly ruling-party-controlled regulator could, over time, acquire enough effectiveness and credibility to qualify as a guarantor institution.
It is likely that the effectiveness of the guarantor institution will be a key factor influencing the credibility of the guarantee. Apart from the polity's general cultural attitude to its constitutional norms, three key design ingredients that will probably inform an institution's effectiveness are (i) its expertise in providing the said guarantee, (ii) its independence from actors who are likely to have the ability and the willingness to frustrate the norm in question, and (iii) its accountability to actors who are likely to have the ability and the willingness to secure the norm being guaranteed. Expertise for guarantor institutions is not limited to technical expertise: as institutions often imbricated in mega-politics, 15 their leaders also need significant political nous. Concerning the necessary independence of guarantors, Kelsen rightly insisted that "no institution is less suitable to perform [the guaranteeing] task than the one upon which the constitution confers the exercise … of the power to be controlled, and which therefore has the best legal chance as well as the strongest political motive to violate the constitution…" 16 Therefore, in most democratic contexts, guarantor institutions will typically need independence from the ruling party/coalition; in other words, in order to be sufficiently independent to provide credible and enduring guarantees, guarantor institutions typically need to be constitutionally entrenched (whether legally or politically). Depending on the norm in question, independence from other actorssuch as large corporationsmay also be required. Actors most likely to have a vested interest in the sound functioning of guarantor institutions include the political opposition, civil society organisations, sub-state or supra-state political organisations, independent media and others. The accountability dimension is critical for the institution's expertise as well as independence-if captured or compromised by a conflicted actor, a guarantor institution not only fails to guarantee the norm, it may even lend institutional legitimacy to the norm's evisceration by hostile actors. 17 It is the need to provide a credible and enduring guarantee for certain (non-self-enforcing) constitutional norms that necessitates guarantor institutions in addition to the three traditional branches of the state. The political executive and the legislature are not independent enough to have the will to guarantee norms that curb or limit the powers of the ruling party/coalition; the judiciary is often insufficiently accountable and lacks both the necessary expertise and the material capacity to guarantee certain norms. 18 Thus, while all constitutional actors have a duty to respect all constitutional norms, only some of them have the additional function of guaranteeing them.

Guarantee versus integrity
To properly appreciate what it takes to guarantee a norm by an institution (or a set of institutions), we need to triangulate three different sets of classifications. In the first set, we distinguish between guaranteeing the content of the norm from the norm's real-world impact. Secondly, each of these dimensions of a norm requires the imposition of primary and secondary duties: primary duties are those imposed directly by the norm and therefore necessary to protect or effectuate its content or impact in the first place; while secondary duties consist of redress mechanisms, which come into play only when the primary duties are breached. 19 In the third set, we can distinguish between those primary and secondary duties that can be performed through expressive capacity from those that require material capacity. So long as the duty-bearer can speak and communicate, they satisfy the minimum requirement for performing expressive tasks. Expressive capacity can be manifested in formal expression (such as pronouncement of a judgment or enactment of a statute) as well as through informal expression (such as a minister's press conference or a speech by the chairperson of the human rights commission). For material capacity, on the other hand, the duty bearer must be able to perform physical actions in order to effect material changes in the world. The complex intersection of these three sets of factors is depicted in Table 1: the second and third rows distinguish between the content and the impact aspects of a norm, the second and third columns separate primary from secondary duties, and the two coloursyellow and purplesignify duties that need expressive capacity and material capacity respectively for their discharge.
With respect to the content of a norm, there are two main primary duties: first, to respect the norm, i.e. to refrain from weakening or erasing it; second, to nourish (i.e. update and strengthen) it so that it does not lose its normative relevance under changing circumstances. Only norms authored or adopted (i.e. positively expressed) by human institutions (such as legal or religious norms) are amenable to weakening or repeal by them. Moral norms may well change because of change in context or circumstances; while human will can bring about a change in the circumstances and thereby indirectly affect the content of moral norms, such norms are not amenable to direct amendment by the exercise of any human will. In order to protect a guaranteed norm against weakening or erasure, the primary duty of relevant actors (against whom the guarantee applies) is to respect the expressive content of the norm by refraining from weakening or erasing it, and to nourish it-if necessary-to keep it relevant to changing circumstances. Interpretive questions about what amounts to a weakening or an erasure of a norm no doubt remain, but the general point should be easy to grasp. 17 On the dangers of unaccountable guarantor institutions, see Mohsin Alam Bhat, 'Between Trust and Democracy: The Election Commission of India and the Question of Constitutional Accountability' in Swati Jhaveri, Tarunabh Khaitan & Dinesha Samararatne (eds), Constitutional Resilience Beyond Courts: Views from South Asia (Bloomsbury Publishing 2022) (forthcoming). 18 See the next section for elaboration of this distinction. 19 See Luigi Ferrajoli, Principia Iuris: Teoría del Derecho y de la Democracia, vol 1 (Trotta 2011) 637-8. Ferrajoli characterizes these duties as primary/substantive and secondary/instrumental 'guarantees'.
The primary duty to respect or nourish the content of the norm applies to any norm-making institution. This is mainly the legislature, although in many states a significant portion of norm-making is also undertaken by appellate courts and the higher executive. Constitutions typically entrench their norms to some extent, by vesting the capacity to change them in specific actors alone, who must follow specified procedures to express a change. All ordinary norm-makers, by implication, are under a primary duty to respect the content of the norm by refraining from weakening or erasing it. Some institutions, often courts, have the additional primary duty to nourish the norm by updating and strengthening it, keeping it relevant under changed circumstances or in the face of unanticipated problems. 20 The effective discharge of secondary duties is essential to secure the primary duties. When breaches of primary duties go un-investigated, uncriticised, or unremedied, the norm itself could change. Secondary duties with respect to the content of the norm include a duty to vigilantly look out for and publicise any suspected breach, and to determine whether there has, in fact, been such breach. If this is the case, there may be additional duties to criticise and remedy such breach.
The duty to publicise a suspected breach can only be performed if the duty bearer has access to the necessary information, regularly examines such information, and can publicly highlight any suspected cases. Determining whether there actually has been a breach of the primary duty to respect the content of the norm is typically an interpretive task: does act x of institution S weaken or erase the norm. In such cases, empirical questions ('did institution S actually do x') tend not to be at issue, at least in sufficiently rule-of-law-compliant regimes. 21 In addition, secondary duties include the duty to criticise and remedy any breach, the latter usually achieved by ordering the purported normchange to be ineffective. Sometimes, the remedy can take the form of an invitation, recommendation, or injunction to another actor (often, the original breacher of the primary duty) to undertake the restoration of the original norm. 22 Criticism of breaches is a key tool for norm-maintenance, but often overlooked. Public criticism expresses a reaffirmation of the norm, and by doing so strengthens it: its salience in overly court-centric constitutional scholarship has been underestimated.
The primary as well as the secondary duties with regard to protecting the content of the norm only require that the duty bearers have expressive capacity. These duties can be performed by any Even when such an empirical fact is at issue, it is usually resolvable on the basis of documentary evidence and deemed legal facts, (such as 'A bill certified by the Speaker of the House to have been duly passed by a majority of its members shall be deemed to have been so passed'). It is, therefore, very rare that investigating a primary expressive breach requires fact-finding by an executive officer. 22 A declaration of incompatibility under section 4 of the UK Human Rights Act 1998 can be understood as a recommendation to Parliament to amend the impugned statute to remove the incompatibility. actor or institution that has the ability to speak and communicate. Most state institutions have some formal ability to communicate in ways that can effect normative change: legislatures do so through statutes and resolutions, courts do so through judgments, ministers do so through orders, government departments do so by promulgating rules. Only formal expression can typically nourish a norm, authoritatively determine breach, or remedy such breach. On the other hand, the primary duty of respecting the content of the norm and the secondary duties of publicising and criticising breaches can be performed by any state institution with formal or informal expressive capacity. In constitutional practice, the primary duty with regard to respecting the content of the norm is imposed on all state institutions except those explicitly vested with a constitutional amendment power; the secondary duty to publicise suspected breaches rests normally with the political opposition in the legislature along with the relevant guarantor instution (if any) and the media, and other secondary duties tend to be vested in constitutional courts. 23 In Table 1, duties that can be performed simply by institutional expressiona stroke of the pen or a key or the utterance of certain wordsare highlighted in yellow.
Guaranteeing the impact of a norm usually requires that it is made effective in the 'real' world, normally through the exercise of material, rather than merely expressive, capacity. They usually require some physical act to be performed: ballots need to be counted, roads built, streets patrolled, taxes calculated, risks assessed, and so on. Occasionally, though, a speech-act may suffice to satisfy a primary material duty: for example, a marriage registrar can discharge her duty to marry a couple who satisfy the necessary legal requirements by simply pronouncing them to be married in the legally recognised form, after following the required process. 24 Norms seek to govern human behaviour and are trivalent in relation to actions: they can mandate certain acts, permit them, or prohibit them. In order to materially guarantee a norm that prohibits or permits some action, primary duties may include a duty to respect the prohibition or the limits of the permission, and (perhaps) a duty to prevent others from doing what is prohibited or from doing what is permitted in the wrong manner or to the wrong extent. Much traditional constitutional scholarship is devoted to negative restraints on the state, in that it focusses largely on prohibitive norms. 25 Preventive and mandatory dimensions of norms have not received the attention they deserve. 26 For mandatory norms, the primary material duty is to fulfil its mandate. 27 Mandatory norms are often aspirational, and concern the realisation of telic objectives such as substantial material equality or providing everyone with a living wage. 28 When a constitution guarantees such aspirational (or, transformative) norms, primary impact-related duties require that relevant constitutional actors shall pursue the realisation of the stated objectives. It is, therefore, meaningful to seek to guarantee an aspirational non-self-enforcing norm even if, or rather especially if, the stated objective is far from being realised. 29 Finally, guaranteeing the impact of a norm may 23 I include both appellate courts with the power of 'diffuse' constitutional review that are usual in common law systems as well as separate and centralised 'constitutional' courts typically found in civilian systems within the term 'constitutional courts'. See  also require the imposition of a duty to educate all stakeholders about the norm. Since education is often a communicative act, one may think that expressive capacity alone could suffice for the discharge of this duty. This, however, is unlikely to be the case. Compliance with non-self-enforcing norms does not come naturally to most actors. They are fragile by their very nature. Educating stakeholders may frequently require advertising campaigns, development and dissemination of informational materials, conducting training sessions, and so on. These acts require material capacity, a mere expressive proclamation in the official gazette or a public speech is unlikely to suffice.
Thus, unlike the primary duties with regard to the content of the norm, guaranteeing its impact typically (although, not necessarily) imposes duties that require physical acts or restraints. While expressive speech-acts can be performed by most state institutions, not all state institutions are capable of physical acts. Normally, it is only some office of the executive that is capable of performing physical acts. Legislatures or courts do often have officers capable of performing limited physical acts, but they ordinarily tend to rely on the executive for the performance of their directives. The implication is significant: institutions that wholly or substantially lack executive capacity will be unable to primarily guarantee the impact of norms. If guaranteeing all constitutional norms required expressive capacity alone, we would not have needed non-judicial guarantor institutions at all. Their need arises because of the need for (small-e) executive capacity that is located outside the (big-E) Executive branch.
Secondary duties in relation to the impact of the norm include: (i) publicising a suspected breach of a primary material duty, (ii) investigating any suspected breach, (iii) determining whether the alleged or suspected breach is in fact a breach, (iv) if a breach is found, criticising it, (v) determining appropriate remedies-if any-to redress it, and (vi) effectuating such remedy. While duties (i), (iii), (iv) and (v) can be discharged adequately by an institution with expressive capacity alone, the performance of duties (ii) and (vi) would normally require an institution that has the ability to perform physical acts. Unlike secondary investigative duties concerning content, investigations into suspected breaches of primary impact-related duties typically require empirical fact-finding. Again, unlike breaches of primary content-related duties which can usually be remedied by undoing the change, breaches of primary impact-related duties often need physical actions to be performed by some actor.
The distinction between the content and impact of a norm is not watertight. It is possible for the content of a norm to change because of its (lack of) impact. Imagine a norm whose contentalthough never formally changedis consistently breached. If these breaches do not invite a response by triggering actions demanded by secondary duties, they couldover timechange the content of the norm itself. The point is most readily understood in relation to constitutional conventions whose normative content is not authoritatively codified. 30 Take, for example, an uncodified constitutional norm that a minister must resign if there is a major corruption scandal involving her department. It imposes a primary content-related duty (on all ministers, among others) to not weaken or erase the norm. It also imposes a primary impact-related duty on all ministers to actually tender their resignations if there is a major corruption scandal involving their departments. 31 Suppose, now, that in breach of the norm, a minister refuses to resign when such a scandal does erupt in her department, claiming that her resignation is not warranted because she was not herself implicated in the scandal. If her refusal to resign is broadly tolerated by the This, by the way, is an example of a primary material duty that can be performed through a speech-act (of resignation).

S48
Tarunabh Khaitan polity, i.e. it fails to sufficiently trigger actions based on secondary impact-related duties, the norm might begin to change. Unlike expressive change which happens as a momentous event, norm transformation by changes in practice is a process. This renders the precise moment when the content of the norm changes because of its (lack of) impact hard to pin down. But at some point, a sufficient number of uncriticised 'breaches' could indicate that the norm itself has changed to something along these lines: a minister must resign if there is a major corruption scandal involving her department in which she is personally implicated. 32 Triangulating the cross-cutting distinctions between (i) the content and the impact of a constitutional norm, (ii) the primary and the secondary duties that guaranteeing a norm requires the imposition of, and (iii) the expressive and material capacities required to perform these duties reveals several insights. For our purposes, the most important of these insights is that what is referred to as the 'checking function' in separation of powers literature is essentially the performance of secondary duties, paradigmatically (but not exclusively) performed by constitutional and administrative courts. Much of the existing literature on the incipient 'fourth branch' understands the functions of the institutions in this 'branch' as some form of checking function, akin to the judiciary. Ackerman, for example, identified the emergence of what he characterised as the 'integrity branch' in certain jurisdictions, whose main task was to check political corruption. 33 Fombad recognized the emergence of 'a possible fourth branch' in African constitutions inspired by South Africa, aspiring to 'make up for the increasing inability of the traditional checks and balances … to ensure that African governments operate in an open, transparent, accountable and responsive manner'. 34 In the same vein, Brown identified an 'integrity branch' in Australia, defining it thus: 'The integrity branch of government consists of those permanent institutions, established with a degree of political independence under a Constitution or by statute, whose function is solely or primarily to ensure that other governmental institutions and officials exercise the powers conferred on them for the purposes for which they were conferred, and in the manner expected of them, consistent with … the legal and wider precepts of integrity and accountability…' 35 Integrity institutions only perform secondary duties, those that only kick in when primary duties have already been breached (although, occasionally, they may also be mandated to nourish the norm). They include constitutional and administrative courts as well as anti-corruption watchdogs, ombudsoffices, electoral commissions (when performing a judicial or quasi-judicial function) and so on. Checkers are primarily reactive; their preventive potential is largely indirect inasmuch as the possibility of their intervention deters breach. But the scope of guaranteeing a norm extends beyond ensuring the integrity of other actors. Unlike checkers who enforce secondary duties, guarantors may have to shoulder the primary duties (especially at preventing a breach or fulfilling the demands of a norm) directly by doing whatever is required by the norm itself. A guarantee that ø may require the guarantor itself to ø. For example, knowledge institutions tasked with producing and disseminating certain types of knowledge relevant to public policy and constitutional governance tend not to perform checking roles. 36 Perhaps appreciating this distinction, Klug characterises only three institutions mentioned in chapter 9 of the South African Constitutionthe electoral commission, the Public Protector, and the Auditor Generalas 'integrity' institutions. 37 Other institutions in the chapter, such as the broadcasting authority and the Commission for Linguistic, Cultural and Religious communities surely are guarantors, even though they may not qualify as integrity institutions.
Thus, guarantor institutions are distinct from integrity institutions, although the same institution can sometimes be a guarantor as well as an integrity institution, as is the case with anti-corruption watchdogs. Furthermore, unlike courts (including many constitutional courts), guarantors typically require considerable material as well as expressive capacities. Without material capacity, depending on the norm in need of a guarantee, they would often end up being ineffective in the discharge of their constitutional function of supplying a guarantee. But they may need expressive capacity too: sometimes to clarify or strengthen the norm in discharge of their primary content-related duty, or to determine, criticise, and remedy breaches of the primary duty by other actors. It is this combination of expressive as well as material (typically executive) capacities that is the hallmark of guarantor institutions.

Guarantors versus regulators
Guarantor institutions perform executive, and (often) legislative and judicial functions. The hybrid character of their functions, however, is not sufficient for their classification outside the three traditional branches. Contemporary states have a host of regulators who are recognised as being located well within the executive branch despite combining quasi-judicial, and occasionally legislative, powers alongside administrative ones, and enjoying varying degrees of autonomy from the partisan higher executive. Unlike the generalist bureaucracy, these regulators also tend to have specialist expertise in the matter under their regulatory concern. These more or less independent regulatory agencies can be classified based on the subject matter they tend to regulate: 38 public utilities (water, energy, transport, telecoms) welfare provisions (health, housing, justice, education, social services) professional standards (law, medicine, veterinary medicine, engineering, accountancy, built environment) the economy (banking, finance, insurance, competition) business standards (food, products, health and safety, environment, labour) criminal justice (policing, investigation, prosecution) 39 Functionally, these regulators may also perform a combination of primary and secondary duties. What really distinguishes a guarantor institution from other regulators is, therefore, not functional modalities but the status of the underlying norms they seek to effectuate as well their own status as constitutional institutions. We have already learnt that in order to provide guarantees that are enduring and credible, guarantors need to be independent. This requirement of independence in turn requires guarantor institutions to be constitutionally entrenched. Thus, guarantors are doubly 36 See Jackson (n 4) 166. Several jurisdictions, however, treat police and/or prosecution services as part of their judicial rather than the executive branch.

S50
Tarunabh Khaitan constitutionalised: they are themselves entrenched as constitutional institutions and they effectuate norms that have a constitutional character. Ordinary regulators are either (i) not constitutionalised at all (professional standards regulators often fall in this category), or (ii) themselves not entrenched as institutions even though the norm they protect is constitutional (welfare regulators in jurisdictions that recognise social rights but do not entrench their institutional operators fall in this category; so do many criminal justice regulators), or (iii) themselves entrenched constitutionally but the underlying norm they protect can be changed easily by the government of the day (this category is rare because the constitutional character of the institution often implies the constitutional character of the norm it is designed to protect; possible examples may include public service appointment commissions that ensure that the extant criteria for appointments to public offices are strictly and impartially applied, even as governments of the day may fully determine what these criteria might be). As such, ordinary regulators are unable to provide credible and enduring guarantees.
Only if the norm as well as the institutional regulator are constitutionalised, we are dealing with a guarantor institution. My account of guarantor institutions, therefore, presupposes a prior constitutional commitment to the relevant norm that is being guaranteed. Being constitutional, the norm is likely to be fundamental to the polityrather than solely to the government of the dayin some respect. Needless to say, the same norm (say, environmental protection or welfare provision) may be constitutional in one polity, and a matter of ordinary policy in another. A polity may well decide to treat any normsay provision for housingas constitutional. Constitutions that set up independent knowledge institutions such as statistics bureaus and census bureaus may be seen as (usually impliedly) seeking to make an epistemic commitment to truth in the public discourse. 40 Equality commissions of various descriptions seek to guarantee constitutional commitments to equality. Central banks may be a guarantor if constitutions wish to guarantee that certain aspects of monetary policy shall be preserved irrespective of the ideology of the ruling dispensation. The council of guardians guarantees the commitment to an Islamic state in the Iranian constitution. 41 All constitutions commit to respect, protect, or fulfil some norms; they are, to some extent at least, instruments to organise and plan for a country's future. 42 The need to guarantee constitutional commitments is important in any constitutional context, irrespective of the content of the constitutional norms. In the liberal democratic tradition, constitutional commitments tend to include the norms of legality, democracy, liberty, and probity. However, there is no conceptual reason why other norms, such as theocracy, socialism, or laïcité, cannot also be constitutionalised. Unlike Tushnet, therefore, I do not therefore restrict my understanding of guarantor institutions to 'institutions protecting constitutional democracy'. 43 A caveat is apposite: to count as constitutional, it is not essential for a norm to be found, explicitly or implicitly, in a big-C constitutional code. Small-c norms of the material constitution are also 'constitutional', 44 although there may be greater controversy over the constitutional character of small-c norms. 45 The key test is whether the polity treats the norm in question as legally or politically entrenched to a sufficient extent, such that something more demanding than the ordinary 40 Jackson (n 4). Tushnet, 'Institutions Protecting Constitutional Democracy' (n 2). For Tushnet, this seems to be a pragmatic, rather than a conceptual, restriction. 44 On the distinction between the big-C and the small-c constitution, see Anthony King, The British Constitution (Oxford University Press 2009) 5-7. 45 It is difficult to say with certainty whether public provision for universal healthcare that is free at the point of use is a small-c constitutional norm in the UK, for example. political or legal process is required to weaken or erase it. 46 We should also note that political practice can be fickle. Even in long established democracies, constitutional shamelessness of a single-term political executive can do significant damage to its softer, small-c, norms. It will take a polity very confident of the endurance of its conventional restraints to entrust the protection of its fundamental norms to political culture alone. Having said that, legal scholars are prone to forgetting that even law enforcement in any jurisdiction is, ultimately, as good as its political culture permits it to be.
Apart from enjoying relative entrenchment, another characteristic feature of constitutional norms is that they tend to admit to a considerable degree of indeterminacy, and often require constant nourishment by further specification and interpretation. This should be unsurprising: since constitutions are intended to endure, a greater level of indeterminacy and abstraction is often wise, if only to avoid unintended consequences. Guarantor institutions, at least those created by big-C constitutions, tend to have their purposes settled in broad and abstract terms (such as 'administration of free and fair elections' rather than 'enforcement of the provisions of the Elections Act'). 47 This provides flexibility to their operation and room to evolve over time as well as the ability to respond to new circumstances with new tools without necessarily needing prior authorisation of the other branches. This is a feature that they tend to share with the top constitutional institutions in other branches. Although ordinary regulatory agencies may also have their powers defined in relatively broad terms by their parent statutes, this evidences a policy choice of the enacting legislature and can be narrowed by a future legislature following the ordinary legal process.
Abstract or indeterminate constitutional norms have their content and meaning specified over time by enforcing statutes or judicial interpretation. 48 Guarantor institutions also undertake the expressive task of norm-specification as well as norm-interpretation, either on their own, or in concert with other branches. When performing their executive or judicial functions, these institutions must, out of necessity, engage in interpretation. For example, the South African Public Protector had to decide whether her inability to conclude an investigation into alleged breaches by the President within a statutorily prescribed 30-day period invalidated her investigations. She interpreted the statutory mandate, in light of relevant constitutional norms, as designed "to ensure that an investigation against a member of the executive regarding a violation of the Executive Ethics Code should be expeditious, if possible. An executive member's public accountability, in my view, cannot lapse if the 30 days period is not met." 49 Thus, she interpreted the statutory time limit as a guideline rather than a rule, relying on the broad constitutional provision. Background constitutional norms facilitate this type of purposive interpretation of more finegrained statutory rules governing guarantor institutions in ways that may not normally be available to ordinary regulators and executive agencies. If there is insufficient legislative supplementation of the relevant constitutional norm, guarantor institutions can also exercise considerable legislative powers of norm-specification. The Model Code of Conduct 'enacted' by the Indian Election Commission is 'enforced' by the Commission against political parties, even though it technically 46 In the UK, as an example of political entrenchment, it has been the political practice since the second half of the 20th century that "first-class constitutional bills" are considered by a committee of the whole House of Commons rather than sent to a standing committee at the committee stage: Select Committee on the Modernisation of the House of Commons, Scrutiny of the Draft Legislative Programme (first report) (HC 2007-08, 81) paras 74-76. 47 See, eg, Constitution of South Africa, s 190(1); Constitution of India 1949, art 324; Constitution of Nepal, s 246. 48 lacks the force of law. 50 Again, by contrast, ordinary executive bodies are normally able to make such specifying rules only if explicitly granted the power to make delegated legislation, and such power is usually strictly policed by the doctrine of ultra vires.
That guarantor institutions effectuate constitutionalrather than merely statutory or policynorms and have constitutional protection themselves distinguishes them from ordinary regulatory agencies. Ordinary legal and political norms and sub-constitutional institutions may be changed, at least in most parliamentary democracies, by the government in power acting mostly on its own steam. To change a constitutional norm or institution, on the other hand, some buy-in from the political opposition, i.e. the government-in-waiting, is typically required. This entrenchment of constitutional normswhether by legal or political meanstherefore imposes the primary content-related duty on all ordinary constitutional actors to respect the content of the norm by refraining from weakening or erasing it. One key implication of the discussion in this section is that constitutionalising a norm without also constitutionalising its enforcer is often unwise, for it allows hostile actors, who cannot directly erase or weaken the constitutional norm, to indirectly frustrate its operation by undermining or abolishing its enforcer. 51 Non-self-enforcing norms A norm requires an institutional guarantee only if it is not a self-enforcing norm, i.e. it is a norm that existing powerful actors in the polity are likely to have reasons and capacity to frustrate. Assuming that 'power' is an actor's ability to enforce its will, a self-enforcing constitutional norm is one with respect to which powerful actors are likely to have the will as well as the ability to effectuate it. Maintaining peace and order, for example, is typically a self-enforcing norm in most states. Given that the state usually seeks to monopolise the use of violence, sees all unauthorised violence as a threat to itself, and commands significant police and military force, it has both the will and the ability to maintain peace and order under ordinary circumstances. Difficult cases here tend to be those where a state's ability (rather than its will) to maintain order is compromised. Likewise, a constitutional norm that the (formally expressed) will of a simple majority of members in the legislative assembly will constitute an act of the assembly does not typically require any constitutional guarantee to be effectuated in most contexts. A simple majority of assembly members are ordinarily likely to have sufficient political power to ensure that their will counts. On the other hand, special mechanisms may be necessary to guarantee norms that protect opposition rights in an assembly, because they are less likely to be self-enforcing. Essentially, the point is this: power is the ultimate self-enforcing phenomenon. When norms track power, they are also self-enforcing and will likely be taken care of. When norms seek to constrain power, on the other hand, some combination of institutional guarantees and a culture of restraint is likely to be required.
The need for guarantor institutions arises because constitutional norms often require powerful agents to act against, or refrain from acting towards securing, their self-interest (at least in its narrow, short-term, rational sense). Legal-political norms that seek to channel and constrain power significantly are unlikely to be self-enforcing; power, after all, is self-perpetuating, convertible, and resilient. 52 Note that we are not speaking of self-enforcing constitutions; rather our interest here is in appreciating specific non-self-enforcing norms-the former is a package of norms, the latter its constituent elements. A package of norms, such as an entire constitution, can be made selfenforcing ifon the wholeit is rational for powerful actors to accept the package and largely play by its rules, rather than reject it in toto. They may have reasons to accept a constitution even as they remain unhappy with aspects contained therein. However, even if the constitution as a whole is self-enforcing in a particular polity, actors may nonetheless be motivated to frustrate those of its norms that are against their selfish interests, so long as they can get away with it.
A specific non-self-enforcing norm may well be adequately guaranteed through checks and balances between and within the three traditional generalist branches of the state. Whatever the main purpose of the tripartite separation of state power, as Waldron argues, the separation of state power into three branches itself credibly and enduringly guarantees a constitutional commitment to legality. 53 Similarly, constitutional courts and/or upper chambers of federal legislatures usually suffice to guarantee a constitutional commitment to federalism in most contexts. 54 However, the three brancheswhether taken severally or jointlycannot guarantee all constitutional norms between them. Take democracy. To guarantee (aspects of) democracy, some of the apex institutions in the traditional branchesusually the political executive and the legislatureare made responsible to the people. The reason why popular responsibility is not demanded of all institutions is simple: democracy is not the only norm democratic constitutions commit to. Constitution makers generally recognise that an elected judicial branch would compromise the demands of legality too much to be advisable. Therefore, most of the democratic world (at least outside the United States) chooses not to elect its judges through popular votes. If we want both legality and democracy, some such optimisation becomes necessary.
A democrat might accept this necessary compromise, but does she have reasons to believe that even this limited democracy can be guaranteed over time simply by this tripartite separation of powers? Although putting in place a well-designed electoral system for electing members of the legislature is a good start, 55 that alone is not sufficient to guarantee the democratic norm. A constitution cannot plan for every eventuality-new norms will be needed to augment old ones. And all norms, new and old, will have to be enforced. These two dimensions track the content and the impact of the democratic norm, and the respective duties they impose. The need to respect and nourish the content of these norms and to effectuate them raises a credibility problem-if the system entrusts the control of the legislature and of the higher executive to the ruling party/coalition of the day (as parliamentary democracies tend to do), then the ruling party could use its temporary power to entrench itself in the institutions of the state and make its power permanent. In the legislature, it will be tempted to bend electoral rules in its own favour; as the executive, it will want to administer norms in a manner that benefits the ruling party. Presidential systems, far from solving the problem, accentuate it-when the same party controls the executive and the legislature, it can enact partisan norms not only with the same ease as in parliamentary systems, but these partisan norms remain far more resilient because it is likely to take another united government by the opposition party to undo the partisanship of the electoral norms. In practice, if yesterday's opposition comes to control both elected branches, it has no incentive to stop at de-partisanising electoral norms; most likely it will also seek to skew the electoral field in its own favour. Democratic states are, therefore, internally unstable, for the government of the day can perpetuate its power only by reneging on its commitment to democracy. 56 If democracy is a system where parties lose elections, 57 such a system could soon become undemocratic. Tushnet identifies the problem as one where the traditional 'branches are placed in a situation of conflict of interest'. Within the traditional branches of the state, the chief constitutional mechanisms to protect democratic regimes from ruling parties have been the political opposition in the legislature and constitutional courts in the judiciary. Thus, democratic constitutions typically task the opposition with checking the government politically and presenting itself as the government-in-waiting, and tend to provide it with a protected voice and a modicum of influence to enable it to do so. 59 Meanwhile, Kelsen's answer to the credibility problem was the generalist institution of a constitutional court with the power to review legislative and executive acts for compliance with the democratic constitution. 60 However, as expressive institutions, constitutional courts typically lack the capacity to administer elections, and the political opposition in the legislature has reasons to game the system in its own favour. Kelsen himself conceded that 'No one will claim that [the constitutional court] is an absolutely effective guarantee under all conceivable circumstances.' 61 For good reason then, even jurisdictions such as South Africa that created a constitutional court have nonetheless also created additional 'institutions supporting constitutional democracy' alongside it. 62 Administering elections is, after all, a primary impact-related duty arising out of the democratic norm, whose enforcement demands material capacities. This is where we run out of options within the tripartite framework: vesting the opposition with electoral administration will simply replicate the problem of partisan enforcement of electoral rules, this time to benefit the opposition. Allocating electoral management jointly between the current government and the opposition can still lead to a cartelised exclusion of other partisan players. 63 Bangladesh has even experimented, for a brief period, with vesting interim electoral administration with the judiciary, in a system where the Chief Justice of the country's top court formed an interim government to oversee elections. 64 Vesting such vast material capacity in the head of the judiciary is perhaps a compromise too far with the functional separation of powers between the traditional three branches.
The upshot of this discussion is that try as we may, it is difficultif not impossibleto guarantee the constitutional norm of electoral democracy by merely tweaking the architecture of the three traditional branches of the state. Unless a polity wishes to risk counting solely on its cultural respect for democracy the management of free and fair democratic elections requires an electoral commission with material capacity, and one that is sufficiently independent of political parties. There may be good reasons to vest such a commission not only with the executive power to conduct elections so it can enforce its primary impact-related duties, but also the power to supplement electoral rules to enable it to uphold its primary content-related duty to nourish the norm. There may even be grounds to allow a judicial division of the electoral commission to adjudicate upon electoral disputes towards the performance of some of the secondary duties concerning the norm. Whether sufficiently guarantee all constitutional commitments. Sometimes, they may have the will, but lack time or capacity. Sometimes, their design is ill-suited for the task at hand. It may be that they lack the required expertise to perform the required function. These are all good reasons, besides a conflict of interest, to entrust the task of guaranteeing a constitutional norm (or an aspect thereof) to a custom-built and dedicated fourth branch institution. 59 rights lawyers, but also personnel with other set of skills and expertise, if they are to effectively discharge their (specific but broad) guarantor role.
As specialist institutions, the internal design of any given guarantor institution predictably varies in relation to other guarantor institutions in the same polity, for the relative importance and the credibility-demands of constitutional norms vary. Some guarantor bodies are federated, others are not. 68 Some are a single-member institution 69 , others are multi-membered. 70 Some oversee a vast country-wide administrative apparatus, others fit into a small office in a government building. 71 They can look like a court (anti-corruption watchdogs which often possess powers akin to those of a civil court), 72 or an executive body (broadcasting authorities or statistic authorities), 73 or some combination of the two (such as information commissioners 74 or electoral commissions with judicial powers 75 ). Such variation exists not only between different types of guarantor institutions but can also exist for the same guarantor institution across different jurisdictions. 76 Generalist institutions, on the other hand, follow relatively fixed templates, significant variations notwithstanding: large legislative assemblies decide by majority votes; smallish judicial benches decide by reasoned judgments after hearing interested parties; 77 small, highly coordinated groups of political executives lead a vast state administration. These are common features of the apex institutions in the three generalist branches almost everywhere.
The specificity and design-variability of guarantor institutions is connected with their actual and perceived legitimacy in a polity. An institution whose point is to guarantee probity in high constitutional offices will be legitimate if and only if (i) probity in high constitutional offices is itself a legitimate constitutional norm, and (ii) its design makes it sufficiently effective in credibly guaranteeing such probity. A single, non-specific, guarantor institution (such as a constitutional court) is unlikely to satisfy the tailor-made-design requirement of legitimacy with respect to all the norms it is meant to guarantee; it is also unlikely to be able to perform both primary and secondary sets of duties that require material, rather than expressive, capacity. There is thus a trade-off between functional generality and purposive specificity of an institution-the more general its functions, the less effective it is likely to be in serving particular functions. If constitutional courts are jacks of all trades, specific guarantor institutions are masters of one. To criticise an institution for its failure to pursue or satisfy a norm other than the one it is designed to guarantee is to fail to appreciate that specificity is a 68 characteristic feature of guarantor institutions. Contemporary constitutions, quite rightly, care about many things-democracy is but one of them.
An important consequence of their tailored specificity might be a distinctive manner of reasoning by guarantor institutions. The political executive and the legislature tend to reason like most ordinary folk-perhaps excepting individual self-interest, they are permitted to appeal openly to any reason and make an all-things-considered judgment on most issues. Political reasoning is, therefore, unbounded. Judicial reasoning is formal and stylised in a particular way: they must show that their decision is in accordance with law. 78 As such, judicial reasoning is bounded perspectivally-judges must decide from the point of view of the law. But within this framework, they ought to nonetheless consider all relevant reasons before arriving at their judgment. As generalist institutions, this all-things-considered mode of reasoning of the apex institutions in the three traditional branches should not be surprising. The nature of their reasoning is reflected in the eligibility criteria for each apex institution: members of the political executive and legislators do not usually need to satisfy any subject-matter expertise requirements, but are drawn from or required to interact with people from all walks of life to examine issues from multiple perspectives. Apex court judges need technical expertise in legal reasoning and knowledge of law, but need not have any substantive expertise in particular subject areas.
Decisions of guarantor institutions are likely to be bounded substantively, because their raison d'être is to protect a specific constitutional norm. This is not to say that electoral commissions will never seek to balance the demands of democracy against other values, or that an anti-corruption watchdog will always pursue probity, whatever the cost. The point is that as institutions dedicated to protecting a single, specific norm, effective guarantor institutions are likely to give its protection enormous, and frequently decisive, weight in cases where this value clashes with other concerns and interests. Compromises they make are likely to be strategic or pragmatic in naturesuch as avoiding a course of action if the threat of backlash from another institution is high, or if the administrative costs of adopting the best means are prohibitive. Principled balancing of the norm being guaranteed with other norms are likely to be rare-for example, an anti-corruption watchdog is not likely to go easy on high-end corruption even if, as Tushnet argues, such corruption can sometimes be good for governance. 79 Tushnet criticises this feature as the 'mission commitment' of the leaders of these institutions, which encourages them to 'believe that their experience dominates all other considerations.' 80 For our conceptual purposes, we will refrain from commenting on the desirability of such mission-committed specialists, but only note that their specialism is a defining feature of guarantor institutions, with real consequences, both good and bad.

Conclusion
I have argued that a guarantor institution is a tailor-made constitutional institution, vested with material as well as expressive capacities, whose function is to provide a credible and enduring guarantee to a specific non-self-enforcing constitutional norm (or any aspect thereof). This explanation adds conceptual clarity to an emerging set of important constitutional institutions, especially in the Global South. Much further investigation is still necessary. One important remaining conceptual issue is this: do guarantor institutions, taken together, constitute a fourthguarantorbranch? Other concerns include how, given their individual, specific, mandates, do guarantor institutions relate to other guarantor institutions? What sort of legitimation concerns arise in relation to these unelected, political, but post-partisan institutions? Empirically, what has been their track 78 Waldron (n 26) 224-5.

S58
Tarunabh Khaitan record in delivering on their specific promise? Are they normatively desirable, and if so, when? These, and no doubt many other, unanswered questions concerning guarantor institutions suggest that constitutional scholars have their work cut out.