The Remaking of South African Administrative Law

Abstract This article explores the remaking of administrative law review in South Africa since the introduction of constitutional democracy in 1994. It characterizes the construction of the constitutional and legislative framework, as well as the courts’ interpretation of that framework, as the first phase of the remaking. The second phase encompasses the courts’ recognition of a constitutional principle of legality based on the rule of law, and their swift development of the content of this principle. This judicial creativity has resulted in an elaborate avenue to review, parallel to the Promotion of Administrative Justice Act 3 of 2000, and has caused problems of rivalry and avoidance. The article identifies and discusses some of the more significant implications of each of these phases of reconstruction. It also proposes corrective measures likely to advance the coherence and effectiveness of judicial review and discourage the adoption of a doctrine of non-justiciability.


Introduction
In South Africa, judicial review has been the dominant administrative law remedy since the days of British colonial rule.Under apartheid, however, the effectiveness of common law grounds of review against breaches of administrative justice was greatly reduced by the sovereignty of a whites-only Parliament, coupled with a largely executive-minded judiciary. 1he legal terrain, and the very basis of administrative law review, shifted dramatically with the constitutional revolution of 1994.As the Constitutional Court described it, the grundnorm of administrative law now resided in the democratic constitution and the courts "no longer ha [d] to claim space and push boundaries to find means of controlling public power". 2The Bill of Rights, in both its interim and final form, famously included rights to administrative justice. 3Section 33 of the 1996 Constitution (the Constitution) not only entrenched rights to just administrative action but also mandated the enactment of national legislation to give effect to those rights and to provide for judicial review of administrative action.In due course, that legislation was enacted as the Promotion of Administrative Justice Act 3 of 2000 (PAJA).These twin developments, together with the courts' interpretation of PAJA, may be regarded as the first phase of the remaking of administrative law, a process in which the constitutional drafters, the legislature and the courts all played important roles.
However, section 33 is not the sole constitutional foundation for administrative law review.The other is the rule of law, a founding value of the Republic of South Africa, and the principle of legality that forms part of this value.Together they give rise to "legality review", which applies to the exercise of all public power, and which the courts have enthusiastically developed to include most of the grounds of traditional administrative law review.The result of this further remaking of administrative law by the courts is nothing less than the creation of a new, constitutional common law. 4 This article explores what seem to be some of the main implications of this post-democratic remaking of administrative law.It begins by outlining the reconstruction of administrative law pursuant to South Africa's constitutional revolution.That description includes a brief account of the constitutional and statutory framework (the first phase) as well as how and why the legality principle was recognized and then elaborated into a highly significant avenue to review alongside PAJA (the second phase).It explains problematic results of the second phase: rivalry and avoidance.Then, with an eye to the effectiveness and coherence of administrative law review, it identifies and discusses some of the most important implications of each of the two phases and proposes a few corrective measures.A brief conclusion follows.

The remaking of administrative law
The first phase: Section 33 of the Constitution and PAJA In 1994, judicial review (a common law institution) was transformed by the adoption of constitutional rights to just administrative action.Those interim rights were replaced a few years later by section 33 of the Constitution and its mandate for the enactment of national legislation. 5hile section 33 has apparently inspired the enactment of even more expansive rights to administrative justice in other countries, 6 the rights it confers are generous by any measure.Section 33(1) proclaims, in unqualified form, 7 that "[e]veryone" has the right to "lawful, reasonable and procedurally fair" administrative action.Indeed, the point of mandating the enactment of national legislation was only partly to flesh out the rights in section 33(1) and (2).Another aim was to qualify their broad sweep to avoid imposing "paralysing burdens on effective administration in South Africa". 8The legislature's fear of causing administrative paralysis became apparent when it enacted PAJA, for it made several changes to the Administrative Justice Bill proposed by the Law Commission, 9 almost all of them aimed at reducing the burden on the government.
For example, the legislature spurned the ground of review of unreasonableness proposed by the Law Commission, which explicitly included disproportionality, and replaced it with a far more 4 Hoexter "From pale reflection", above at note 1 at 182. 5 Sec 33 provides: "(1) Everyone has the right to administrative action that is lawful, reasonable and procedurally fair.(2) Everyone whose rights have been adversely affected by administrative action has the right to be given written reasons.
(3) National legislation must be enacted to give effect to these rights and must -(a) provide for the review of administrative action by a court or, where appropriate, an independent and impartial tribunal; (b) impose a duty on the state to give effect to the rights in subsections (1) and (2); and (c) promote an efficient administration."6 Including Kenya, Zimbabwe and Fiji; see further, Hoexter "From pale reflection", above at note 1 at 180-81.7 Like all rights in the Bill of Rights, however, sec 33 is subject to reasonable and justifiable limitation as contemplated in the Constitution, sec 36(1).8 Report on Administrative Justice (August 1999, South African Law Commission), para 3.3.9 The Law Commission's bill appears as annexure A to its Report: id at 15-43.restrictive Wednesbury clone. 10Innovations in connection with rulemaking and various other reforms, such as the creation of an administrative review council, were similarly rejected or attenuated.(Section 10(2) of PAJA does still allow for ministerial regulations to be made on such topics, but this power has not been used.)A provision that became section 2 of PAJA was added to allow both for ministerial exemptions from the statute and "permissions to vary" duties as to procedural fairness and reason giving.Most significantly for present purposes, the legislature did its best to limit PAJA's sphere of application.
This aim was achieved via the threshold concept of "administrative action", for, like its parent rights, PAJA only applies to such action.During the parliamentary process, several elements were added to the relatively simple definition proposed by the Law Commission, including a couple of ingredients borrowed rather arbitrarily from other jurisdictions. 11As a result of these additions, under section 1 of PAJA, conduct qualifies as administrative action only if it is: a decision of an administrative nature; by an organ of state or another person; exercising a public power or performing a public function; in terms of any legislation or an empowering provision; that adversely affects rights; that has direct, external legal effect; and does not fall within any of the listed exclusions.
Had it not been for the courts' intervention, this "palisade of qualifications" 12 would no doubt have achieved its aim of a vastly reduced burden of administrative justice.However, the courts did intervene, partly by means of a liberal interpretation of the statutory definition of administrative action and partly by discovering and swiftly developing a constitutional safety net to take care of cases of non-administrative action.Indeed, although the realm of non-administrative action was expanded by the PAJA definition, the question of what to do with such action predated the national legislation.The problem was the inevitable result of demarcating a constitutional right or rights to just "administrative action", and the Constitutional Court (South Africa's highest court) began to address it some months before the drafting of PAJA began.The court did so by deploying a constitutional principle of legality: a principle that would apply to every exercise of public power, thus guaranteeing some level of court-based accountability in respect of non-administrative action.The principle of legality thus began as a safety net.However, as appears from the account that follows, it soon became something far more extensive and elaborate.
The second phase: The principle of legality and the development of its content The courts' remaking of post-1994 administrative law took place from 1998 in a series of cases involving non-administrative action.It was in Fedsure, 13 concerning the making of original legislation by a municipal council, that the principle of legality was first identified and characterized as an aspect of the rule of law.The case was decided under the Interim Constitution, which, unlike the 1996 Constitution, 14 made no explicit reference to the rule of law.The Constitutional Court was undaunted, however, and reasoned that the rule of law, and thus the principle of legality (the more general counterpart of the right to lawful administrative action), was necessarily implicit in the Interim Constitution. 15In essence, the principle required public power to be exercised within the authority lawfully conferred on its holders. 16In SARFU 17 a few months later, in the context of executive action taken by the president, the court expanded this meaning to require those who exercise public power to act in good faith and not misconstrue their powers.The following year, in Pharmaceutical Manufacturers, 18 the same court laid down a basic requirement of rationality for every exercise of public power.While this entailed merely establishing a rational relationship between an exercise of power and the purpose for which the power was conferred, that link was judged to be missing in this instance.The court found that the president had acted irrationally in an objective sense when proclaiming the Medicines Act 19 into force before essential regulations and schedules were ready, for this premature proclamation could only impede the purpose for which the power was conferred. 20he safety net of the legality principle was thus in place by the time PAJA entered into force in November 2000.However, as already mentioned, it did not remain a mere safety net for long.In May 2003, in Pepcor, 21 the Supreme Court of Appeal 22 boldly added to the principle's coverage by reading into it a ground of review that had never previously been recognized in South African law: material mistake of (non-jurisdictional) fact.According to the court, the doctrine of legality expounded in Fedsure, SARFU and Pharmaceutical Manufacturers required that "the power conferred on a functionary to make decisions in the public interest … be exercised properly, ie on the basis of true facts". 23PAJA was not applicable in this case as the facts had arisen before the statute came into force.Nevertheless, the court suggested that the new ground of material mistake of fact could also be accommodated by a suitably expansive interpretation of section 6(2)(e)(iii) of PAJA, a ground dealing with relevant and irrelevant considerations. 24This was a significant moment: the first occasion, although not the last, on which the legality principle led the way in terms of the creation or expansion of grounds of review.
After a lull of several years, the next stage in the development of the legality principle was the elaboration of the basic requirement of rationality introduced in Pharmaceutical Manufacturers.In the pivotal case of Albutt 25 the Constitutional Court held that, as a matter of rationality, the president should have heard victims or their families before exercising a power to pardon offenders who had been convicted of politically motivated offences.Rather confusingly, however, both before and after this judgment, the same court denied that procedural fairness was part of the legality principle. 26This tension was resolved eventually, although not very satisfactorily, in Law Society of South Africa, 27 where the Constitutional Court reserved the label "procedural fairness" for hearings based on fairness to an affected party and insisted on "procedural rationality" to describe hearings for the sake of rationality.Unfortunately, this solution contradicted the court's previous jurisprudence in which "procedural fairness" was informed by both the traditional rationales: better and more rational decision-making, as well as fairness to those likely to be affected by a decision. 28he concept of procedural rationality had itself emerged from Simelane 29 in 2013.This case concerned a challenge to the president's controversial appointment of Menzi Simelane as National Director of Public Prosecutions (NDPP).Here the Constitutional Court distinguished between substantive rationality, relating to the decision itself, and procedural rationality, relating to the process of making the decision.The latter included steps in the process and, indeed, "everything done" in the process. 30The court held that both decision and process must bear a rational relationship to the purpose of the power, and furthermore that a misstep, if sufficiently irrational, might taint the entire process and thus the decision itself. 31In this instance, it concluded, the president's failure to take into account a relevant consideration (cogent evidence pointing to dishonesty on Simelane's part) was clearly at odds with the purpose of appointing a credible and conscientious NDPP. 32nterestingly, in this development the legality principle was again in advance of PAJA: it was only in 2020 that the Constitutional Court decided to read procedural rationality as such into the statute.In PG Group, 33 again in the context of a failure to consider relevant material, a majority reasoned that, while section 6(2)(f)(ii) of PAJA seemed to address only substantive or decisional rationality, 34 it made no sense for rationality under PAJA to have a more restrictive meaning than rationality under the legality principle.Failure to take into account a relevant consideration was thus read into section 6(2)(f)(ii), notwithstanding the separate provision made for that very failure, arguably in broader terms, in section 6(2)(e)(iii) of the statute.As the majority saw it, some evaluation of process was essential to a rationality inquiry for, without it, "we are simply asking whether a decision is right or wrong based on post hoc reasoning". 35eanwhile, in a case decided in 2013, 36 the Supreme Court of Appeal had broken another barrier by requiring reasons to be given for a non-administrative decision of the Judicial Service Commission (JSC).Once again this was for the sake of rationality, or more accurately for the sake of demonstrating rationality.The JSC had decided not to recommend certain well qualified candidates for judicial office and had refused to give reasons for this controversial decision.While conduct of this kind is expressly excluded from the definition of administrative action in PAJA, the court reasoned that the JSC would effectively be immune from a rationality challenge if, "when properly called upon to do so", it was not required to give reasons in such a case. 37his obligation is presumably another manifestation of procedural rationality.

Problematic results of the second phase: Rivalry and avoidance
The courts' deployment and rapid expansion of the principle of legality have profoundly affected administrative law review in several ways.
First and most obviously, in view of its increased content, the legality principle can no longer be described as a mere safety net for non-administrative action.While there are still some differences between the grounds of review available under the legality principle and PAJA, those differences are far less pronounced than they were two decades ago.In relation to the grounds associated with lawfulness, which is to say the areas of authority, jurisdiction and abuse of discretion, there is really nothing to choose between the two avenues to review.At the other end of the spectrum, the only PAJA ground that remains entirely unavailable under the legality principle today is unreasonableness, including disproportionality.In between these extremes, the differences are a matter of degree and sometimes also of terminology.Review for substantive irrationality is available under both avenues but is somewhat more searching in terms of the four legs of section 6(2)(f)(ii) of PAJA than under the legality principle.Furthermore, while "procedural fairness" as redefined in Law Society of South Africa38 is said not to be a ground of review under the legality principle, hearings and other procedures traditionally associated with fairness may, as we have seen, be required as a matter of procedural rationality, as may the giving of reasons.
This enhanced coverage means that there are few cases now in which it will matter much whether a review challenge is brought under PAJA or the legality principle.Furthermore, the attractions of the latter are considerable.One advantage for both litigants and the courts is not having to bother with the complicated administrative action inquiry that is one of the drawbacks of PAJA: legality review is always available provided only that the conduct under scrutiny qualifies as an exercise of public power. 39Another advantage is that, procedurally, legality review is governed by relatively liberal and flexible common law doctrines.Litigants opting for legality review thus also avoid the strict 180-day delay rule set out in section 7(1) of PAJA and the stringent duty to exhaust internal remedies in section 7(2).The advantages extend to remedies, too, for section 172(1)(b) of the Constitution is even more liberal than section 8 of PAJA.Without imposing any restrictions, the constitutional provision simply empowers a court deciding a constitutional matter to make "any order that is just and equitable". 40From the courts' point of view, the legality principle also has the attraction of flexibility.Legality, a much cleaner slate than PAJA, can more easily be made to mean whatever the court wants it to mean in a particular case.
Given these distinct advantages, it is not surprising that the legality principle has increasingly become a rival to PAJA. 41The statute has frequently been avoided or sidestepped in favour of legality review, 42 by which is meant that that avenue has been employed in respect of conduct that possibly qualifies as administrative action.At first, avoidance of the statute tended to operate in favour of section 33 or the common law.Then came Albutt, 43 in which the Constitutional Court rather recklessly encouraged avoidance of PAJA in favour of legality review.As the court admitted frankly, it would otherwise have had to engage in a particularly tricky administrative action inquiry. 44nstead of doing so, the court reasoned that the inquiry was unnecessary and thus "ancillary" because, on its unique and context-specific facts, the case was capable of being resolved by means of the legality principle. 45Indeed, the court went so far as to criticize the court a quo for having "reached" and answered the administrative action question unnecessarily, thus wasting time on an ancillary issue. 46s explained in the next section, this expedient approach was inconsistent with constitutional subsidiarity to the extent that it preferred a general constitutional principle over a more specific legislative norm, and a constitutionally mandated one at that.The Albutt philosophy might indeed have resulted in the near redundancy of PAJA, for almost every case could similarly have been decided on its own facts with reference to the legality principle. 47Fortunately, within a few years the Constitutional Court reverted (albeit in an unobtrusive footnote, and without mentioning its contrary approach in Albutt) to the more prudent position that recourse may only be had to the legality principle once it is clear that PAJA does not apply. 48That has been the orthodoxy ever since. 49However, PAJA avoidance continues to be widely practised and tolerated, and the Constitutional Court has not been consistent about condemning it. 50These mixed messages from the highest court naturally foster uncertainty, and a High Court judge has suggested that it is still not completely clear "whether recourse to the principle of legality applies only residually". 51hree other moves by the Constitutional Court may be thought to have contributed to the privileging of the legality principle and thus the undermining of PAJA.The first is the court's unpopular conclusion in Gijima that cases of "self-review" by an administrator are ordinarily governed by the legality principle and only exceptionally by PAJA. 52While it seems that at least one of the two possible exceptions is easily capable of overcoming the Gijima rule,53 the judgment has been criticized for aggravating the bifurcation of South African administrative law. 54he second development is the court's identification of a public law claim of unconscionable state conduct that is said to be based on rights other than section 33, and thus is not capable of engaging PAJA. 55Rather tellingly, however, this public law claim bears more than a passing resemblance to the legality principle, 56 and the case in which it was first recognized was one that should have been but was not argued under PAJA. 57The Constitutional Court has denied having avoided PAJA in that case, explaining that "this court did not ignore PAJA, but rather chose to dispose of the matter without having to answer the question whether PAJA applied". 58However, critics are unlikely to be placated by this rather feeble rationalization, especially since an almost identical explanation was given for the court's more obviously ruthless approach in Albutt. 59Nor are the authors convinced by it.Avoidance that is "chosen" is surely avoidance par excellence.
More recently, the highest court has again promoted the legality principle at the expense of PAJA.In Ledla it held unanimously that the Special Tribunal established under the Special Investigating Units and Special Tribunals Act 74 of 1996 enjoys review jurisdiction, but under the legality principle, not PAJA. 60While section 1 of PAJA contemplates review by "any independent and impartial tribunal established by national legislation for the purpose of judicially reviewing an administrative action in terms of this Act", the court did not consider whether the Special Tribunal could be regarded as falling within that definition.Taking the definition literally, the court simply ruled that the tribunal had not been established for that purpose. 61his article now turns to the main implications that seem to flow from the remaking of South African administrative law, starting with those that relate to the first phase of the reconstruction (the first two items discussed below) and then moving to those that concern the second phase (the last three items discussed below).It also points to a few corrective measures that can be taken to mitigate the problems of rivalry and avoidance identified above.

Implications and some corrective measures
Constitutional entrenchment and the legislative text One implication of the constitutionalization of administrative law is that the legislature is barred from adopting ouster clauses (that is, statutory provisions that are intended to prevent courts from reviewing specified administrative acts). 62The courts' power to review administrative action is thus secure against legislative curtailment.However, South Africa's constitutional framework means that the courts are secure in another more pervasive sense when they review administrative action.The fact that the courts' review power is entrenched in section 33 of the Constitution read with PAJA means that courts do not have to justify the exercise of that power through some contested device such as the ultra vires doctrine or the common law theory.Importantly too, the courts are less susceptible to assertions of judicial overreach when exercising a power with clear constitutional and legislative endorsement.When they act to control public power within this framework, there is less need for courts to "claim space and push boundaries" than was the position under the common law. 63he Constitution, however, does not simply confer rights to administrative justice, it mandates that these rights be given effect through legislation.That being so, the review of administrative action now ordinarily takes place under PAJA. 64In exercising this review power, courts must interpret and apply the text of that statute, albeit through the prism of the values of the Bill of Rights. 65wo opposing risks arise from this reshaped legal context.The first is that the restrictive elements of PAJA might be applied in a manner that undermines administrative law principles and underlying values.An example is provided by the Constitutional Court's judgment in Walele, 66 which concerned a procedural fairness challenge by a neighbour to the approval of building plans for a block of flats.Writing for the majority, Jafta AJ found that the approval of the plans did not trigger the threshold requirement for procedural fairness in section 3(1) of PAJA ("materially and adversely affects the rights or legitimate expectations of any person"). 67He held that, while the erection of the block might ultimately have such an impact on the neighbour, the mere approval of the plans did not. 68However, as the minority judgment in Walele demonstrates, 69 the procedural fairness argument could have been rejected in this case without adopting the majority's formalistic, narrow approach to section 3(1).
The second risk, which arises at the other end of the spectrum, is that the court might, in attempting to give effect to sound administrative law principles, adopt an overly strained interpretation of the PAJA text.The oft-cited judgment of the Supreme Court of Appeal in Grey's Marine 70 is arguably an example of such a case, for here the court appeared to find that PAJA's definitional element of "adversely" affecting rights was met where a benefit was conferred on the lessee of waterfront property.
The challenge for South Africa's courts is to interpret PAJA to advance the constitutional rights to administrative justice, and the values underpinning those rights, while respecting the legislative text.This approach is not only consistent with well-established constitutional principles, 71 it also allows for administrative law to develop in a manner that best promotes the underlying values of accountability, participation, transparency and administrative efficiency. 72his balanced approach is not to be mistaken for judicial timidity.On the contrary, an emphasis on underlying constitutional values should encourage South African courts to be innovative and bold, as demonstrated, for example, in the judgments of the Constitutional Court in Bato Star 73 and Joseph. 74In the former case, the court interpreted section 6(2)(h) of PAJA as involving review for mere unreasonableness even though the wording of this provision ("so unreasonable that no reasonable person could have so exercised the power or performed the function") might more naturally be taken to suggest unreasonableness of an egregious sort.The court did so with reliance on the unqualified language of the right to "reasonable" administrative action in section 33(1) of the Constitution.O'Regan J stated: "Even if it may be thought that the language of section 6(2)(h), if taken literally, might set a standard such that a decision would rarely if ever be found to be unreasonable, that is not the proper constitutional meaning which should be attached to the subsection.The subsection must be construed consistently with the Constitution and in particular section 33 which requires administrative action to be 'reasonable'.Section 6(2)(h) should then be understood to require a simple test, namely that an administrative decision will be reviewable if, in Lord Cooke's words, it is one that a reasonable decision-maker could not reach." 75e judgment in Joseph dealt with whether tenants had the right to make representations before the decision of a municipal entity (City Power) to disconnect their electricity supply, in the absence of a contractual nexus between City Power and the tenants.In finding that the tenants did have such a right, the court located the "right" that was adversely affected for the purposes of section 3(1) of PAJA in local government's constitutional and statutory obligation to provide basic municipal services. 76Skweyiya J, writing for a unanimous court, emphasized that proper regard to the constitutional right to administrative justice confirmed "the need for an interpretation of rights under section 3(1) of PAJA that makes clear that the notion of 'rights' includes not only vested private law rights, but also legal entitlements that have their basis in the constitutional and statutory obligations of government". 77He went on to describe the right at play in this case as a "public law right" to receive a basic municipal service. 78hese two cases show the benefits of substantive reasoning that is based on the Constitution. 79Both cases developed administrative law in progressive ways without, in the authors' view, unduly straining the text of PAJA.They illustrate that, while the legislative text might constrain judicial decision-making, the constitutional context may provide the impetus for the creative remaking of administrative law in appropriate cases.
As explained immediately below, the distinct constitutional and legislative setting in which administrative law review now takes place in South Africa also potentially has implications for the continued use of English case law.

Reliance on English precedent
English law has had a profound impact on South African administrative law.It formed the basis for the common law of judicial review in the pre-1994 era, and the South African grounds of review were largely derived from English law.However, the retrogressive tendencies of the common law as applied under apartheid moved a leading commentator, writing in 1984, to describe the administrative law of the time as "a pale reflection of the English law of a bygone era". 80Against this backdrop, English law was a crucial source of progressive administrative law principles that liberal minded judges could employ to constrain the power of the state.These principles included the power to review delegated legislation for unreasonableness 81 and the extension of procedural protection to legitimate expectations.The influence of English law has continued in the democratic era, with South African courts regularly drawing on case law from that jurisdiction. 82While this comparative approach is to be encouraged, one effect of South Africa's constitutional remaking of administrative law is that case law from other jurisdictions should be employed with careful regard to the different constitutional and statutory setting within which administrative law review now takes place in this country.
Bato Star 83 is a good example of such an approach.While noting that the wording of section 6(2)(h) of PAJA drew directly from the unreasonableness test in Wednesbury, 84 O'Regan J rejected that restrictive test in favour of a more straightforward and constitutionally sound formulation ("one that a reasonable decision maker could not reach"). 85Notably, this formulation was itself derived from another, more recent, English source: a dictum of Lord Cooke in Trader's Ferry. 86nother area that calls for a nuanced, context-sensitive approach to English law, and indeed the common law, is the meaning to be ascribed to the concept of legitimate expectations in modern South African law.In its seminal 1989 judgment in Traub, 87 the Appellate Division adopted the English law doctrine of legitimate expectations and apparently approved of Lord Fraser's dictum in Council of Civil Service Unions 88 that a legitimate expectation arises from either a promise by the public authority or a regular practice. 89The adoption of this doctrine meaningfully liberalized the scope of procedural fairness.As Corbett CJ put it, it provided a legal remedy "where the facts cry out for one" but where the common law formula of "liberty, property and existing rights" offered none. 90he legitimate expectation doctrine was regularly employed by South African courts in the ensuing decade before being appropriated by PAJA's drafters.Section 3(1) of the statute provides that procedural fairness applies where "rights or legitimate expectations" are adversely affected.A question that arises is whether the phrase "legitimate expectations", when used in section 3(1), is limited to the traditional conception (an expectation arising from a promise or regular practice) or whether it might bear a broader meaning.In Walele, the majority of the Constitutional Court suggested that it was so limited. 91Jafta AJ noted that "legitimate expectation" is not defined in PAJA and should therefore be given its ordinary meaning, as understood in previous judicial decisions. 92The problem with this approach is that it appears to foreclose the possibility that a person might be said to have a legitimate expectation where the impact of a decision on that person's interests is such that fairness would require a hearing. 93uperficial reliance on the concept of legitimate expectations in English law might suggest that Walele was correctly decided.Courts and academics in that country, for the most part, regard legitimate expectations and the impact of a decision on one's rights or interests as distinct bases for procedural fairness.As a leading English text puts it: "The distinction between the legitimate expectation and protectable interest may not always be clear, particularly if the two overlap.However, the underlying principles justifying one or the other are distinct.The legitimate expectation derives its justification from the principle of allowing the individual to rely on assurances given, and to promote certainty and consistent administration.Such a justification is distinct from that which permits a person to participate in a process of reaching a decision which may threaten his rights or interests." 94rists might thus regard the extension of legitimate expectations to cover decisions that have a dramatic impact on a person's interests as a perversion of doctrine. 95However, the constitutional and legislative setting in South Africa points in the opposite direction.Given that section 3(1) of PAJA refers only to "rights or legitimate expectations" and makes no mention of interests, if "legitimate expectations" is confined to its traditional conception, an affected person would not necessarily be entitled to a hearing even if the decision had a dramatic impact on her interests.Such an outcome would seem to be constitutionally problematic, particularly since section 33(1) of the Constitution grants everyone the right to procedural fairness in unqualified terms. 96here is thus much to be said for holding that a legitimate expectation arises, for the purposes of section 3(1) of PAJA, where the decision's detrimental impact is such that fairness requires a hearing. 97While this approach might not, as a matter of form, be consistent with English doctrine, it would, in substance, find support in the English law approach of granting procedural protection in deserving cases under the broad rubric of the duty to act fairly.Substance is, as always, more important than form, even if it might muddy doctrinal waters somewhat.

The proper use of the two main avenues to review
As explained above, avoidance of PAJA in favour of the legality principle has become commonplace.This is understandable given the considerable attractions of legality review, which include relatively indulgent procedural requirements and freedom from the burdensome administrative action inquiry.However, such avoidance is constitutionally unacceptable for more than one reason.
First, by contemplating the use of a more general constitutional principle in preference to a more specific legislative norm, the non-residual use of legality review violates the principle of constitutional subsidiarity. 98This principle has been affirmed by the Constitutional Court in a range of contexts, including the relationship between section 33 and PAJA. 99Subsidiarity contemplates the invocation of a higher or more abstract norm "only where the more local institution, or concrete norm, or detailed principle or remedy, does not avail". 100The legality principle is clearly a more abstract norm: from the start it was characterized as the more general equivalent of the right to lawful administrative action, 101 and the enactment of PAJA effectively widened the gulf between the general and the particular.Today, notwithstanding the subsequent inflation of the legality principle, , describing as "heretical" two pre-PAJA judgments that suggested that a legitimate expectation might arise from the nature of the decision and proposing that those sorts of cases should rather turn on "the duty to act fairly".96 The majority in Walele, above at note 66, para 30, noted that the applicant did not challenge the constitutionality of PAJA, sec 3 and that the court should thus proceed on the assumption that the provision did not offend the Constitution.This seems implicitly to acknowledge the potential constitutional difficulty brought about by the majority's own understanding of the phrase "rights or legitimate expectations".97 Klaaren and Penfold "Administrative justice", above at note 62 at 63-94.98 See, for example, Hoexter "The rule of law", above at note 42 at 65; Hoexter and Penfold Administrative Law, above at note 11 at 170-77; and Murcott and van der Westhuizen "The ebb and flow", above at note 42 at 49-53.99 See the discussion in My Vote Counts v Speaker of the National Assembly 2016 (1) SA 132 (CC), paras 54-66.100 Id, para 46 (Cameron J). 101 Fedsure, above at note 13, paras 58-59.
PAJA remains a far more specific and more detailed statement of administrative justice than legality review.
Furthermore, PAJA enjoys all the democratic legitimacy of a statute.Even in the absence of a constitutional mandate to provide for judicial review, there is an argument that "in the interests of democracy a body of law of this kind should receive the imprimatur of a legislature and should not be fashioned by the courts alone". 102PAJA has that imprimatur and, while it is not a particularly voluminous piece of legislation, it cannot be described as minimalistic.In comparison with the more laconic US Administrative Procedure Act of 1946, for instance, the grounds of review in PAJA are set out in considerable detail.Its makers also went into detail about the ingredients of procedural fairness, both in individual cases and in cases affecting the public.Accordingly, sidestepping the statute in favour of legality review, which is entirely judge-made, not only flouts subsidiarity but also disrespects the legislature. 103Moreover, it undermines the constitutional structure of administrative law review, since section 33(3)(a) of the Constitution makes plain that legislation is to provide for the review of administrative action. 104s various commentators have urged, 105 the courts should be applying this orthodoxy uniformly and consistently.In recent years the strongest statements about the proper use of PAJA have tended to come from the Supreme Court of Appeal.It is judges of that court who have pointed out that the "proper place for the principle of legality in our law is for it to act as a safety net or a measure of last resort … It cannot be the first port of call or an alternative path to review, when PAJA applies", 106 and similarly that litigants and courts must "start with PAJA and only when PAJA does not apply should they look to the principle of legality and any other permissible grounds of review lying outside PAJA". 107What is needed now are similarly unequivocal statements from the Constitutional Court.Sad to say, South Africa's highest court seems to have done more than any other actor to foster doubt about whether administrative justice will be best served by "embracing" or "sidelining" PAJA. 108king the boundaries of administrative action seriously Avoidance of PAJA in favour of legality review is not only constitutionally objectionable, it is also deleterious from a more practical point of view.Applying legality review to acts that might or would qualify as administrative action under PAJA is an inflationary practice.This is because using legality review where PAJA ought to apply exerts pressure on the courts to make the legality principle do the work that would ordinarily be done by PAJA.In the absence of limiting doctrine or precedent, the inclination of any court is to apply the standard of review that it considers appropriate to the matter before it.The temptation, then, is to make legality review mimic or mirror the grounds of review enjoyed under PAJA, thus leading to increasingly wide grounds of review.While the authors can only speculate, the misapplication of legality review in this manner surely explains at least some of the inflation of the content of legality that has occurred in the last decade or so.
A related problem is that there is apparently no going back once a ground of review has been created or expanded.Whether in PAJA or elsewhere, grounds of review seldom shrink or decrease but only seem to expand their content or sphere of application.Given its extraordinary development in recent years, the legality principle has arguably become more rigorous and invasive than it needs to be to act as a safety net for non-administrative action: the executive or distinctively political conduct for which the principle was originally devised.That amounts to overkill, which may in turn lead to accusations of judicial overreach in such matters.
The concept of legality review reflects an eminently principled approach: the exercise of all public power must comply with basic administrative law principles.This unqualified approach has, up to now, held its ground even in cases involving sensitive political decision-making, including a presidential decision to appoint a commission of inquiry 109 and decisions involving foreign policy. 110owever, a recent Constitutional Court judgment suggests that the universal application of the legality principle may be under threat, for the court left open the question whether a decision to reshuffle the Cabinet was subject to rationality review, despite the president's concession that it was. 111This threat is likely to be increased by the expansion of the grounds of legality review beyond its safety net function, particularly if those review grounds apply in the same manner to all conduct that falls within the ambit of legality review.
One way out of this conundrum, as argued below, is to restore some balance by means of the notion of variability.Meanwhile, the authors suggest that this inflationary effect is a strong argument for drawing the boundaries of administrative action properly, instead of simply relying on legality review as a substitute.
The decision of the Supreme Court of Appeal in Scalabrini 112 perhaps illustrates the point.The court in this case held that a decision to close the Refugee Reception Office in Cape Town did not amount to "administrative action", owing to its policy-laden nature. 113Applying the legality principle, the court then struck down the decision on the basis that it had been irrational not to consult with certain non-governmental organizations that had special knowledge about asylum seekers in Cape Town. 114While Nugent JA emphasized that this did not imply "a general duty on decision makers to consult organizations or individuals having an interest in their decisions", 115 it will be difficult to distinguish between the organizations in Scalabrini and parties affected by a host of other types of decisions who may well have special knowledge relating to the subject-matter of the decision.Rather than applying procedural rationality in this expansive manner under the banner of legality review, it might have been more appropriate for the court to hold that the closure decision was susceptible to PAJA review and then to apply suitable deference in assessing the reasonableness of the decision. 116veloping variability under the legality principle For the reasons given above, there is a pressing need to develop the concept of variability to ensure that the degree of judicial scrutiny is appropriately calibrated to the relevant type of administrative act.
Variability, closely related to deference, expresses the notion that the grounds of review "need not be applied in an all-or-nothing fashion, and that the intensity of judicial scrutiny may vary according to the context". 117In relation to PAJA and its universe of administrative action, this proposition has received ample recognition from the South African courts.As in other parts of the common law world, 118 the variable content of procedural fairness and the variable meaning of reasonableness are well established.Section 3(1)(a) of PAJA recognizes that what is fair "depends on the circumstances of each case", and the Constitutional Court has acknowledged that both procedural fairness and reasonableness are "context specific". 119Variability is increasingly being accepted in other areas too; it is fairly clear, for instance, that the answer to whether there has been "compliance" with a statutory formality or whether an administrator has provided "adequate" reasons will necessarily vary with the factual and legal context. 120nder PAJA, the chief advantage of variability is that it discourages all-or-nothing reasoning and the concomitant risk of no entitlement to administrative justice.In this regard, the highest court made it plain in Joseph that the threshold concept of administrative action ought not to be deployed with the aim of achieving administrative efficiency or, stated more negatively, in order to avoid the "spectre of administrative paralysis". 121As Skweyiya J explained for the court, it is preferable for practical concerns such as these to inform the content of duties imposed by administrative law (in that instance, procedural fairness) rather than the scope of administrative action: a concept "fundamentally determined by the relationship between the administrative State and its citizens", and one that ought not to be delimited too strictly. 122Under PAJA, then, variability works to discourage courts from drawing the boundaries of administrative action too narrowly and thus setting too high a threshold for the application of the statute.
Turning to legality review, the existence of this avenue may itself be regarded as a manifestation of variability: that is, a kind of variability between rather than within standards of review. 123When it was first conceived, legality review made a sharp contrast with PAJA: it operated as a basic safeguard that permitted a court to "defer to the government at the margins without relinquishing its supervisory role completely". 124There was no real need to ponder possible variability within its grounds of review, which were then of the most rudimentary kind.However, now that the principle has developed into a much more elaborate safeguard, one that rivals PAJA and matches its content to a large extent, it seems that the recognition and development of variability within the grounds of legality review has become a necessary and perhaps inevitable step, particularly given the wide range of public conduct that is subject to legality review.national security, economic policy, and so forth), the court is bound to exercise considerable deference when deciding whether the purpose of the decision is legitimate and whether the decision is likely to serve that purpose.In such contexts, the considerations of democratic principle and institutional competence are particularly pressing and consequently oblige the court to pay a higher degree of respect to the empirical and normative judgments of the executive … By contrast, in less 'political' contexts -that is, in cases where the considerations of institutional competence and democratic principle are less weighty -the courts need not tread as lightly when evaluating the purpose and effect of the law or conduct in question." 125e development of a doctrine of variability along these lines would surely help to counteract the inflationary effect discussed earlier: the results of the near-constant temptation to make legality review do all the work of PAJA.This option also seems far more attractive than the obvious alternative, the development of a doctrine of non-justiciability for certain exercises of public power.Such a doctrine carries the risk of unprincipled application, is inherently less principled, undermines accountability, and would in any event be inconsistent with South Africa's existing jurisprudence.After all, the Constitutional Court established the justiciability of every exercise of public power two decades ago. 126

Conclusion
This article has explored the remaking of South African administrative law and, more particularly, judicial review, since constitutional democracy was introduced in 1994.It characterized the construction of the constitutional and legislative framework, and the courts' interpretation of that framework, as the first phase of the remaking.The second phase encompasses the courts' recognition of the legality principle and their subsequent development of its content, resulting in an elaborate avenue to review parallel to PAJA, with attendant problems of rivalry and avoidance.
The article went on to identify and discuss some of the more significant implications of each of these phases of reconstruction.As to the first phase, it argued that the constitutional and legislative context has important implications for judicial interpretation and reasoning, as well as for the use of English precedent.One product of this phase is that, while the constitutional right provides fertile ground for the progressive development of administrative law, the text of PAJA constrains judicial decision-making somewhat.As to the second phase, it argued in favour of respecting constitutional subsidiarity in using the two competing avenues to review, and pointed out the need to take the boundaries of administrative action seriously and to develop the concept of variability under the legality principle.
In the authors' view, all these measures are likely to advance the coherence and effectiveness of administrative law review and its jurisprudence.They should assist in ensuring that the standard of review is appropriately calibrated to the nature of the administrative act, and that the principled position that legality review applies to the exercise of all public power is not jettisoned in favour of a doctrine of non-justiciability.