The doctrine of legitimate expectation operates as a control over the exercise of discretionary powers conferred upon a public authority. The typical reason why discretionary powers are conferred upon a public authority is to ensure that they are exercised having due regard to the particular circumstances of individual cases coming before the decision-maker – ie in circumstances where Parliament was not confident at the time of passing legislation in predicting all circumstances and how individual cases should be resolved.
It is often difficult to tell in advance of concrete situations arising precisely how an authority should act; and that may be as true for the authority as for Parliament itself.
This reasoning is inherent in the rule forbidding a public authority which has a discretion and adopts a policy as to its exercise from following that policy without having due regard to the specific facts of the particular case: British Oxygen  AC 610.
The doctrine of legitimate expectation is the converse of this situation: it operates to say that, subject to certain conditions, a public authority which adopts a policy should be required to follow and apply that policy in cases subject to it, without being permitted to depart from it.
It applies in cases where the decision-maker has committed itself in advance to a particular course of conduct in a particular class of case (defined in more or less general terms), without reference to the specific facts of individual cases.
The key issue, therefore, in legal terms, is to explain by reference to normative considerations when and why (or why and, therefore, when) the policy maker will be required to follow its own pre-determination of the outcome of the case without being free just to change its mind when the specific facts of an individual case are before it.
In addressing that issue, it is important to understand that the problem which presents itself is one of public law, and that private law analogies are of limited assistance: see R (Reprotech (Pebsham) Ltd v E Sussex CC  1 WLR 348, HL, at  per Lord Hoffmann (indicating an important shift from the cases in the infancy of the development of the doctrine, where the private law analogy was used as the normative foundation for the doctrine: see esp. Re Preston  AC 835, 865-7 per Lord Templeman).
Viewed from the perspective of public law, the protection of legitimate expectations is both greater but also less than the protection of comparable interests under private law (particularly in contract or estoppel cases):
The protection is greater, in that in some cases no absolute promise or consideration or detrimental reliance may be required. It may be sufficient to say that the adoption of a policy by the decision-maker informs and strengthens the application of general requirements of lawful behaviour on the part of public authorities, including in particular (in the case of representations as to the procedure to be adopted) the requirement that a public decision-maker act fairly and (in the case of representations as to the substantive outcome in a class of case) the requirement that a public decision-maker act consistently and rationally (cf Matadeen v Pointu AC 98.
In each sort of case, it seems that the specific adoption of a policy by the decision-maker will strengthen rights which the individual enjoys under general public law. But then, it is important to identify to what extent the adoption by the public authority of a policy in advance adds to the rights which the individual already enjoys under general public law;
The protection is less, in that the interest of the individual in having the policy or representation complied with in their case may be overridden by countervailing public interest considerations. Short of a binding private law contract or estoppel, the hands of the public authority are not absolutely bound by its advance declaration of how it will deal with particular cases.
It can depart from its policy, if at the time of decision it considers that there is some overriding reason of the public interest which requires an approach different from that declared in general terms by it in advance. But then, it is important to identify to what extent it is open to the public authority which has made representations or adopted a policy in advance should have the freedom to depart from it.
In institutional terms, where a public authority has conferred upon it a wide discretionary power covering a large number and variety of cases, it may be important for the public authority’s own coherent and consistent approach to decision-making that it adopts a policy as to how it will act – at least in the usual run of cases.
But, the greater the extent that the doctrine of legitimate expectations may result in the public authority’s hands then being held by the courts to be tied in law, the greater the wariness authorities may have about adopting policies as to how they approach their decision-making in a particular area.
Also, a public authority may, in practical terms, face the same sort of difficulty in predicting precisely in advance how its decision-making powers should be exercised in particular cases that Parliament faced when conferring discretionary powers upon it in the first instance: ie it cannot be known with certainty in advance how particular cases ought to be decided (which is at least part of the rationale for conferring a discretion in the first place, rather than Parliament itself prescribing in advance in legally binding legislation what the outcome should be in particular cases).
Parliament having chosen to confer a discretion rather than creating a binding duty, it would be odd if the doctrine of legitimate expectations could operate so that the public authority itself, by adopting a policy, creates a binding duty for itself.
Again, the more the doctrine operates to bind the hands of the public authority, the more it implicitly engages the courts and the decision-maker in creating something approaching the kind of binding duty which Parliament chose not to create.
Given these general points, one would expect the doctrine of legitimate expectations to develop by adaptation in the light of the weight of the conflicting public law reasons in favour of greater or less decision-making flexibility for public authorities at the point of actually deciding what to do in specific cases. In broad terms, it is submitted that this is what one finds in the case law.
A distinction between procedural and substantive legitimate expectations?
In general terms, it is desirable for public authorities to do what they have declared they will do. That assists citizens to plan their affairs and fosters trust and confidence in the administrative authorities. It can generally be said to be a feature of good administration that public bodies “deal straightforwardly and consistently with the public” (see Abdi and Nadarajah v Secretary of State for the Home Department  EWCA Civ 1363, per Laws LJ at , where he describes it as a “requirement”).
But these are factors which have to be balanced, at the level of legal principle, against competing factors, pointing in favour of the desirability of public authorities having flexibility to adapt to changing circumstances and changing assessments of the public interest – which is the main underlying reason they will have had a discretion conferred upon them. It can generally be said also to be a feature of good administration that public bodies accurately assess the overall public interest as to how their powers are exercised, in the light of all the relevant facts available to them at the time when they decide what to do.
When taking account of this latter sort of consideration, I would suggest that it is very much stronger in the case of representations as to substantive outcomes rather than representations as to the decision-making procedure to be followed. That in turn suggests that it is important to maintain a distinction between procedural legitimate expectations and substantive legitimate expectations.
It is well known that the Court of Appeal in its seminal exposition in R v N and E Devon Health Authority, ex p Coughlan  QB 213 emphasised the difference between procedural legitimate expectations (based on representations as to the procedure a public authority will follow before it takes a decision on the substantive merits of a particular case) and a narrower class of protected substantive legitimate expectations (based on representations as to the decision which a public authority will take on the substantive merits of a particular case).
Where a public authority represents that a particular procedure will be followed before it decides what to do on the merits of an individual case, the intrusion on its decision-making discretion to decide what to do in that case having regard to the overall public interest is less where it is held to be bound by its representations than where it is held to be bound by its representations as to the substantive decision it will take.
In the latter class of case (representations as to substantive outcome), the public authority is being held to its assessment of the overall public interest taken in advance (sometimes, a long time in advance) and in more or less general and abstract terms, and that advance and general assessment is found to trump the specific and precise assessment of the public interest which it makes when the particular facts of the individual case are directly before it – even though the public authority has not become functus officio simply by adoption of a policy.
Indeed, this difference between procedure and substantive outcome has for a long time lain at the heart of the development of public law more generally, and is reflected in the development by the courts of comparatively full and binding standards of fairness in the decision-making process (where the courts have taken themselves to be the primary decision-makers as to the content of those standards: see eg R v Panel on Take-overs and Mergers, ex p. Guinness plc  1 QB 146) as compared with relatively light standards of review as to the substantive merits as expressed, in particular, in the Wednesbury formula.
The conflict of substantive legitimate expectations with the ability of a public authority to assess the balance of the public interest and how best to behave in the light of the actual circumstances of a particular case becomes particularly acute where the enforcement of the substantive legitimate expectation would involve the court in overriding the public authority’s judgment how scarce resources should be utilised in the face of competing claims.
That is a judgment which, on the face of it, Parliament has allocated to the public authority rather than the court, and where the public authority rather than the court is best placed to weigh all the competing considerations.
Given the nature of the analytical exercise when deciding whether and to what extent a public authority should be treated as bound by representations it has made in advance, and the balance which the doctrine of legitimate expectation has to strike in concrete terms between treating a public authority as bound by its earlier (rather than present) assessment of the overall public interest and allowing it freedom to decide how to behave in the light of its present assessment arrived at in light of the specific facts of particular cases, one would expect in principle to find a difference in approach as between procedural and substantive cases.
The pressures for preserving freedom of choice for a public authority in the present are far less in the case procedural legitimate expectations than in the case of substantive legitimate expectations (directed to the outcome on the merits).
The more a public authority is to be treated as bound by its advance decision as to the substantive merits, the more one would expect the conditions for that binding outcome to reflect the ability of the public authority to have accurately assessed in advance what the balance of the public interest should be and the clarity with which the public authority had in fact addressed that relevant question. That suggests that, where one is dealing with substantive legitimate expectations, the courts should focus very carefully on factors which affect the quality of the advance “decision” taken by the public authority, and the appropriateness of treating that advance “decision” as in some sense binding upon the public authority when taking an actual decision in a concrete case in the present.
The more the public authority has, at the advance stage, had clearly in mind the particular ramifications of its representations or policy for the type of concrete decision which it eventually has to take and has deliberately chosen a particular outcome, the more readily one would regard it is as legitimate for the courts (having regard to other considerations, including the general desirability of public authorities abiding by pre-declared policies) to treat the public authority as in some sense bound by that advance “decision”. Conversely, the further away one is from that situation, the stronger the argument for retention of freedom of action for the public authority in the light of full information about current circumstances.
This, I would suggest, is an important part of the analytical background to the stress in substantive legitimate expectation cases upon the quality of the information available to a public authority before it is treated as prima facie bound by advance representations it makes as to the substantive outcome in a case (see eg R v IRC, ex p. MFK Underwriting Agencies Ltd  1 WLR 1545), and upon the precision, clarity and limited focus of the representation relied upon (MFK Underwriting; Coughlan at -).
Since the impact upon the general interest in an authority having freedom to determine and act upon the public interest as it determines it to be at the point of actual decision is much less where the authority’s representation only goes to the procedure to be followed rather than the substantive outcome of any case, these strict requirements can be relaxed.
It is more acceptable to require the authority to abide by its representations and policy when made in general terms and to the world at large. Similarly, it is more acceptable for a legitimate expectation to be founded upon an established course of conduct in the past, rather than any specific promise as to conduct for the future: see eg O’Reilly v Mackman .
2 AC 237, 275 (there is no authority which has found a substantive legitimate expectation to have emerged from past conduct: the closest example is R v IRC, ex p Unilever plc  STC 681, which the Court treated not as a legitimate expectation case - even in the context of an extremely precise and focused course of conduct known by the IRC to have engendered very specific expectations on the part of the taxpayer claimant - but a case of conspicuous unfairness, tantamount to irrationality, in failing to follow the same established practice in relation to permitting tax rebate claims to be made in a particular way, so that the claims were defeated at huge cost to the taxpayer).
For reasons of both authority and principle, therefore, one might respectfully doubt whether an approach which seeks to assimilate the test for holding that a protected legitimate expectation has been created by representations by a public authority as to procedure ought to be adopted in relation to representations as to substantive merits: contrast Abdi & Nadarajah v Secretary of State for the Home Department  EWCA Civ 1363 at - per Laws LJ, esp. at  (“The dichotomy between procedure and substance has nothing to say about the reach of the duty of good administration”).
On the analysis presented above, the dichotomy between procedure and substance does have important things to say about the reach of the duty of good administration; and there is good reason why that dichotomy should inform the test for determining when a protected legitimate expectation will be held to have arisen.
How may a legitimate expectation be overridden?
In Abdi and Nadarajah, Laws LJ said at , “The principle that good administration requires public authorities to be held to their promises would be undermined if the law did not insist that any failure or refusal to comply is objectively justified as a proportionate measure in the circumstances.” There are three problems with this part of the judgment which I would respectfully wish to focus on.
First, to the extent that (as Laws LJ does earlier in his judgment) one assimilates the test for a protected legitimate expectation to arise in a substantive expectation case with that for a procedural expectation, the adoption of a standard of proportionality represents a greater inroad by the courts into the practical decision-making powers of the public authority charged by Parliament with taking the decision in question. In relation to procedural expectations, the courts’ authority to apply their own standards of fairness is well established (see Guinness, supra), and this represents a clearly acceptable practical compromise between fairness as a value of good administration and accurate substantive decision-making by the relevant public authority designated by Parliament as a competing value of good administration.
It seems straightforward for the courts to extend that sort of approach to cover procedural legitimate expectations, and insist upon a high standard of justification (perhaps expressed in a proportionality rubric) before such legitimate expectations are overridden. To say that a departure from a substantive legitimate expectation can only be justified on a basis of proportionality, rather than by reference to an ordinary rationality standard, is much more questionable.
To the extent that relatively restrictive criteria are applied before a substantive legitimate expectation is found to have arisen, such an intensive standard of review may be justified – and this was, in effect, the balance struck by the Court of Appeal in Coughlan.
But if those restrictive criteria are to be relaxed (as Laws LJ suggests), it is not immediately obvious that the proportionality standard of review (if that is said to be different from ordinary rationality review) will be the appropriate one. At some point, argument about the relevant standard shades into the arguments about whether proportionality should be adopted as a general test of lawfulness under domestic administrative law.
Second, one may doubt just how useful a test of proportionality will be in these cases. To say that the test is one of “proportionality” begs the question, proportionate to what? If it is said, “proportionate to a legitimate aim”, one has to acknowledge that in some cases the legitimate aim will be extremely diffuse, to the point of not being a helpful analytical tool (eg a public authority, on consideration of the full facts, may simply decide it does not wish to expend limited public funds in a particular way; or there may be a change in the political complexion of a public authority, and it simply makes a different assessment of the public interest in a given case).
Perhaps the test should be expressed more broadly, as one directed to severe and unjustifiable substantive unfairness which would be suffered by an individual if a public authority departs from its representation, with a focus then upon spelling out the factors (eg detrimental reliance, the nature of the public interest being pursued) relevant to such an assessment.
Detrimental reliance was identified by Peter Gibson LJ as an important factor in substantive legitimate expectation claims in R v Secretary of State for Education and Employment, ex p. Begbie  1 WLR 1115, 1124.
A proportionality approach might be said to be rather narrowly expressed for other reasons as well. Sometimes, the reason that a legitimate expectation may be departed from is that the public authority behaves fairly in doing so, rather than because the departure is itself proportionate to some overriding public interest.
It may by sufficient for a public authority to afford an individual to whom a representation has been made an indication that the public authority is minded not to adhere to its representation and an opportunity to make representations why the public authority should adhere to it. If the basis for a claim to a legitimate expectation is a representation by a public authority how it will act, there may well be situations in which it will be fair for the public authority to retreat from that position, provided that it affords the individual concerned an opportunity to make representations to persuade it otherwise.
If there has been no detrimental reliance by the individual, and the public authority has acted fairly in this way, it is difficult to see why (in the ordinary case) it should be held bound by what it said some time ago. This point is underlined by the approach of the courts in relation to relief, particularly where a substantive legitimate expectation is concerned: see paragraph 21 below.
Third, a proportionality approach is difficult to reconcile with recent House of Lords authority. In R (Mullen) v Secretary of State for the Home Department  1 AC 1 the Secretary of State recognised that the claimant fell within the scope of his policy on payment of compensation to persons wrongfully convicted, but (since it was established that the claimant had in fact been involved in a terrorist related offence) did not wish to pay him compensation in respect of his wrongful conviction (ie wished to depart from his policy statement in light of the particular features of the claimant’s case).
The Secretary of State gave the claimant an indication that he proposed to do that, invited representations on the point, and having considered the representations received decided that he would depart from his policy in that case. He did not withdraw or restate the general policy.
The House of Lords held that the Secretary of State was entitled to decide not to apply his policy in the claimant’s case, having afforded the claimant a fair opportunity to make representations on the point: on the substance of the decision, the House applied a rationality test (see esp. - per Lord Steyn).
The form of relief where there is a violation of a legitimate expectation
Ordinarily, in a procedural legitimate expectation case, the public authority will (absent some sufficiently compelling and overriding public interest) simply be required to comply with its representations (or established practice) as to the procedure it would follow. A failure to do so will be treated like a failure to comply with common law standards of fairness.
The position is more complicated in a substantive legitimate expectation case, where a representation as to the substantive outcome on the merits is concerned. It may be that the public authority will be held to be required actually to confer the substantive benefit as it has said it would. But that will bring the court into tension with other principles of public law, which require that the primary operative decision should be that of the public authority in the light of all the circumstances known at the time of the decision.
Therefore, it is submitted that usually a particularly compelling interest of the individual will be required before the court may feel confident that it is appropriate to override the (current) decision of the local authority. The sort of interest which means that the fair balance to be struck between the interests of the individual (and general value in public authorities acting in accordance with their prior representations) and the public interest of flexibility of decision-making in the light of all the circumstances at the time of decision comes down in favour of compelling the public authority to act contrary to its present judgment and in line with its previously declared policy will usually be some form of detrimental reliance on the part of the individual.
One may then become involved in a debate about the sort of detriment which ought to count for these purposes: compare R (Bibi) v Newham LBC  1 WLR 237 and R (Rashid) v Secretary of State for the Home Department  EWCA Civ 744.
The tension referred to above becomes all the more acute when the enforcement of the substantive legitimate expectation according to its terms involves, in practice, the court allocating to the claimant some scarce resource, which will mean that other possibly deserving individuals will lose the opportunity to have that resource allocated to them. That was the position in Bibi, supra.
The solution of the Court of Appeal was to meet that tension by declining to grant relief in substantive form by requiring the local authority to honour its promise, and instead granting relief in procedural form (obligation on local authority to reconsider claimant’s case, with opportunity for representations, and a requirement to give reasons explaining any decision to depart from its promise).
An important area for the future development of the law in relation to substantive legitimate expectations will be the articulation by the courts of the principles according to which the proper remedial reaction to a substantive legitimate expectation will be procedural in form, and those according to which the proper remedy will be direct enforcement of the public authority’s promise or policy on the mandatory order of the court. Again, it is submitted that detrimental reliance is likely to be an important consideration in relation to this dividing line.
-----------------------  For a more complete expression of some of the underlying ideas in this lecture, see P. Sales and K. Steyn, “Legitimate Expectations in English Public Law: An Analysis”  Public Law 564.
 Cf Pascal, Pensées (trans. A.J. Krailsheimer, Penguin 1966), no. 619: “All their principles are true, sceptics, stoics, atheists etc … but their conclusions are false, because the contrary principles are also true.”