faculty of law, university of oxford

Almost every constitution in the world confers a power to pardon.  Pardon powers are found in the constitutions of old states and new states, Western states and non-Western states, states with a Christian tradition and states without one. Pardon powers are part of the constitutions of states as diverse as France, Indonesia, Peru, Russia, the United Kingdom, and the United States. 

Pardon powers share several features. First, the exercise of the power has the result of lifting or lessening criminal liability. Second, the power achieves this result not by changing the law, but by setting it aside in a particular case. Third, the power is held by a branch of government other than the judiciary – the executive, normally, or the legislature, less commonly. Finally, in its traditional mould, a pardon power is wholly arbitrary. Its use is unreviewable and unconstrained. (For recent departures from this traditional form of the pardon power, see my post here.) 

So understood, pardon powers seem to conflict with two of the most basic principles of constitutionalism. Contrary to the separation of powers, a pardon power gives to a branch of government other than the judiciary a role in determining criminal liability in particular cases. Contrary to the rule of law, a pardon power is traditionally neither controlled by nor ruled by law.  

Hence the pardon paradox: one of the most common constitutional power is at odds with some of the most fundamental constitutional principles. Pardon powers are everywhere but seem to properly belong nowhere. I argue that the paradox has a solution. Appearances notwithstanding, pardon powers threaten neither the separation of powers nor the rule of law. My argument has three steps.

Gap cases

The first step is that even an ideal optimal legal code will require suboptimal results (ie, results which there is overall reason not to achieve absent the code) in some cases. In such cases there is a “gap” between what is required by the code we should have and what we should do absent the code.  

Gap cases are inevitable for familiar reasons. A code which perfectly tracks the underlying reasons in every case will be very complicated. It will be time-consuming to apply, as a result. People will make mistakes about what it requires. That will lead to uncertainty about what others will do. These costs may be significant enough to favour a simpler, easier to use code instead. Crude and usable may beat accurate but hard to apply. 

Consider the regulation of assisted suicide. Sometimes, it is morally permissible for one person to help another person commit suicide. Think of cases in which a person has a clear desire to die and is of sound mind, terminally ill, in terrible pain, etc. But let us assume – purely for the sake of argument – that were the law to make a limited exception for this sort of case, some people would be unfairly pressured to kill themselves. Overall, it could be best to have a blanket prohibition on assisted suicide. 

Selective deviation

The question then becomes: how should we respond, when an optimal code requires a suboptimal result? There are broadly three options.

At one extreme is adherence to the code in every gap case. We would never set the prohibition on assisted suicide, say, even if it was clear that it yielded a bad result. This strategy the advantage of preserving the code as a source of guidance and stability. But it comes at too high a price. A code will require many suboptimal results. Deviating in the occasional case does not significantly undermine the code’s predictability and stability. 

At the other extreme is deviation from the code in every gap case. Whenever it was clear that it was morally permissible for a person to help another to commit suicide, for example, we would set aside the legal prohibition on doing so. The problem is that, if we always deviate from the code in gap cases, then people will come to expect deviations in these circumstances. As a result, they will not look to the code for guidance; they will look to official decisions. We will have ended up altering the code, not in form, but in substance. 

So, we shouldn’t always adhere to an optimal code. Nor should we always deviate from it. We should adhere in some cases and deviate in others. We should “selectively deviate”, in other words. The most important thing about selective deviation is that it leads to treating like cases differently. For example, in some cases when a person is terminally ill, suffering terribly, etc., we would set aside the prohibition on assisted suicide. In some like cases, we would not. Treating like cases unalike may sound terribly unjust. But the alternatives – adhering to the code come what may, and deviating so often guidance and stability are threatened – are even worse. This is the second step in the argument. 

Justifying the pardon power

The final step is to show that pardon powers are the right vehicle for selective deviation. 

What we want is a power to set aside a code without altering or changing it. A pardon power fits the bill. The power we want must not be regulated by any rules. That is because, were it subject to rules, its use would be predictable. Were its use predictable, it would undermine the code. Again, a pardon power is exactly what we want. There is an irony here. Were the pardon power ruled by law, the code would not rule over citizens. So, the rule of law itself favours an uncontrolled pardon power over a controlled one. The implication is that we should not extend judicial review to pardon decisions, at least on rule of law grounds.  

It is not enough that powers to set aside the law *can* be used inconsistently; the must *in fact* be used that way. The judiciary is institutionally diposed towards consistency in its  decision-making. This is precisely what we do not want. It follows that we should give a power to set aside the law to a different branch of government, and thus to either the executive or legislature. So, the separation of powers, as a principle which requires power to be allocated to the institution best suited to wielding it, itself favours a pardon power. 

Overall, the features of pardon powers which seemed to be in tension with the rule of law and the separation of powers are in fact justified based on those same principles. The paradox is dissolved. 

I set out this argument in detail in a draft article, and I would be very grateful for comments or criticisms. 

This post first appeared on I-CONnect here.

I’ve been lecturing lately on constitutional conventions, one of my favourite topics in the constitutional law course. A couple of weeks ago I sketched what I think is a new argument for why conventions aren’t laws. I’m reasonably confident of the argument, but far from certain, so I thought I’d share it more widely. If you spot any flaws, please let me know!

Dicey’s Argument

Dicey distinguished laws and conventions based on enforcement. Laws, he said, ‘are enforced by the Courts’. Meanwhile, ‘conventions, understandings, habits, or practices’ aren’t ‘laws at all since they are not enforced by the Courts’.

Taken literally, this is silly. Some laws are never litigated. Section 467 of the No One Cares Act  will never make it to court, so it won’t be enforced. Let’s be charitable and take Dicey to mean that a norm is a law only if courts would enforce it were an appropriate case to arise it. Also, as Colin Munro noted, ‘enforce’ doesn’t capture how courts engage with many laws. It’s awkward to say that a court ‘enforces’ a permission or the conferral of a power, for example. ‘Apply’ or ‘act on’ are better, more encompassing terms. So, again, let’s be charitable. Let’s take Dicey to distinguish between laws and conventions based on their applicability or courts’ willingness to act on them.

Putting these two points together, here’s the argument:

  1. A norm is a law only if courts would apply it, in an appropriate case.
  2. Courts don’t apply conventions, even in appropriate cases.
  3. So, conventions aren’t laws.

I’ll call this “Dicey’s argument”. There are two ways to attack the argument: attack premise 1 or attack premise 2. I think that premise 2 is true in the United Kingdom. That leaves premise 1.

A good objection

 The usual way to attack premise 1 is via counterexample. There seem to be laws which courts won’t apply. As a result, courts’ unwillingness to apply conventions doesn’t distinguish them from laws. This objection originated with Sir Ivor Jennings, who identified a couple of counterexamples. Scholars since have added others. Drawing them together, we have the following list of laws which courts are unwilling to apply:

  • Laws protected by an ouster clause. Suppose there is a statutory duty imposed on an official. Suppose there is also a very clear “ouster clause” – a statutory provision instructing courts not to consider or question whether the official fulfilled the duty. Courts are bound by the ouster clause, so they wouldn’t apply the statutory duty. Nonetheless, the statutory duty is a legal duty. Until recently, this was theory. Since R (Privacy International) v Investigatory Powers Tribunal, it is practice.
  • The law and custom of Parliament. Parliament has a wide range of norms governing its own affairs and procedures. These norms are known as the ‘law and custom of Parliament’. Even though many of the norms are laws, they are not enforceable by courts.
  • Directive principles. Directive principles impose a legal obligation on a legislature or government to pursue a social value (e.g., housing, healthcare, equality) or an equally broad aim. Even though they’re legal norms, directive principles aren’t enforceable by courts. Many constitutions have directive principles, including those of India, Ireland, and South Africa, as well as the United Kingdom.

To these three traditional examples, I’ll add a fourth:

  • Disapplied statutes. The Merchant Shipping Act 1988 (MSA) was ‘disapplied’ under the European Communities Act 1972 (ECA). As things stand, the MSA is not applicable by courts. Indeed, the MSA has never been applicable by courts, because it has never been consistent with EU law. And yet the MSA was validly enacted, and it has never been repealed. The norms it generates are valid legal norms, even though they are not applied by courts. This is not a rare case: there are many disapplied statutes.

Perhaps someone would suggest that the MSA is not a valid statute. They would argue that the MSA was invalidated by the ECA, just as if it had been repealed. That would be a mistake though. Suppose the ECA itself were repealed (and no relevant transitional arrangements were enacted). What would happen to the MSA? If the MSA had been invalidated or repealed, it would not come into force upon the ECA’s repeal. A repealed statute does not “revive” upon the repeal of the repealing statute. But the MSA would come into force upon the ECA’s repeal. The legal norms it creates would then be applicable. That’s one of the differences between disapplication and repeal: repeal of a disapplying statute “revives” the disapplied statute, whereas repeal of a repealing statute doesn’t revive the repealed statute. Because the ECA’s repeal would revive the MSA, we can infer that the MSA is a valid statute.

A revised argument

If even one of these examples holds good, then Dicey’s argument fails. We need to try again to distinguish conventions from laws. Here’s my best effort.

Look again at the examples in the last section. There’s a common feature, beyond unenforceability. It’s that there are two laws, structured like this:

Law A: imposes a duty on so-and-so to do such-and-such.

Law B: directs courts not to apply Law A.

We see this structure in each case:

Law A Law B
Ouster Statutory duty (e.g. in the Regulation of Investigatory Powers Act 2000) Norm ousting the courts from reviewing whether that duty is fulfilled
Law & Custom Law and custom of Parliament Norms of parliamentary privilege, telling courts not to ‘question’ what happens in Parliament (e.g. Article IX of the Bill of Rights 1689)
Directive Principles Principle imposing a duty to pursue a broad aim Norm telling courts not to enforce or act on the principle (Lael Weis has a good explanation here)
Disapplied Statutes Statutory duties (e.g. duties in MSA) Section 2(4) of the European Communities Act 1972

Constitutional conventions are different. Courts don’t enforce or apply conventions. But that’s not because of a legal norm which prohibits courts from enforcing convention. There isn’t any such norm. How do I know that? Well, Parliament hasn’t laid down a rule that courts shouldn’t enforce conventions. No statute says that. Courts have expressed their unwillingness to enforce conventions, but they don’t present this as a legal finding. Customary law obviously doesn’t contain such a norm either. And, as a final check, we can ask ourselves: would we say that courts act “unlawfully” by enforcing a convention? The answer is clearly “no”.

So here’s my proposal. For a norm to be a law a court would have to be willing to apply that norm, were the court presented with an appropriate case, and were it legally permitted to apply that norm. (This is a slight simplification. There are complicated questions about the legal status of rules of recognition, which aren’t relevant here.) Were a court presented with a case involving a statute, without an ouster clause, it would apply the rules it finds there. Were a court presented with a ‘law’ of Parliament, and were parliamentary privilege no bar, I take it the court would apply the law. Were a court to be presented with a case involving a directive principle, and were it permitted to enforce the principle, it would enforce it. Were the ECA no bar, the courts would apply the MSA. My counterfactuals are speculative, of course. But think about it pragmatically: why would we need Article IX, or an ouster clause, or a statement warning courts away from directive principles, or s 2(4) of the ECA unless courts might otherwise act on these norms?

If this is right, then conventions aren’t laws because courts aren’t willing to enforce them even though there’s no legal bar to them doing so. The revised argument would go like this:

  1. A norm is a law only if courts would apply it, in an appropriate case, were they permitted to apply it.
  2. Courts don’t apply conventions, even in appropriate cases, and even though they’re permitted to do so.
  3. So, convention aren’t laws.

Further questions

I said there were two ways to attack Dicey’s argument. The other way would be to try to show that courts do apply or enforce conventions. I think that line of attack would fail in the UK, as things stand now. But of course things could change. Courts in the UK might suddenly start enforcing conventions. Would that make conventions laws?  Courts elsewhere already enforce conventions. Are conventions in those places already laws?

Not necessarily. Courts sometimes enforce or apply norms which aren’t part of their  legal system. Examples include norms of logic or maths, contractual norms, foreign laws, and so on. All of these norms are acted on by courts. None are laws of their system. Something distinguishes laws from norms which courts are willing to act on; but what? It’s a core question in legal philosophy. But it’s a question that legal philosophers have neglected, a bit to my surprise. A topic for another blog post!


Parliament gives an official a power, and the official makes a decision you don’t like. So you go to a judge to have the decision reviewed. With any luck, the judge will agree with you and give you a remedy. That’s the normal way of things. But if Parliament is sovereign, then Parliament has the power to prevent the judge from reviewing the official’s decision. All Parliament has to do is include an “ouster clause”: a clause communicating its intention to make the official’s decisions unreviewable.

In the past Parliament has enacted what looks like an ouster clause. It did that most famously in Anisminic Ltd v Foreign Compensation Commission. But judges have tended to say that what looks like an ouster clause is, on closer inspection, not an ouster clause. It seems as if Parliament wants to oust the courts, but really it doesn’t intend that. Now, the clause in Anisminic was a bit ambiguous. It wasn’t as clear as one would wish. So, ever since Anisminic, it’s been a matter of speculation what judges would do when faced with a more clearly drafted clause.

With that as background, the interest in R (Privacy International) v Investigatory Powers Tribunal is understandable (DC | CA). It looks like exactly the sort of test case people were waiting for. There is a putative ouster clause, more clearly drafted than the one in Anisminic. What would judges do? Well, we know the Divisional Court and the Court of Appeal would do – they’ve both said that there is an ouster clause. As a result, they’ve found that certain decisions are immune from judicial review. But it’s likely that the case will be appealed to the Supreme Court, so we’re still waiting for the last word.

I want to show that Bayes’ Theorem provides a helpful way to think of ouster clauses. Bayes’ Theorem is much-discussed in evidence law, but rarely invoked in statutory interpretation. I think it suggests – I don’t put it any more strongly – that Privacy International may have been wrongly decided. (more…)

I’m an administrative law scholar, but I often suffer from private law envy. In private law, I see lots of doctrinally-oriented theory. I don’t mean theories of this or that doctrine. (There are those, too, of course.) I mean theories of whole areas of law: tort law, contract law, and the law of unjust enrichment. I mean promissory theories of contract law, corrective justice-based theories of tort law, and the like. I’m not saying these theories succeed. For all I know they don’t. It’s the ambition that impresses me, because in administrative law, things could hardly be more different. There are articles proposing theoretical accounts of specific doctrines: legitimate expectations, say, or error of law. There is the endless, exhausted ultra vires debate. And that’s about it, theory wise. I’m not sure why this is. Maybe it’s that administrative law is a relatively young area. Maybe public law scholars are preoccupied with devolution, human rights, and now Brexit. In any case, there’s a gap, and I want to help fill it.

In a new article I outline a theory of the most significant part of administrative law – the law of judicial review. Here’s the abstract:

There is no general theory of the grounds of judicial review (e.g., the rule against bias, the doctrine of legitimate expectations, unreasonableness). Here I try to fill the gap. My theory draws on ideas from the philosophy of law and the philosophy of action, but it’s a simple theory. Officials make decisions for the community. Their decisions are subject to requirements of instrumental rationality. Ideally, officials would figure out for themselves how to live up to these requirements. Because officials aren’t perfect, the law also has “Plan B”, which is for judges to ensure that officials do what rationality requires of them. The grounds of judicial review are simply the grounds on which it’s rational for officials to reconsider, retain, suspend, or apply their decisions. The “Plan B theory”, as I call it, doesn’t account for every detail of judicial review. Even so, the theory is powerful, parsimonious, fruitful, and sharply at odds with almost everything else written about judicial review.

You can download the paper here.

Comments and suggestions, no matter how small, are very welcome!

I have a new draft article about constitutional conventions, co-authored with Farrah Ahmed (Melbourne Law School) and Richard Albert (Boston College Law School). The article is about how courts can, do, and should engage with constitutional conventions. Here’s the abstract:

The study of constitutional conventions is anchored in an assumption that has so far remained unchallenged: Commonwealth courts will recognize and employ conventions but never enforce them. We show in this Article that the dominant view today is doubly mistaken: there is no such shared “Commonwealth approach” to the treatment of constitutional conventions nor do Commonwealth courts refrain from enforcing conventions. Drawing from Canada, India and the United Kingdom, we disrupt the foundations of the scholarly understanding of conventions by demonstrating that Commonwealth courts have recognized, employed and indeed also enforced conventions. Beyond this new discovery, we make the normative claim that Commonwealth courts sometimes should enforce conventions, an additional contrast between the dominant view and ours. We argue that courts should act as executors of the will and judgment of constitutional actors, and limit themselves to enforcing only power-shifting conventions, which transfer power from those who have legal power to those who can legitimately wield it. This new role of an executor court brings clarity, stability and predictability to the exercise of official powers that are rooted in constitutional convention rather than constitutional law.

Download the SSRN version of the article here.

If there were ever a prize for “least examined ground of judicial review”, I would nominate the flexibility rule. The flexibility rule says that administrative policies must be flexible not rigid. The rule is nearly a century old. It’s part of the law of judicial review in England and Wales, Canada, Australia, New Zealand, and many other common law jurisdictions. It’s invoked in case after case. And yet the rule has received hardly any serious scrutiny. Judges rarely question it. Academics rarely write about it.

This neglect extends to the rule’s justification. Take the case of British Oxygen Co v Minister of Technology. The Board of Trade had a discretionary power to make grants for equipment purchases. To manage all the applications it received, the Board made a policy not to make grants for purchases of less than £25. British Oxygen had purchased a lot of oxygen cylinders for £20. The company didn’t quite fit under the policy, but it applied for a grant anyway. The Board refused, and the company went to court, arguing that the Board hadn’t taken into account the particulars of its case. The House of Lords ultimately sided with the Board. But it did so because it believed that the Board had ‘carefully considered’ the company’s application. Had the Board’s policy been rigid, British Oxygen would have won.

British Oxygen is a famous case. It’s probably the leading case on the flexibility rule. Yet the House of Lords never explains the rationale for the rule. So I wonder: what’s so great about flexibility? What’s so terrible about rigidity? What, if anything, is to be said for the flexibility rule? (more…)

I’ve been working on a series of papers about mercy. My latest is a discussion of when the government acts mercifully, if indeed it ever does.

Here’s the abstract:

A pardon is an act of mercy according to the law, but is a pardon mercy in an ordinary or genuine sense? What distinguishes a pardon from a lenient judicial sentence, which is not mercy by the law’s lights? These are questions about what mercy as it is understood in law has to do with mercy as it is understood outside of law, and about who in government acts mercifully and when, if indeed anyone in government ever does. Here I propose a general analysis of mercy, then bring that analysis to bear on government action. Three features of my analysis are noteworthy. First, almost all existing analyses say that mercy is unconstrained in a normative sense, but I argue that mercy is unconstrained in the way that arbitrary power is unconstrained. Second, although it’s often assumed that mercy must be motivated by compassion, I show that mercy only requires acting with the intention to benefit the recipient. Third, my analysis says that mercy requires the giver of mercy to overcome a motivation to treat the recipient harshly. Given this analysis, few government acts are merciful, but pardon is an institutional approximation or analog of mercy.

Read the full paper here.

The paper is a draft, and comments and suggestions, no matter how small, are very welcome!