Faculty of law blogs / UNIVERSITY OF OXFORD

A New Referendum is a Constitutional Requirement


Time to read

7 Minutes

The most recent primary legislation on the relations between the European Union and the United Kingdom is the European Union Act 2011. This Act, who some scholars called at the time an ‘unprecedented constitutional experiment’ is known for the fact that it establishes a very unusual ‘referendum lock’ before an amendment of the EU Treaties can be ratified by the UK. It introduces a rule that all serious amendments of the EU treaties will have to be approved both by an Act of Parliament and by the electorate in a referendum. The intention behind that Act was to make sure that the UK had thought long and hard before participating in further EU integration. The Act was seen as controversial among lawyers, because it changed the ‘manner and form’ in which the sovereignty of parliament is exercised.    

In my view, the Act adds an important dimension to the debate about the effects of the EU referendum of June 23. The 2011 Act has been written in such an expansive way as to encompass, in my view, not only treaties that amend the EU treaties, but also treaties that the UK is due to enter as a result of withdrawing from it. This may be a surprising suggestion, because the drafters of the 2011 Act probably had not thought about this prospect. Nevertheless, the Act is to be applied on the basis of what it says, not on the basis of what its drafters were thinking at the time. On the basis of the words on the page, a new referendum is most likely legally necessary before the UK withdraws from the EU.

The argument is simple. The 2011 Act provides at section 2(1) that it applies to a ‘treaty which amends or replaces’ the EU treaties. Treaties that ‘amend’ the EU treaties affect, of course, only members of the EU. The treaty between the EU and the withdrawing UK will not amend the main treaties, namely the TEU and the TFEU (a separate treaty among the remaining 27 will have to do that). But, as we saw, the Act, in its attempt to include perhaps everything about the EU, covers also those treaties that ‘replace’ the EU treaties. What does ‘replace’ mean? ‘Replacing’ is something different from ‘amending’. Because it is different, it must be something different from the ordinary amendment of the current EU treaties. It is therefore obvious that a new treaty between the withdrawing UK and the remaining EU could be one ‘replacing’ the European treaties, at least as far as the UK is concerned.  It is a treaty that replaces the UK’s rights and obligations towards the EU.

This conclusion is strengthened by the fact that the Act defines (in section 1(4)) part of what it means by ‘amends’. There is no definition at all, however, of what is meant by ‘replaces’, which is left entirely open. It is therefore clearly plausible that both a withdrawal agreement that the UK will agree with the EU under Article 50 (the ‘withdrawal agreement’) and an in principle separate (and hopefully contemporaneous) future trade relations agreement (‘the trade agreement’) will be treaties that ‘replace’ the EU treaties as far as the UK is concerned. This is because all the rights and duties of the UK towards the EU would be replaced by the rights and duties created by the withdrawal agreement and the trade agreement. If this is the case, then both these agreements will fall under the scope of the 2011 Act.

Not all agreements require a referendum according to the 2011 Act. Section 4 of the Act outlines a list of cases where a new treaty that in principle falls under the scope of the Act ‘attracts a referendum’.  If any one of the conditions mentioned there is met, a referendum is required. These conditions are very broad, perhaps excessively so. It is likely that those who drafted them did not have in mind a treaty that is part of a process of withdrawal from the EU. Most likely, they had in mind its opposite, a treaty bringing further integration. But the words mean what they say. Any court will have to give effect to them as they are in the statute book.

This was clearly the intention of the government at the time. Introducing the Second Reading of the Bill for this Act in the House of Lords Lord Howell of Guildford (who was then Minister of State, Foreign and Commonwealth Office) said about section 4: ‘The Government make no apology for the complexity of the provisions. We want to make it clear for Parliament, the British people and, indeed, our EU partners and the EU as a whole where a referendum would be required under the Bill. We feel that a short, vague statement would leave any future decisions more open to challenge in Parliament and the courts’. So every one of these detailed provisions was carefully thought out.

The most obviously relevant cases seem to me to be in paragraphs 4(1)(i) and 4(1)(j), although others may also prove relevant:

(i) the conferring on an EU institution or body of power to impose a requirement or obligation on the United Kingdom, or the removal of any limitation on any such power of an institution or body;

(j) the conferring on an EU institution or body of new or extended power to impose sanctions on the United Kingdom.

The conditions do not refer to the balance of power or any arrangement of reciprocity. They are triggered if an EU institution or body is given the power to impose a single ‘requirement’ or ‘obligation’ on the UK, or if any limitation on existing powers is removed. They are also met if an EU institution or body is given the power to impose ‘sanctions’ on the UK.

Whether these tests of section 4 will be met by a future UK/EU treaty on withdrawal and future access to the single market depends on what those treaties will say.  But it is in my view highly likely that at least one of these very broad tests will be met by any treaty creating reciprocal obligations in the process of withdrawing from the EU and establishing a future trading deal. For example, a withdrawal agreement will certainly impose obligations on the UK regarding EU citizens currently in the UK. This is especially so if the UK decides to replace its membership in the EU with membership of the EEA. As is well understood by now, being a member of the EEA involves accepting the rules of the single market without voting for them at the Council. So the test of section 4(i) will be met in both its limbs, because by virtue of the EEA agreement EU institutions will have been conferred a new power to impose requirements on the UK, while the UK will have lost both its power to vote in the Council and its right to bring an action to the Court of Justice against an institution of the EU regarding these obligations.

For different reasons, the tests of section 4 (i) and (j) may also be met by a bilateral treaty or set of bilateral treaties between the UK and the EU (the ‘Swiss’ model).  First, among other things, any such agreements will most likely remove the protection that the UK currently enjoys not to be discriminated against by virtue of 18 TFEU. The agreement will thus remove a ‘limitation’ on the powers of EU institutions. Second, any new treaties will almost certainly create powers in bodies or institutions of the EU to impose a requirement or obligation on the UK (paragraph i) or even create some form of sanctions (paragraph j) as in the WTO model of judicial panels, or by way of arbitration.

They will do so, most likely, in international and not in EU law, but the Act does not draw such a distinction. It says, for example, that what matters is the conferring of ‘power’ not ‘power in EU law’. And if this applies to the set of bilateral treaties, it is also likely to apply to the WTO option as well.

Of course I am only summarising here issues that will require full and detailed consideration, if and when these treaties take shape. But the literal reading of the 2011 Act suggests that the tests are wide-ranging and therefore the threshold for meeting them is low. And, if any one of the tests of s.4 is met, a new referendum will be required by law. In short, because the scope of the 2011 Act is so wide and because it requires a referendum practically whenever there is any change in the allocation of powers in the relation between the EU and the UK, by changing these powers in the process of withdrawal, a new treaty or treaties between the withdrawing UK and the remaining EU will almost certainly meet some of the tests of the 2011 Act. This was not perhaps the intention of the drafters, but this will not matter (under the well-established Pepper v Hart criteria) where the meaning of a provision is clear.

If this argument – or a version of it - is correct, then it follows that practically any new agreement with the EU as a result of the EU referendum will have to be approved by a new referendum. Of course, the current government – if it could secure a majority in Parliament – could seek to enact new legislation to abolish the 2011 Act (or perhaps try to ignore it, relying on a rival account of parliamentary sovereignty). This is a feature of the omnipotence of our Parliament. But the retrospective change of the law will be seen as obviously self-serving, and as something offensive to the rule of law. The current government had a chance to address these issues when it secured the enactment of the European Union Referendum Act 2015. The terms of the referendum were set then by the current parliamentary majority. To retrospectively change these terms after the referendum has taken place in order to circumvent the legal obligation of a second referendum, will be rightly seen as constitutionally inappropriate. When the referendum was called, the obligation to hold a second referendum was already in the statute book.

I conclude that a proposed new treaty between the UK and the EU that seeks to replace EU membership with a set of new trade agreements will most likely have to be approved by way of a new referendum according to the framework created by the 2011 Act. The only secure way to avoid such a referendum is not to enter into any new treaty with the EU at all. This would entail that the UK would leave unilaterally, without regard for future trade and in very bad terms. This, however, is an unthinkable scenario, one to bring certain economic disaster for the UK and one that has not been contemplated by the leave campaign or the government.

So what would a new referendum decide?  The question would be about the proposed new treaty or set of treaties with the EU, which would have been agreed after a lengthy and probably laborious process as the best possible deal available to the UK. If the answer to this referendum is affirmative to the new treaty, then the new agreement replacing membership of the EU with another type of relationship will have been approved and will take effect. But if the answer is ‘no’ to the new agreement, then the status quo, namely continuing membership of the EU, will have received a renewed mandate by the electorate as a preferable state of affairs to the best available alternative. The electorate would have chosen to remain. At that point the government would have to revoke the Article 50 notification and the UK would continue being a member of the EU.

Pavlos Eleftheriadis is a Fellow in Law at Mansfield College, Oxford, and a barrister at Francis Taylor Building, London.


With the support of