When should the UK government hold a border poll in Northern Ireland?

The question is essentially a political one, but there are some firm legal limits

February 08, 2024
From left to right: Northern Ireland secretary Chris Heaton-Harris, first minister Michelle O’Neill, deputy first minister Emma Little-Pengelly and Rishi Sunak at Stormont Castle, Belfast, following the restoration of the powersharing executive. Image: PA Images / Alamy
From left to right: Northern Ireland secretary Chris Heaton-Harris, first minister Michelle O’Neill, deputy first minister Emma Little-Pengelly and Rishi Sunak at Stormont Castle, Belfast, following the restoration of the powersharing executive. Image: PA Images / Alamy

What is the legal position relating to a referendum in Northern Ireland on Irish unification? In particular, can the United Kingdom be obliged as a matter of law to hold a referendum, regardless of any wishes the UK government has to maintain the union?

The starting point for answering these questions is the very first sub-section of the Northern Ireland Act 1998, the statute that implemented the Good Friday Agreement of the same year. Here there is a declaration twinned with a qualification: 

“It is hereby declared that Northern Ireland in its entirety remains part of the United Kingdom and shall not cease to be so without the consent of a majority of the people of Northern Ireland voting in a poll held for the purposes of this section in accordance with Schedule 1.”

The legislation thereby expressly envisages what is sometimes called a border poll, and that unless and until there is such a poll Northern Ireland remains part of the United Kingdom. A referendum in favour of unification in Northern Ireland, together with a corresponding vote in Ireland, would lead to both the British and Irish governments introducing bills to their respective legislatures for unification.

This sub-section is similar to the first sections of previous UK legislation on Northern Ireland. The 1949 Ireland Act, for example, provided that it was “affirmed that in no event will Northern Ireland or any part thereof cease to be part of His Majesty's dominions and of the United Kingdom without the consent of the Parliament of Northern Ireland.”

And the successor legislation of 1973 provided that “in no event will Northern Ireland or any part of it cease to be part of Her Majesty’s dominions and of the United Kingdom without the consent of the majority of the people of Northern Ireland voting in a poll held for the purposes of this section in accordance with Schedule 1 to this Act.”

That UK legislation provided a requirement for a vote, either of the then Northern Irish parliament or, after the parliament came to an end, a direct poll, was therefore not an innovation of the Good Friday Agreement. The emphasis of the wording perhaps changed, but the notion of a border poll did not. And indeed, there was such a vote in 1973, though it was largely boycotted by the nationalist community who (rightly) saw it as a device to maintain the union and placate the unionists.

The innovation introduced by the 1998 agreement was the conditions under which such a referendum may be introduced. Here, tucked away in the first schedule to the 1998 Act, are two important things. First there is a general power of the UK government to hold a referendum:

“(1) The Secretary of State may by order direct the holding of a poll for the purposes of section 1 on a date specified in the order.”

This provision confers a bare, seemingly unqualified discretionary power. However, in invisible ink, the law implies that the power must be exercised in good faith and the public interest, and—as we will see—implies an obligation to do so on an even-handed non-discriminatory basis, giving effect to the intentions of the Good Friday Agreement.  

But it is the second thing that is more significant. For set out next is this significant wording, which should be read carefully:

“(2) Subject to paragraph 3, the Secretary of State shall exercise the power under paragraph 1 if at any time it appears likely to him that a majority of those voting would express a wish that Northern Ireland should cease to be part of the United Kingdom and form part of a united Ireland.  

(3) The Secretary of State shall not make an order under paragraph 1 earlier than seven years after the holding of a previous poll under this Schedule.”  

The “shall” in (2) is immense, for it means that this passage imposes a duty, rather than conferring a mere discretion. If the condition expressly mentioned is met, then the UK government has no choice: there has to be a border poll in Northern Ireland. 

These paragraphs are no accident: they are explicitly provided for in the Good Friday Agreement and the UK undertook to place the wording on the statute book.

The condition is framed in wide—very wide—terms. The UK government, through the secretary of state for Northern Ireland, has to hold a border poll when “at any time it appears likely to him” that the majority would vote for unification. The breadth of this wording means that the courts will be slow to gainsay the judgement of the secretary of state. The provision also does not stipulate what evidence or considerations should be used by the minister in forming a view of how the majority would vote.

But while the provision may be broad, it is not unlimited. There are certain extreme circumstances where the courts—however reluctantly—would step in. One parameter would be if the UK government invoked this power in bad faith, calling a border poll (arguably like in 1973) with the desire and expectation that unification would be defeated and a further such referendum could not happen for a further seven years.

At the other parameter there are conceivable circumstances where it would become unreasonable for the UK government to ignore the accumulation of evidence that a majority of those in the Northern Ireland would vote for unification. Say, for example, that elections both to the Northern Irish Assembly and the Westminster Parliament showed overall majority support over a number of contests for avowedly nationalist parties, both in terms of votes cast and seats won.   

At a certain level it would become what lawyers call Wednesbury unreasonable for the UK government not to call a referendum: so unreasonable that no reasonable government could maintain that position.

In such a circumstance, however, it is not clear what the courts would actually do. Judges are reluctant to interfere in political matters generally and Northern Irish politics in particular. There is also the general truth that courts in “public law” matters are more comfortable quashing erroneous decisions than mandating public bodies to do things. A mandatory order to hold a referendum would be unlikely.

What a court would probably do in this extreme situation is simply make a declaration that the condition for a border poll has been met, and that judicial remedy would probably be sufficient.   

But even that level of intervention is a remote possibility. The reluctance of the courts to intervene can be shown in a couple of judgments brought about by the spirited litigation of the victims’ campaigner Raymond McCord. While not binding on the UK government or courts on when a referendum should be held, these two judgments indicate the likely judicial attitudes to intervening in the event the UK government resists calling a border poll.

In 2017, McCord brought a case contending that the UK government should set out how it would approach making a decision to hold a referendum. In particular, McCord argued that the UK government should publish and then abide by a policy on what information would be considered, who would be able to vote, what question would be asked, and so on.

This was not an unreasonable request, and there are precedents where a government holding a wide and consequential power will set out when that power would and would not be used, and how it would be used.

Yet the courts said no. It was not for them to compel the government to publish such a policy. In two judgments that should be read by anyone interested in how the courts will treat the legal aspects of the next unification referendum, the judges stated that the issues are intensely political, and so should be resolved by politicians and not judges.

At first instance, the judge Sir Paul Girvan stated:

“any decision as to the holding of a border poll involves extremely complex political considerations and if not carefully handled taking account of prevailing circumstances it could give rise to great instability… The precise circumstances and the political context of a decision are variable and highly political. Decision-making in this area requires a political assessment on the part of the Secretary of State and in this context political flexibility and judgment are called for… the Secretary of State must determine what she considers to be the relevant considerations to be taken into account or left out of account in deciding the political question whether the calling of a border poll would in the circumstances be appropriate.”

And on appeal, three judges unanimously stated:

“We consider that it is essentially a political judgment. It is assigned to and is to be performed by the respondent, a politician who is to form an assessment as to the political views of others. The political judgment as to the likely outcome of a border poll is not a simple empirical judgment driven solely by opinion poll evidence. It is also not a simple judgment based purely on perceived religion. 

“The judgment depends on what are the prevailing circumstances at any given time. For instance a likely outcome may involve an evaluation as to whether there are other factors which will impact on voting intentions crossing traditional party or perceived religious lines and if so as to their impact.  

“Instances of such factors are changes in social attitudes North and South, relative economic prosperity North and South, the taxation structures North and South, the outcome of Brexit and the nature of future trading relations between both parts of Ireland which in turn depends on any agreement between the United Kingdom and the European Union.”

McCord’s application was refused: the UK government was under no legal obligation to publish a policy on how it would approach the referendum question, let alone abide by any such policy.

Reading the two judgments one gets the impression that the judges really want this to be an entirely political matter, with no involvement of the courts at all.

But the judges do not get to make that choice. Whether the courts like it or not, the Northern Ireland Act 1998 imposes a legal obligation if a certain condition is fulfilled, and that obligation was spelled out expressly in the Good Friday Agreement. The parties to that agreement intended this to be—at least ultimately—a legal matter and not just something left to politics. That was part of the comfort that the nationalists were seeking when they agreed to the 1998 accord. 

So the judges cannot duck out of this entirely, as much as they would prefer to do so. And reading the two judgments carefully you will see that the judges know and accept that there will be circumstances where the decision to hold or not hold a referendum could become a matter of law.

The UK government has to approach the question honestly, in good faith and with even-handedness. The courts were expressly careful to also say the secretary of state’s discretion is subject to Wednesbury limits (though Girvan’s first instance judgment does so rather subtly).

What all this means is that if, year-on-year and election-after-election, it becomes undeniable that the hard voting figures show that “a majority of those voting [in a border poll] would express a wish that Northern Ireland should cease to be part of the United Kingdom and form part of a united Ireland”, then eventually there will come a stage where an ongoing UK government refusal can only be Wednesbury unreasonable and/or in bad faith. In these circumstances the courts would eventually have to intervene.

Section 1 of the Northern Ireland Act therefore means that time is on the side of the nationalists in Northern Ireland—if their current electoral success continues. Although the courts will be deferent to the political decision-making process—and will certainly not grant any remedy easily (or happily)—that deference is not and cannot be inexhaustible.  

This balance of politics and law is what the parties to the Good Friday Agreement intended. And, by enacting section 1 and its accompanying schedule to the Northern Ireland Act, it is what the UK parliament intended too.