Referendums in UK democracy: how should they work in practice?

The Independent Commission on Referendums, established by the Constitution Unit to review the role of referendums in UK democracy, has now met twice. One of the issues they are considering is rules for how referendums should work in practice. The Commission’s Research Assistant, Jess Sargeant, summarises the issues for consideration.

In a previous blog post I explored some principles that could be used for deciding when a referendum might be appropriate. The Independent Commission on Referendums is also considering how referendums should work in practice. The following post explores some key practical questions.

Should there be rules for when a referendum is required, permitted or prohibited?

The UK is unusual among comparable democracies in that referendums are held ad hoc: there are very few standing rules on when referendums are to be held. This means, at least in theory, that there are no restrictions on matters that a referendum may be held on: it could be held on any issue within parliament’s legislative competence.

Many other democracies have provisions in their constitutions setting out when a referendum must be, can be, or cannot be held. Constitutional issues are the most common category of issues on which a referendum is required. For example, Ireland, Australia and Japan require referendums on any bills amending the constitution. In Austria, Spain, Lithuania and Iceland amendments to certain key parts of the constitution must be approved in a popular vote. There are also examples of referendums being required on other issues: Denmark has mandatory referendums on transfers of sovereignty and changes to the voting age.

Where referendums are not required on constitutional amendments, there is often a mechanism allowing a parliamentary minority to trigger one, as is the case in Italy, Austria and Spain. In some democracies, legislation can be put to a referendum if requested by a body so empowered by the constitution. This could be the parliament, as in Denmark and Austria, the president, as in Ireland and Iceland, or groups of citizens, as in Italy and the Netherlands. Where referendums are permitted on legislation, certain types of legislation are often exempt: most commonly, finance, budgetary and tax laws or legislation implementing treaties.

One question for the Commission to consider is whether the UK should seek to adopt such rules. These could either be institutionalised, in a Referendums Act or other legislation, or simply be expressed as political conventions, or recommendations of best practice. It is worth noting that the idea of legislation in this area is not completely alien to the UK: requirements to hold referendums are codified in the Northern Ireland Act 1998, the European Union Act 2011, the Localism Act 2011, the Scotland Act 2016 and the Wales Act 2017.

The UK differs from most democracies in that it does not have a codified constitution, and this presents further complications. Should it be decided that referendums should be held on constitutional issues, as is the recommendation of the Constitution Committee, the Public Administration and Constitutional Affairs Committee and Constitution Society, the difficulty becomes defining what should be considered a ‘constitutional issue’.

Where should referendums sit in the policy-making process?

The UK’s constitutional arrangements make it impossible to hold legally binding referendums: the principle of parliamentary sovereignty means that parliament cannot be bound in law. However, referendums can be de facto binding if the enabling legislation for a referendum not only sets out provisions regarding the referendum itself, but also contains the new rules that will come into being if the referendum is passed. Referendums of this kind are known as ‘post-legislative’ referendums, and have the advantage of providing clarity on the options being put to voters. Further, this allows the substance of the proposed change to be subjected to the same parliamentary scrutiny as any other legislation. Holding a referendum post-legislation, however, does not necessarily mean that the proposals have genuine parliamentary support. Parliament could pass the legislation on the basis that there should be a referendum on the proposals rather than because it is in favour of them, as was the case with the AV referendum. It might even approve the proposals in the hope (or expectation) that the referendum will fail to pass them.

In some circumstances, a referendum may be pre-legislative: voters are asked to vote on a principle rather than a concrete proposal. The 1997 devolution referendums, the 2014 Scottish independence referendum and the 2016 EU referendum were all pre-legislative. The disadvantage of pre-legislative referendums is that the details of proposals are not known and voters may be unclear exactly what they are voting for. Nonetheless, when it is necessary for the government to obtain the consent of the population before proceeding with a particular aim – such as beginning negotiations – it is difficult to see how a referendum could be held only post-legislation.

Steps can nevertheless be taken to minimise uncertainty and provide clarity: for example, by requiring a white paper on government plans. Another possibility could be holding two referendums: one pre-legislative, to mandate the government to proceed with an aim, and another post-legislative, to obtain consent for the final proposal – though such arrangements could have a number of problematic political consequences. Consideration might also be given to the possibility of holding referendums after a deliberative process, where a group of citizens might determine the topic of the referendum and/or the options available. This approach has been practised in Ireland and was responsible for the 2015 referendums on marriage equality and the age threshold for the presidency. It has also led to the current proposal for a referendum on the liberalisation of abortion rules.

Should there be additional constitutional safeguards?

It is often argued that referendums should require additional safeguards to prevent major changes being made on the basis of narrow or temporary majorities. One way to do this would be to require a change to be approved by different actors in the political system at multiple decision points – as is achieved in various other democracies through constitutional amendment procedures. For example, in Ireland a bill amending the constitution must pass through the Irish parliament, be approved by a majority in a referendum, and then be signed by the President before it is made into law. Such an approach can be criticised on the basis that overly onerous procedures can make constitutional modernisation difficult; on the other hand, it can be argued that big decisions such as constitutional change should be difficult, and require wide and considered consent.

Another safeguard often proposed is to vary or supplement the default 50% threshold necessary for a referendum to be valid. Threshold requirements that take turnout into account are often justified on the premise that sufficient participation should be required for a referendum to be considered a legitimate expression of public will. One such threshold type requires that a certain percentage of the electorate – often 50% – turn out to vote in order for the result to be valid. Another type is an electorate threshold, where a specified percentage of the total electorate (including non-voters) is required to support a change before it passes. In Denmark, for example, 30% of the electorate must vote to reject a bill that has passed through the legislature in order to block that bill. However, thresholds of these kind have been subject to a number of criticisms: they can have paradoxical effects, incentivising campaigners against change to demobilise voters; and inaccuracies in the electoral roll could potentially change the referendum result.

Other types of threshold take different approaches. In the case of supermajority thresholds, the basic threshold is varied so that more than half of votes cast is required for the referendum to pass. Internationally these are rare, but 60% supermajority requirements have been applied in referendums on electoral reform in some Canadian provinces. A different kind of additional threshold requires multiple majorities in different areas. The SNP proposed that this should apply to the UK’s devolved nations prior to the EU referendum. A similar requirement is in place for Australian constitutional referendums.

One final constitutional safeguard could be to hold two referendums on the same question, with a ‘cooling off period’ in between. This could ensure that preferences are stable and also have the benefit of providing additional time for information and deliberation.

Next steps

It is too early to say what the Commission will make of these various issues. They will be discussed in the coming months, and we also welcome the views of anyone else who thinks they have insights that the Commission’s members should attend to.

This post draws on background research prepared for the Commission and does not represent the views of the Commission or any of its members. Should any reader feel they have expertise that would assist in examining the questions raised above, they are welcome to contact Jess Sargeant at

About the author

Jess Sargeant is Research Assistant for the Independent Commission on Referendums, based at the Constitution Unit.

2 thoughts on “Referendums in UK democracy: how should they work in practice?

  1. Pingback: I·CONnect – What’s New in Public Law

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s