Referendums in the United Kingdom - Constitution Committee Contents

Memorandum by Professor Vernon Bogdanor, Professor of Government, Brasenose College, University of Oxford


  Almost all democracies employ the referendum. Amongst countries which have been continuously democratic since the 1940s, only Germany, India, Israel, Japan and the United States have not used referendums at national level.[2]

  But the referendum is not addictive. Switzerland, which holds on average around one national referendum a year, is very much the exception to the general rule. Indeed, Switzerland has held around half of all of the national referendums that have ever occurred.[3] Australia and Italy are the only other democracies to have used referendums at national level at all frequently. No other democracy has held more than 45 nationwide referendums. The typical democracy, like Britain, holds referendums but very infrequently.

  The dichotomy between "representative" and "direct" democracy is, therefore, highly misleading. For the referendum, even in Switzerland, is used not to replace, but to supplement representative democracy. There is little danger that it will come to subvert parliamentary government.

  Referendums are used primarily to resolve constitutional issues. For fundamental changes, so it is argued, ought not to be implemented without the consent of the people. They should not be implemented simply at the will of the government of the day. If they are to secure legitimacy, they need the endorsement of the people as well as that of the legislature.

  Britain, of course, lacks a codified constitution. This means that there are no legal rules requiring a government to hold a referendum on particular issues. And no definition of what is to count as a "constitutional" issue. The referendums that have been held or promised have been decided upon by the government of the day at its discretion. An elastic constitution, so it seems, implies an elastic use of the referendum. But this gives rise to a problem. For the referendum, in countries with a codified constitution, is intended to constrain the government of the day. In Britain, by contrast, if use of the referendum lies at the discretion of government, it can be used to augment the power of government rather than limiting it, by allowing a government to bring the people into play against Parliament. That was perhaps the case with the devolution referendums. The referendum could then become a tactical device, "the Pontius Pilate" of British politics.[4]

  In fact, however, conventions have grown up as to when the referendum ought to be used. These conventions, though in no sense legally binding, may serve to act as precedents constraining future governments.

  The examples of the devolution referendums in 1979 and 1997 in the non-English parts of the United Kingdom, together with the referendum on regional devolution in the north-east in 2004, would seem to imply that a referendum needs to be held before there is any significant devolution of powers away from Westminster.

  The 1973 border poll in Northern Ireland, together with the commitment, first made in the Northern Ireland Constitution Act of 1973, and reiterated, most recently, in the 1998 Belfast Agreement, would seem to indicate that a referendum should be held in the area concerned before any part of the United Kingdom is allowed to secede. It seems generally agreed that, even were the SNP to win a majority of Scottish constituencies in either Westminster or Holyrood, a referendum would be held before independence would be conceded.

  The examples of the European Community referendum of 1975 and the referendum in London in 1998, seeking approval for the first directly elected mayor in British history, together with the promises to hold a referendum before joining the eurozone or changing the electoral system for elections to the House of Commons, would seem to show that a referendum is required when a wholly novel constitutional arrangement is proposed.

  The referendum, then, is used not on bills which propose changes, however radical, in the laws, but for legislative proposals which provide for a radical alteration in the machinery by which the laws are made. The rationale for this requirement lies deep in liberal thought and was well stated by John Locke in his Second Treatise of Government, para. 141. "The Legislative cannot transfer the power of making laws to any other hands. For it being but a delegated power from the People, they who have it cannot pass it to others". Voters entrust their power to representatives, but they give them no authority to transfer those powers, to make radical alterations in the machinery by which laws are made. Such authority can be obtained only through a specific mandate, that is a referendum.

  Locke's doctrine would seem to imply that a referendum is required, not only when power is transferred downwards, as with devolution, but also when it is transferred upwards to the European Union. However, there has not been a referendum on any of the five amending treaties to the Treaty of Rome—the Single European Act of 1986, which involved a very wide transfer of powers, the Maastricht treaty of 1992, the Amsterdam treaty of 1997, the Nice treaty of 2000, nor the Lisbon treaty of 2007. The promise of a referendum on the now defunct European constitution but not on the Lisbon treaty could perhaps be defended on the grounds that the former was a wholly new constitution for the European Union, while the Lisbon treaty was a mere amending treaty like the Single European Act and succeeding amending treaties. Significantly, while nine member states, including Britain, either held referendums or were proposing to do so on the constitution, only Ireland held one on the Lisbon treaty, and that because she was required by her constitution to do so.

  Nevertheless, by analogy with the referendums on devolution, there does seem a strong case in logic for arguing that there should be a referendum before major legislative powers are transferred upwards to the EU as well as downwards to devolved bodies. Were that doctrine to be accepted, it would have made for referendums on the Single European Act and Maastricht, but perhaps not on the Amsterdam, Nice or Lisbon treaties, none of which involved major transfers of powers—moreover, the Lisbon treaty provided for opt-outs for the United Kingdom for many, though not all, of the transfers. But there is of course much room for argument as to which transfers of power are "major" and which are not.

  There are, then, persuasive precedents and these may in the future come to constrain governments. But, even if they do, national referendums are likely to be held only very infrequently.

  Might it be possible to go further in making the referendum a weapon of retrenchment, as early advocates of it such as Dicey hoped that it would be? One possible way of entrenching legislation might be to suggest that any future amendment or repeal of a particular statute e.g. the Scotland Act, should require a referendum. It would be natural to apply such a provision to legislation of fundamental constitutional importance such as devolution or the Human Rights Act. It might, for example, have been provided that the devolved bodies in Scotland, Wales and Northern Ireland, which were established after referendums, could not be repealed without a referendum. Upon one interpretation of parliamentary sovereignty, this could not be done, since Parliament could simply ignore the referendum requirement and abolish the devolved bodies without any recourse to the people. The decision of one Parliament cannot, it might be argued, bind a later Parliament. Nothing can prevent later legislation from repealing earlier legislation. But it could be argued that the referendum requirement could be made a condition of a bill purporting to abolish a devolved body receiving the Royal Assent. The referendum requirement would then redefine what was to count as valid legislation on a particular topic. The Parliament Acts of 1911 and 1949 redefined what was to count as valid legislation, by providing that a money bill could be passed without the consent of the House of Lords, and that a non-money bill could be passed without the consent of the House of Lords, provided that the same bill had been passed by the House of Commons in two successive sessions. From this perspective, the referendum requirement would be doing nothing more than laying down a further rule for what was to count as valid legislation. There seems no reason in principle why such a requirement should not be possible.


  In countries with codified constitutions, the outcome of a referendum generally binds both parliament and government. In Britain, however, with an uncodified constitution, the position is much less clear. For, although neither Parliament nor government can be legally bound by a referendum result, the government could agree in advance that it would respect the result, while a clear majority on a reasonably high turnout would leave Parliament with little option in practice other than to endorse the decision of the people. Shortly before the European Community referendum in 1975, Edward Short, the Leader of the House of Commons insisted to the House that "This referendum is wholly consistent with parliamentary sovereignty. The Government will be bound by its result, but Parliament, of course, cannot be bound". He then added, "Although one would not expect honourable members to go against the wishes of the people, they will remain free to do so".[5] Presumably Short meant that the government would be morally bound. It seemed then that it could not be legally bound. For it seemed then as if the British constitution knew nothing of the people.

  It is worth asking, however, whether a referendum could be mandatory rather than advisory, or whether the people must be held to have irretrievably delegated the authority to legislate to their elected representatives in the House of Commons. In 1653, Oliver Cromwell's Instrument of Government declared that legislative power resided in the person of the Lord Protector "and the people". Such a doctrine, however, does not seem to have survived the fall of Cromwell. Yet there seems no reason in principle, despite the doctrine of the sovereignty of Parliament, why a referendum result should not be mandatory in the sense that legislation passed by Parliament would automatically come into effect if there were a vote in favour, and automatically be rejected if there were a vote against.

  There is a precedent, perhaps not of very great significance, seeming to show that a mandatory referendum is not incompatible with the British constitution. When the Callaghan government, in February 1977, produced its referendum amendment to the Scotland and Wales bill, the New Clause 40 originally provided for a mandatory referendum. The clause originally declared that "If the decisions in the referendum are that no effect is to be given to the provisions of this Act, this Act ... shall not take effect".[6] Were the referendum outcome to be favourable, the government would have been under a legal duty to bring forward a commencement order so that the devolved bodies could be established, although Parliament would still enjoy the purely theoretical right to reject the commencement order. The government, however, changed its view during the course of the debate, and decided that the referendum should be advisory and not mandatory. It seems, nevertheless, that it might be perfectly possible to frame a referendum provision by which legislation was required to come into effect with a "Yes" vote, and required to be repealed with a "No" vote, in other words, a mandatory referendum. Whether it is desirable to provide for a mandatory referendum is, however, another matter. The next section provides arguments to show that it would not.


  A threshold can be in the form either of a minimum turnout level or a minimum percentage of the registered electorate. There are strong arguments against thresholds. It is difficult to be precise on what constitutes a sufficient turnout or a sufficient majority. Suppose there were a 50% turnout threshold, and the outcome of a referendum was that 49% of the electorate voted "Yes" and 10% "No". Then it would probably be reasonable to implement the measure concerned. If, on the other hand, the result were to be 25% "Yes", and 22% "No", it would probably be reasonable not to proceed. There was a 34% turnout in the referendum on the London mayor and assembly in 1998, which had no threshold requirement, and the government took the view that, with a 72% "Yes" vote, the measure should be implemented.

  Similar considerations hold when the threshold is in the form of a minimum percentage of the registered electorate being required to vote "Yes". In Denmark in 1939, there was a referendum on the abolition of the upper house, proposed by the government. 92% of those voting supported this measure. But, because turnout was below 45% of the electorate (the then mandatory requirement in Denmark for constitutional change—it has since been lowered to 40%), the government could not implement the change, even though over nine out of 10 of those voting had supported it.

  A threshold requirement in the form of a proportion of the registered electorate is likely to depress turnout. For a "No" voter might believe that an abstention was equivalent to a "No" vote, and that she need not, therefore, bother to turn up at the polls. Simply by staying at home, she would in effect be voting "No". An extraneous factor such as the weather on polling day may influence the result. Suppose that in the Scottish devolution referendum of 1979, when the threshold was 40% of the electorate, turnout had been 80%, and the outcome had been 41% "yes" and 39% "No", but one-quarter of the abstainers, i.e. 5% of the electorate, had stayed at home in the belief that abstention was the same as voting "No". The true strength of the Noes would be not 39% but 44%, and the "Noes" would have won. A threshold, therefore, may confuse voters and produce an outcome which does not reflect their true intentions.

  There is, however, a strong case for using a threshold in Northern Ireland where a simple majority, if composed almost entirely of the majority, Unionist community, might not be thought sufficient. Instead, a majority of those in both communities may be needed to secure legitimacy. It is, however, difficult to ascertain the precise composition of a majority other than by asking voters to label themselves "Unionist" or "Nationalist". Therefore a qualified majority large enough to ensure that at least a substantial proportion, if not a majority, of the minority community, as well as of the majority community, would be needed. Where there is to be a threshold in a referendum, it is better to implement it in the form of a specific percentage of votes cast, rather than a percentage of the eligible electorate.

  Outside Northern Ireland, however, it seems more in accordance with the constitution as it has developed to allow the government and Parliament to make the final decision after a referendum, using its own judgment where there is a narrow majority on a low turnout. The government and Parliament decided, perhaps rightly, not to allow Scottish devolution to go ahead in 1979 despite the small positive majority for it—33%-31%—far below the 40% threshold which Parliament had set. On a figure of 39% to 31%, however, the government would almost certainly have proposed to allow devolution to go ahead. But it is difficult to specify in advance the precise margin or size of turnout which would justify the government and Parliament in its decision whether or not to accept the outcome of a referendum. That is a matter perhaps best left to the discretion and judgment of MPs.

  There might, however, be a stronger case for a turnout requirement in local referendums. For the average turnout in local elections is under 40%, and there is some danger of vociferous local minorities imposing their policies on the apathetic majority, who do not bother to vote. It might be reasonable perhaps to require a turnout of, say, 35% for the outcome in local referendums to be accepted as valid.


  There is much more scope for the referendum at local than at national level. The Local Government Act, 2000, allows, but, following the Local Government and Public Involvement in Health Act of 2007, does not require, a local authority to hold a referendum before introducing a directly elected mayor system. But the Local Government Act, 2000, also introduced, for the first time into British politics, the initiative. The Act provided that any 5% of registered local electors could, by signing a petition, require a local authority to hold a referendum.[7] The purpose of introducing this device was to overcome the opposition of the local authority establishment, and in particular, local councillors, to the introduction of directly elected mayors. So, for the first time, voters were given the power to override the wishes of a local council which was unwilling to hold a referendum on the mayor option.

  The referendum allows voters to repair sins of commission by government. The initiative allows them to repair sins of omission. The referendum, as it currently operates in Britain, is a weapon that can be used only by the political class. The initiative is a weapon to be used by the people.

  There seems no reason why the initiative should be confined to just the one issue of a directly elected mayor. It could be argued that if voters are to be entrusted with the decision as to how their local authority is to be governed, they might also be entrusted with the decision, for example, as to how their local authority should be elected. 5% of local authority electors could be allowed to petition for an alternative voting system for local elections. They could be allowed to petition also on such matters as the shape and size of the local authority budget, or the organisation of the schools in their local authority. The initiative is an innovation with very radical possibilities.

  I would hope, therefore, that the Committee will examine the use of the referendum at local as well as national level. Local initiatives perhaps lack the glamour of national referendums, but they can yield real "double devolution", that is devolution not merely from central government to local authorities, but from local authorities to the citizen. They could prove an instrument to encourage participation at local level, so contributing to the regeneration of our democracy.

31 December 2009

1   This short note summarises material elaborated upon in chapter 7 of my book, The New British Constitution, Hart, 2009. Back

2   There have of course been many referendums and initiatives at state level in the United States. Indeed, every state except, for some reason, Delaware, requires a referendum to amend its constitution. Back

3   David Butler and Austin Ranney, eds, Referendums Around the World, American Enterprise Institute, Washington, 1994. Back

4   S.E.Finer, ed, Adversary Politics and Electoral Reform, Anthony Wigram, 1975, p. 18. Back

5   House of Commons Debates, vol. 888, col. 293, 11 March 1975. Back

6   House of Commons Debates, vol. 926, cols. 275ff, 15 February 1977. Back

7   There is also provision for voters to require a referendum on the abolition of grammar schools. Back

previous page contents next page

House of Lords home page Parliament home page House of Commons home page search page enquiries index

© Parliamentary copyright 2010