Written evidence from The Centre for European Reform

Opinion piece (Public Administration and Constitutional Affairs Committee)
22 September 2016

Public Administration and Constitutional Affairs Committee

Lessons Learned from the EU Referendum inquiry


Executive Summary

Referendums on constitutional questions are too important to be settled by a simple majority of those who vote on the day.

There should be more effective measures to ensure that false information used by campaigns and the media is corrected quickly, so that voters are not misled.

A neutral ‘fact-checking’ agency would put more pressure on campaign groups in referendums (and political parties in general election campaigns) to be specific about their proposals and to acknowledge any shortcomings.

Because Leave campaigners never had to specify what their alternative to EU membership was, the Civil Service could not make contingency plans of any value, other than a plan to deal with market uncertainty.


1. The Centre for European Reform is a think-tank devoted to making the European Union work better and strengthening its role in the world. The CER is pro-European but not uncritical. The CER has published extensively both before and since the EU referendum on ways in which the EU could be improved, the UK’s role in the EU and the implications for the UK and the rest of Europe of Britain leaving the EU. Though the CER is not a campaigning organisation, it took a close interest in the referendum campaign, its conduct and the arguments made by both the Leave and Remain sides.

The role and purpose of referendums

2. Even in a representative democracy in which Parliament is sovereign, there will be some important constitutional questions which Parliament wants to hand back to the population to decide in a referendum. But if a government commits itself to acting on the result of the referendum (whether or not in legal terms the vote is advisory or binding), the outcome should meet a higher threshold than the support of 50 per cent plus one of voters on the day. A referendum should succeed if it passes two tests:

a. Did a large number of people think that the issue mattered to them? If turn-out in a referendum is very low, it could mean that there is a vocal minority interested in the subject, but the majority is indifferent.

b. Did the result show that the population had a clear-cut view in favour of one option? Implementing major constitutional change on the basis of 50 per cent plus one could create or exacerbate divisions in society.

3. British law already acknowledges that some decisions need more than a simple majority of those voting if they are to be regarded as ‘legitimate’. Under the Trade Union Act 2016, a vote for industrial action on London Underground (for example) will only be valid if turn-out is over 50 per cent, a majority of those who vote are in favour, and those who vote in favour constitute more than 40 per cent of the eligible voters. It is absurd that a constitutional decision with permanent and far-reaching consequences does not have to meet any such threshold.

4. It is not surprising that in the case of the EU referendum, where the number of votes in favour of leaving amounted to about 37 per cent of eligible voters, many on the losing side continue to challenge the legitimacy of the result. Indeed, in advance of the vote, Nigel Farage, then the leader of the United Kingdom Independence Party and a leading campaigner to leave the EU, suggested that those in favour of Brexit would not have accepted a narrow vote to remain in the EU either: he told the Daily Mirror “In a 52-48 referendum this would be unfinished business by a long way. If the remain campaign win two-thirds to one-third that ends it.”

5. There are a number of ways in which the outcome of a referendum could be given greater legitimacy. One would be to use the model of the Trade Union Act 2016, so that a referendum would only change the status quo if it met thresholds for turn-out and support in the electorate. The Scotland Act 1978 (the first attempt at devolution of powers to Scotland) took a similar approach: the Act would have come into force if 40 per cent of the total Scottish electorate had voted in favour. Another possibility (hinted at by Mr Farage) would be to ignore turn-out, but require a super-majority of those voting, whether of two-thirds or another figure, to change the status quo. The referendum on Montenegrin independence from Serbia in 2006 had both a turn-out threshold (50 per cent) and a super-majority (55 per cent).

6. When the House of Lords Select Committee on the Constitution reported in 2009-2010 on referendums, they concluded that thresholds and super-majorities either encouraged manipulation of the system (a side trailing in the opinion polls could invalidate the vote by getting its supporters to boycott it) or gave extra weight to votes in favour of the status quo. The committee suggested, however, that such measures might be justified in exceptional circumstances, such as in ‘divided societies’. The example used was the split between Unionist and Nationalist communities in Northern Ireland.

7. Though the government chose not to include a threshold or a super-majority in the EU referendum act, the result showed that British society was indeed divided. The national majority to leave the EU masked very considerable differences between Scotland and Northern Ireland (voting by significant percentages to remain) and England and Wales (voting by smaller percentages to leave). It remains to be seen what impact these divided voting patterns will have on the long-term cohesion of the United Kingdom, but with the benefit of hindsight it would have been better to set the bar for changing the status quo higher.  

The regulatory system for referendums, and the campaign

8. A referendum campaign is not like a general election campaign, and the purdah rules that apply to the latter do not work well for the former. In an election campaign, parties have manifestoes which set out what they plan to do to change the status quo. In the EU referendum, the government had a policy, which was to endorse the status quo (though ministers were allowed to campaign against it in their personal capacities); the Leave campaign avoided specifying what the alternative to the status quo was (different Leave campaigners offered different and often incompatible policy prescriptions for life outside the EU). The picture was further complicated by the fact that there was an officially designated Leave campaign (Vote Leave) and a rival, unofficial campaign (Leave.EU), as well as a variety of advocates against EU membership who were not aligned with either of these movements, and who put forward their own views on what sort of Britain they wanted to see post-Brexit.

9. The material published by the Civil Service during the campaign was as impartial as it could be, though the In campaign sometimes used it in misleading ways. The Treasury issued a scenario-based central estimate that GDP per household would be £4,300 lower by 2030 if the UK left the EU than if it remained. The then Chancellor of the Exchequer used this figure to suggest that every household would be £4,300 poorer as a result of Brexit. But GDP per household and household income are different things. The latter is much lower than the former. George Osborne could have made the more accurate but less eye-catching claim that household incomes would grow more slowly as a result of Brexit; the Civil Service was not to blame for the fact that he did not. 

10. The Civil Service was unable to rebut misleading or entirely false claims by the various Leave campaigns and by pro-Brexit media outlets. Despite the efforts of the Chair of the UK Statistics Authority, Sir Andrew Dilnot, Vote Leave continued to claim, falsely, that the UK contributed £350 million per week to the EU. Leave.EU claimed, falsely, that Turkey was on the point of joining the EU and that 75 million Turks would have access to the NHS. The Daily Express published false claims about the number of Turks intending to migrate to the UK, and about a non-existent EU regulation banning the sale of eggs in dozens. In some cases it withdrew or corrected these stories only after the referendum had taken place. It is impossible to know how many voters were influenced by these stories, but some people have subsequently told the media that they were.

11. If governments intend to make more use of referendums in the future, there is a case for looking at a number of steps to ensure that campaigns and the media do not deliberately or inadvertently use false information, and that if they do they have to correct it quickly and prominently. One model for neutral ‘fact-checking’ is the Netherlands Bureau for Economic Analysis. This has responsibility during Dutch general election campaigns for checking the manifestoes of political parties to cost their proposals, and to say whether their proposals are realistic. The Bureau assesses the fiscal and budgetary implications of the manifestoes and judges whether the plans are based on sound economic assumptions, or not. Parties voluntarily submit their manifestoes to the bureau, but if they do not, voters may reasonably question the economic feasibility of their plans. In the case of the EU referendum, such a system could have provided advice on the likely economic effects of the various alternatives to EU membership suggested by Leave campaigners, as well as checking the Treasury’s scenarios.


12. It was impossible in the circumstances of the EU referendum for the Civil Service or the Government to plan for a vote to leave (other than the short-term steps to manage market instability on June 24th) because there was no programme from the Leave campaigners. If the UK had had an agency modelled on the Netherlands Bureau for Economic Analysis, it might have forced the Leave campaigners to set out more concrete proposals, the economic consequences of which could be modelled. The Civil Service would still have had to make several different contingency plans, however, responding to the different models for the UK’s relationship with the EU post-Brexit. And at some point after the referendum, the winners would still have had to decide which model to adopt – something which (at the time of writing) has still not happened.

Ian Bond is director for foreign policy at the Centre for European Reform.