According to the latest opinion polls the race to win Britain’s EU referendum is close. In the ‘Poll of Polls’, once undecided voters are excluded Remain is currently on 51% and Leave is on 49%. This raises an intriguing but neglected question: how might changes in the national economy and levels of turnout affect the vote on June 23?
To contribute to the national debate we have examined how any changes in the level of unemployment and turnout might impact on public support for remaining in the European Union. Our measure is public approval or disapproval of EU membership. Our data come from the monthly Essex-YouGov Continuous Monitoring Surveys (CMS) that have been conducted between April 2004 and 2015. In what follows we assume that people who disapprove of Britain’s EU membership will vote to leave the EU while those who approve of membership will vote to remain.
Let’s begin by looking at the relationship between unemployment and public attitudes toward the EU. In recent months unemployment in the country has fallen to reach its lowest rate in more than a decade. Unemployment was 5.1% in the three months to November 2015, its lowest rate since the three months to October 2005. Since then, it has risen slightly (to 5.2%). How might future changes impact on the referendum result?
This is an important question because we know that unemployment is a key indicator that voters use to assess the country’s economic performance. As past research shows, rising unemployment undermines public support for the prime minister, who in this case is also a leading spokesperson for remaining in the EU. Any rise in joblessness between now and the referendum could also trigger a decline in support for EU membership. This is especially the case since Remain’s argument depends heavily on the prudential claim that EU membership brings economic benefits. Rising unemployment will be seen by the electorate as evidence to the contrary, that in fact the economy is not doing well.
Indeed there is a strong negative correlation (r = -.70) between public approval of Britain’s EU membership and the monthly seasonally adjusted unemployment rate—when unemployment goes up, approval of EU membership goes down.
Given this relationship we can use anticipated changes in unemployment to create a forecast of public approval or disapproval of EU membership between January and June 2016. Our forecast uses a simple dynamic model, specified such that the balance of EU approval-disapproval at time t is a function of EU approval-disapproval in the previous time period plus the contemporaneous (time t) level of unemployment. We use the Essex CMS data from between April 2004 and December 2015 to estimate the model and then forecast the balance of EU approval-disapproval for January thru June 2016 conditional on assumptions about levels of unemployment during this period.
When it comes to how unemployment might change between now and June 23, we consider three scenarios: (1) unemployment could remain at its current rate of 5.2%, (2) it might increase to around 5.5%, or (3) it could rise more significantly to around 6%.
What happens in each of these three scenarios to public support for EU membership? Under all three the balance favours staying in the EU—if unemployment (1) remains at its current level then public approval of Britain’s EU membership is likely to be 4.5 points ahead of disapproval (2) if there is a slight increase in unemployment then public approval is likely to be 3.9 points ahead of disapproval, and (3) if there is a significant increase in unemployment then approval is likely to be 3.1 points ahead.
But what about the question of turnout? How might different levels of turnout among Remain and Leave voters affect the vote?
According to recent polls public enthusiasm for the vote on June 23 could play a decisive role in the outcome. Some suggest that there is an ‘enthusiasm gap’—that while Leave voters appear to be more enthusiastic and committed to turning out on polling day Remain voters appear less likely to go to the polls.
According to data recently collected by Ipsos-MORI, for example, Remain held an 8-point lead but once intention to vote had been taken into account this lead dropped to only 2 points, well within the margin of error. We can now shed more light on this.
Differential turnout could play a critical role. In short, if there is an ‘enthusiasm gap’ on the day of the vote—meaning that Leave voters are more enthusiastic than Remain voters—then Brexit could win.
Figures 2 and 3 illustrate this by showing what could happen in two scenarios. Figure 2 shows a ‘high turnout referendum’. In this figure we assume turnout among Remain voters’ reaches 70% while turnout among Leave voters is even higher, at 80%. We continue to take account of unemployment and possible changes in the number of jobless citizens. In this high turnout scenario where Leave enjoys a 10 point lead in turnout then Brexit wins, no matter what the level of unemployment it is between 1.1 and 1.7 points ahead of Remain.
Figure 2 also shows a second scenario, in which the difference in turnout between Remain and Leave is lower, at 5 points rather than 10 points. Under this scenario turnout among Remain voters is 75% while turnout among Leave voters is 80%.
Under this scenario Britain votes to Remain in the EU but only if unemployment is at the lowest two levels (i.e. 5.2 or 5.5%). However, if between now and the referendum the level of unemployment rises to 6% and Leave is able to encourage a 5 point advantage on turnout then the referendum is forecast to be a dead heat. Alternatively, this suggests that if unemployment remains at or near its current rate than Remain could absorb a 5-point lead for Leave in turnout but still finish ahead.
But some will rightly wonder whether turnout on June 23 will reach high levels. Granted, turnout at the recent 2015 general election was 66% while at the independence referendum in Scotland it surpassed 84%. Yet it seems unlikely that voter participation the UK-EU referendum will be this impressive given that the underlying issue of the EU has only rarely featured in the ‘top five’ most pressing concerns for voters.
Turnout at recent EU referendums in some other countries also has been modest, for example 32% at the recent EU-Ukraine referendum in the Netherlands, 53% at the first referendum on the Lisbon Treaty in Ireland, or 42% at the referendum on the EU Constitution in Spain.
Figure 3 reveals what might happen if turnout at Britain’s EU referendum falls to lower levels and there remains an enthusiasm gap of between 5 and 10-points. If overall turnout falls into the low 50s and there is a 10 point enthusiasm gap between supporters of Leave and Remain (i.e. turnout among voters is 50% but among Leave voters is 60%) then Brexit wins under all three scenarios, regardless of the state of the economy.
If, instead, there is only a 5-point enthusiasm gap between the two sides (i.e. turnout among Remain voters is 50% but 55% among Leave voters) then Brexit wins again in two of the three scenarios. Only if unemployment hovers around the lowest level of 5.2% does a dead heat emerge. But it should be underscored that in none of these scenarios is Remain forecast to hold a lead.
So, what does all of this suggest? Clearly, statistical forecasts have uncertainty and should be treated with caution. Yet what we have found is broadly consistent with past research. Any deterioration in the economic picture is likely to strengthen public support for Brexit. Assuming that the health of the economy deteriorates between now and June, consistent with the Office of Budgetary Responsibility forecasts published last month, then the outcome is likely to be influenced heavily by turnout.
For Brexit to occur this gap cannot be trivial. If Leave holds a 10 point advantage and overall turnout falls below 75%, then our forecasts suggest that Brexit is a distinct possibility, even if unemployment stays at 5.2%, where it was at the beginning of 2016. Our forecasts also suggest that Brexit is likely if overall turnout slumps into the low-50s and Leave manages to establish only a 5 point advantage on turnout. For both the Leave and Remain camps, therefore, getting their supporters into the polling booths looks set to be one of their most important tasks.
Harold Clarke is Ashbel Smith Professor, School of Economic, Political and Policy Sciences, University of Texas at Dallas. Matthew Goodwin is Professor of Politics at the University of Kent and Senior Fellow at The UK in a Changing Europe initative. Paul Whiteley is Professor of Politics at the University of Essex.
Disclaimer and Notes: As should always be the case for political-economic forecasts, this one comes with an econometric health warning. Confidence bands for the balance of approval-disapproval over the Jan – June 2016 period are predictably wide. As always, fortune seeks hostages. The forecast equation is a simple regression equation: The model is estimated for the April 2004 – December 2015 period and then a conditional forecast is made thru June 2016. Analysis is performed with EVIEWS 9.5.