Why did we vote to leave? What an analysis of place can tell us about Brexit


So much has happened since the UK voted to leave the EU that it is hard to believe that the result was announced just three weeks ago. Already the pop analysis of what influenced the referendum outcome is becoming received wisdom, with politicians and others competing to see who can best deal with the sources of discontent they believe to be at the heart of the debate. But understanding what really happened will take longer than a few days. And the truth is likely to be complex. With so many issues being touted, can we disentangle the key drivers of the vote to leave?

To do this, new Resolution Foundation analysis uses data on the characteristics of people living in 378 of Britain’s 380 local authorities. By testing the strength of the relationship between each of these factors and the leave vote, while holding all other factors constant and controlling for specific regional effects, we’re able to explain around 90 per cent of the variation in the leave vote across different areas within Britain.

As the summary below shows, living standards, demographics and migration are all significant factors in explaining the leave vote, though not necessarily in the way that many people assume.


There are six key takeaways.

First, living standards matter. Reflecting on the results three weeks ago, we noted that recent changes in living standards hadn’t played much of a role – despite the contention that people were angry about the post-financial crisis squeeze on living standards – but that deeper rooted economic differences did. That finding is confirmed in our more detailed investigation, with employment levels proving the indicator with the strongest link to tendency for an area to vote Leave. A 10 percentage point (ppt) rise in the employment rate within a local authority is associated with a 1.7ppt fall in the leave vote. Areas where high numbers of people are out of work voted to leave.

Second, education matters. A lot. A 10ppt increase in the proportion of people with degrees (more technically, with qualifications at NVQ4 and above) is associated with a 4.5ppt fall in the leave vote. In our analysis education was the best predictor of how an area voted, reflecting the fact that it is closely related to pay, employment, and feelings of cohesion – bringing together both economic and cultural explanations for how areas voted and suggesting that more than just academic achievement is at work here.

Third, demographics matter. A 10ppt rise in the proportion of students within the population is associated with a 5ppt fall in the leave vote. The same increase in the ratio of over-50s to under-50s is associated with a 0.7ppt increase in the leave vote. This chimes with what we know about how people of different ages voted and the widespread support for Remain among students.

Fourth, migration matters, but only in areas where the migrant population has grown rapidly. That is, while the overall share of the population in a given area born outside the UK does not have a significant effect on the leave vote, the pace of change in the decade after 2004 does. Increasing the proportion of migrants in the population by 10ppt raises the leave vote by 3.9ppt. All areas that experienced at least a 7 per cent increase in the proportion of migrants over the last decade voted to leave.

Fifth, culture and cohesion play an important role. Chiming with the findings from Lord Ashcroft on the importance of attitudes at the individual level in explaining the vote, we find that a 10ppt increase in the proportion of people who believe that individuals from different backgrounds ‘get on well’ in their area is associated with a 3.9ppt decrease in the leave vote.

Finally, politics and wider geographical factors also matter. Local authorities in Scotland recorded leave votes that were 12ppts lower than the English average, even once all of the factors above were taken into account. In contrast, we found no specific ‘London effect’ when holding all these other factors constant.

While our analysis tries to include all possible factors that may have affected the vote (we tested many others and dropped them where there was no apparent correlation), not everything of importance can be measured at a local level. It’s also worth noting that this is an investigation of place, not people.

Nonetheless it clearly highlights the complexity of the vote. Yes, economics matters – and there is evidence that communities that have long been left behind helped to drive the referendum outcome. But demographics and culture play important roles too. Our analysis suggests that the key routes to repairing an apparently divided nation include taking action on improving people’s living standards, addressing concerns about the appropriate pace of migration and making areas more cohesive.

In this regard, Theresa May’s maiden speech as Prime Minister is encouraging. Her dual commitment to tackling “burning injustices” and making Britain “a country that works for everyone” has the potential to tackle some of the obvious issues raised over the course of the referendum campaign. While her premiership may ultimately be defined by her ability to deliver on the referendum vote and deal with the Brexit fallout, it is vital that these deeper-routed concerns remain priorities even as the vote itself fades further into the past.