The unexpected results of the Brexit referendum are still very far from being metabolised, all the contrary in fact. An overwhelming wealth of analysis and commentary on the vote has been produced in the past few days and thrown at us from every direction. We took a good look at the best of it, and… decided to produce our own.
It was quite disappointing that the smallest units for which the results were released are local authorities, as that prevents a more detailed analysis of a smaller granularity of the vote. Still, some interesting considerations can be drawn. Starting from the picture above.
As predicted, Scotland was decidedly in favour of remain, while England voted to leave, with the exception of inner London, the cores of the largest metropolitan areas (excluding Birmingham/West Midlands), and most university towns. The result was a narrow Brexit victory, leaving the country deeply divided along a handful of very significant lines.
Here is a quick spatial exploration of the most significant factors in the English and Welsh vote, where Leave prevailed.
The electoral campaign was largely centred on immigration, however as it is often the case with UK elections, the amount of actual immigrants seems to matter very little, and to have had, if any, an inverse relation with the leave vote. More diverse places, with a high proportion of foreign born residents, are more likely to have voted to remain, with a correlation index of -0.53.
The correlation stays negative even if we take London and other urban areas out of the equation.
Education, skills, and occupation
It has been stressed that graduates were enthusiastically pro-EU. Such claim seems confirmed in the plot below, that shows local authorities with higher numbers of highly qualified residents (BA and above) being overwhelmingly more likely to have voted Remain. Correlation here is extremely high r=.89
The opposite is also true. Areas with a high proportion of residents with no formal qualifications have voted massively to leave, although with a slightly lower index r=.76
At the same time, we found very little correlation between areas of high unemployment and worklessness and the Leave vote. Interestingly, a much stronger correlation exists between the leave vote and areas with a high share of residents employed in jobs with a low level of specialization and qualification. The highest correlation was with areas where jobs described in the census as “Lower Supervisory and Technical Occupation” prevailed. These are caretaking, maintenance and craftsmanship jobs with mostly routine tasks. It is also likely that many fishing jobs fall in this category. The East coast of England was strongly pro-Brexit, and criticism towards the EU fishing policy was one of the factors.
However, it seems likely that it was working people, not the unemployed, who voted against the EU for the most part. Especially people working in low paying, insecure jobs, likely to have suffered from welfare cuts, zero hours contracts and other measures pushed by the last Conservative governments.
The prevalence of Leave votes in areas of low qualified jobs and low skills prompted us to to look at the one synthetic measure of “social suffering” existing in the UK Census, that of deprivation. The multiple dimensions of deprivation (relative to housing, income, health and more) are often combined in a single index. We chose to measure the proportion of households experiencing two or more deprivation dimensions in each constituency, and plot it against the Leave vote share. The correlation turns out to be quite weak: only 0.32.
A closer look at the plots is quite revealing though. All the biggest outliers, at the bottom and bottom right corner of the plot, are London boroughs. Without considering London, the correlation jumps to 0.52. This might be a sign of the London factor, or rather, the urban factor: another supposedly big fracture in the brexit vote, that between urban and non urban England (and Wales).
The urban factor
London (or The European People’s republic of London, as someone aptly dubbed the capital) has reported among the highest shares of the Remain vote across England. Not all London voted to remain though. Outer boroughs in east London casted a massive Leave vote, in Barking and Dagenham, Havering, and Bexley – all boroughs with high levels of deprivation. But very similar levels of deprivation can be found in places that voted in favour of Remain, slightly, like Newham, or enthusiastically, like Hackney.
What is the difference between these boroughs?
It is probably gentrification.
Barking and Dagenham, together with areas further afield, are the only remaining portions of East London not to have massively gentrified. They didn’t experience the influx of the young, affluent demographics that Lewisham and Newham saw after the Olympics, and Tower Hamlets had from long before.
In other words, Newham and Dagenham have very similar levels of deprivation, but Newham has many more graduates. That, and little else explains the difference in the vote. Moreover, Barking and Dagenham, along with other outer east London boroughs is where many of the displaced from the neighbouring gentrified boroughs were pushed to, further increasing the pressure on services and skewing the social make-up of the borough.
The same argument holds for urban areas outside of London. If we take London out of the deprivation plot, the next big outliers are Cardiff, Bristol, Brighton, Manchester, Liverpool, and Gwynedd.
Gwynedd is not a city, but is the most Welsh speaking county in Wales (two thirds of the population speak Welsh, compared to 20% in the rest of Wales), so a Scotland-like dynamic might be at play there.
If we plot cities by their proportion of highly qualified residents, we discover what Cardiff, Bristol, Brighton and Hove, Manchester and Liverpool have that other areas lack: graduates.
Manchester displays the very same London trend in a very vivid way, with its central area voting to Remain, surrounded by local authorities with a very similar (if not lesser) level of deprivation and lack of qualification, massively in favour to Leave.
The only real outlier seems to be Liverpool, and it is also worth noting that the top outliers in the opposite quadrant, at the top of the plot below, are Doncaster and Rotheram, among the places with the highest UKIP vote in the country.
In conclusion, without including cities and metropolitan districts, the correlation between Leave and deprivation is much higher, .62, and almost total with lack of qualifications, .80.
Being in a city, apparently makes it less likely for the urban malaise to translate into a Brexit vote. However this is not due to the atmosphere being different. It is not that people living in the cosmopolitan, open minded, diverse environments of city cores have been drawn towards remain even though their profile would match that of a leaver. The city effect is likely to be simply a byproduct of the deep inequality of these areas.
The urban areas that defy the logic of deprivation/lack of skills translating into high Brexit vote, are also those that have been mostly gentrified in the last years, just like central London boroughs. The population of these places is more polarised that elsewhere, high levels of deprivation coexist with affluence and high skills. The two groups are segregated, but happen to fall within the same district. The very oddly shaped Manchester City Council area, for instance, includes some of the most deprived neighbourhoods of the UK, along with some very affluent areas, plus many students and graduates. The intra-urban vote breakout of these places probably reflects more closely the general trend, as Charles Pattie argues in the case of Sheffield
The distance from the regression line could simply be determined by one sector of society having turned out in slightly larger numbers than the other, or being represented in slightly higher numbers than the other. Lack of intra-urban data unfortunately doesn’t allow further exploration, but it is extremely likely that the intra-urban closely reflects the inter-urban.
What was it, then?
It was insecure, medium and low skilled jobs, lack of qualification. That, along with poverty and deprivation. Places where these factors concentrate, in or out of big cities, in London and elsewhere, where young people are more represented and where the older generation is larger. These placese are overwhelmingly more likely to have voted to leave the EU.
Places that defy the rule, the cores of big urban areas – Liverpool, Manchester, Bristol, Brighton – seem to do so, in a sort of trompe l’oeil, only because we can’t take the analysis at a more granular level, but we can safely assume that deprived and working class areas within these local authorities will have voted to leave.
The Brexit vote was mainly a protest vote. When asked to vote in favour or against the status quo, places with low quality jobs, lack of skills and deprivation voted against. In cities, just like far from them. It is a trend that exists across Europe and beyond. The recent Italian elections come to mind. The second round of voting in Turin’s mayoral elections exemplifies the invisible (but very spatially explicit) divide between haves and have nots: areas with low academic qualifications around the elegant and affluent city centre voted en masse against the incumbent mayor Fassino, in favour of anti-establishment candidate Chiara Appendino of the 5 Star Movement.
Such pattern is replicated in most Italian cities, starting with Rome. Our next analysis of the Italian vote will show it in detail.