Dis-United Kingdom? Inequality, growth and the Brexit divide


Much has changed in Britain since the EU referendum, but in many ways the divide that opened up around the vote feels as cavernous today as it was on the morning after the night before. That owes much to the inevitably divisive nature of a binary in/out referendum of course, but many commentators point also to an economic partition in the country that has underpinned very different experiences of, and attitudes to, the workings of our political economy.

For many, the Brexit vote has added to the case for tackling income inequality and, in the words of the Prime Minister, ensuring that our country’s growth “works for everyone”.[1] And yet, on any number of measures, relatively little appears to have altered in the UK in relation to income inequality over the past 25 years. It is undoubtedly high by international standards, but that doesn’t explain why any discontent should be manifesting itself only now.

This essay considers the role economics played in the referendum result and argues that it is not just inequality that mattered, but more particularly the specific recent combination of high inequality and low growth. The fact that the country is now on course for a period in which rising inequality and falling incomes come together in an unprecedentedly toxic combination is therefore especially worrying.

Brexit: not just economics, stupid

At first glance, the link between economics and the vote for Brexit appears obvious. Figure 1 details a clear correlation between median pay and the Leave vote in each of Britain’s 380 local authorities, with lower paying areas much more likely to have voted in favour of Brexit. Yet, as the chart also shows, the relationship was far from uniform. The spread of the Leave vote share varied considerably within both lower-paying authorities (such as Boston and Barnsley in the yellow group, and York and Exeter in the red group) and higher-paying authorities (such as Havering and Watford in the light blue group, and Lambeth and Edinburgh in the dark blue group). To look at what location-specific characteristics actually drove the variation in vote, we need to run a regression analysis.[2] The results are reassuringly complex.

Even after controlling for all other factors, we find that economics did play a role. However, recent differences in experience – such as changes in pay or shifts in the industrial composition of an area – didn’t feature. Such elements, and the wider economic turmoil that followed the financial crisis, may well have increased the propensity to vote Leave across the country, but they didn’t explain any variation in vote. What mattered instead were the economic differences established across local authorities some decades ago.

Figure 1: The Leave vote was correlated with local pay levels, but there was more going on

Source: Office for National Statistics

However, economics was only part of the story. Demographics also mattered with, most eye-catchingly, immigration featuring as a statistically significant predictor of the Leave vote. Not its level, but rather its pace of change. That is, there was no statistically significant difference in the proportion voting Leave between areas of low and high immigration; but significant differences did arise in those communities in which the proportion of non-UK-born residents changed most markedly in the decade before the referendum. The implication is that areas experiencing the most visible change in population were more likely to vote Leave. The number of students in an area also had a statistically significant effect, lowering the Leave vote holding all else constant.

Cultural factors played a role too. The Leave vote was stronger in areas where higher proportions of people said that residents in their neighbourhood didn’t get on with people who are “different.” And specific regional effects were also apparent; for instance, communities across Scotland proved more likely to vote Remain than all of their other characteristics would suggest they should have. In that instance, the political leadership of the Scottish Nationalist Party and a preference for Brussels over Westminster is likely to have been important. But regional effects were evident in other parts of the country too.

The complexity of the drivers of Brexit was neatly encapsulated in the fact that the three themes of economics, demography and cultural forces came together in the biggest single predictor of the vote in an area — namely the proportion of the population holding a university degree (or equivalent). Yet, while this analysis makes clear that the notion of the referendum result as an economic protest vote is way too simplistic, it is apparent that living standards played an undeniably important role. And as noted, the vote highlighted old divides rather than new, begging the question: why now?

Income inequality has changed little over the last 25 years, but a generalised slowdown in growth has given new urgency to the debate

As Figure 2 shows, headline income inequality (as measured by the internationally-recognised Gini coefficient) has been broadly flat since the early-1990s. It did, of course, pick up extremely sharply over the course of the 1980s – and remains high by international standards as a result – but there is nothing to suggest that things have got noticeably worse in recent years. Indeed, if anything, inequality has edged downwards in the post-crisis years. In considering why things might feel different however, three factors appear important.

Figure 2: Household income inequality has been broadly flat over the last 25 years

Source: Institute for Fiscal Studies analysis of DWP, Family Resources Survey

First, it’s not just how the pie is divided that matters to the mood of a nation, but the size of the pie too. Figure 3 details a series of step changes in UK economic growth over the past four decades, with average annual growth in GDP per capita slowing from 3.2 per cent in the 1980s to just 1.2 per cent in the period since 2010. All else equal, we might expect the same level of economic inequality to matter more in a period of slower overall growth.

Figure 3:  Real-terms GDP per capita growth has slowed significantly in recent years

Source: Office for National Statistics

Secondly, the standard measure of income inequality ignores the housing headwind that has played such a major role in determining living standards in the UK over recent decades. Rising housing costs have eaten increasingly into incomes across all parts of society, but the effect has been most pronounced among low to middle income households. In part this reflects the sharp increase in housing costs recorded across all tenures in the 2000s. But it also stems from a  compositional shift, with low and middle income households more likely than in years past to find themselves stuck in the private rented sector and therefore facing higher day-to-day housing cost-to-income ratios than exist among mortgagor households.

Figure 4 shows the effect of recalculating the Gini coefficient after accounting for housing costs. On this measure – which arguably more accurately captures the lived experience of income growth for households – two things stand out: inequality is higher than the standard measure suggests, and it has continued to rise in recent years (albeit not as rapidly as in the 1980s). Irrespective then of what the headline measure shows, the living standards divide across households is likely to have felt like it has grown still wider over the last 20 years.

Figure 4: Measured after housing costs, income inequality has been rising

Source: Institute for Fiscal Studies analysis of DWP, Family Resources Survey

Thirdly, new divisions have opened up in recent years – with age marking a particular distinction. Several problems are visible. Younger people have experienced the biggest pay squeeze in the aftermath of the financial crisis and have found themselves increasingly locked-out of the home ownership that seemed normal for their parents. Working-age families have also borne the brunt of benefit cuts introduced in the name of fiscal consolidation, even as the triple lock has provided protection for the state pension.

As a result, recent income growth has tended to be weaker for working-age households than for pensioners. Figure 5 shows that typical working-age household incomes have barely grown at all since the early-2000s, in direct contrast with solid growth in median pensioner household income. The stagnation of incomes among working-age households is likely to have further focused minds on apparent economic unfairness, with the group feeling like it is working no less hard than in generations past but without the income progression that was previously taken for granted.

Figure 5: Working-age household incomes have grown little in real-terms over the last 15 years

Notes: Incomes are measured after housing costs. CPI-AHC deflator is a variant of the CPI that removes housing elements in order to match the after-housing-cost nature of the income measures shown here.

Source: Resolution Foundation analysis of DWP, Family Resources Survey

Taking these three factors together, we can start to understand why economic tensions might feel somewhat heightened at the moment. That is, while headline income inequality may not have risen in any significant way since the early-1990s, the already-large gap between rich and poor has taken on new meaning. Indeed, we can see that a given level of inequality can feel very different depending on the broader backdrop. Focusing on the working-age population, Figure 6 divides recent UK experience into four broad ‘episodes’ of inequality.

  1. ‘Classic’ inequality in the 1980s (1977 to 1991): Relatively strong growth underpinned a stretching out of the income distribution – with the top moving away from the middle and the middle moving away from the bottom. Average annual income growth of 4 per cent to 6 per cent a year in real-terms at the top end of the working-age income distribution contrasted with average growth towards the bottom of less than 1 per cent a year. While the country became much more unequal however, it is worth noting that income growth was present at all but the very bottom end of the distribution. Indeed, approaching three-quarters of the working-age distribution enjoyed real-terms growth in excess of 2 per cent a year.
  2. Strong, shared growth in the 1990s (1991 to 2002-03): Average income growth broadly matched that recorded in the 1980s, but gains were much more evenly felt. Average annual income growth of roughly 3 per cent a year was recorded across the entirety of the working-age distribution.
  3. Pre-crisis slowdown in the mid-2000s (2002-03 to 2007-08): Income growth slowed markedly between 2002 and 2007, averaging just over 0.5 per cent a year in the middle of the working-age distribution. Those at the top fared better, though average annual growth of just 1.5 per cent at the 90th percentile remained much weaker than the norms that had prevailed in previous decades. There was also evidence that the bottom fifth fell further behind in these years, with incomes falling year-on-year from around the 13th
  4. Post-crisis stagnation (2007-08 to 2016-17): Incomes fell sharply between 2007 and 2012 before recovering slowly thereafter. Some protection at the bottom end and the tendency of the highest earners to lose most during financial crises meant that inequality actually fell a little in the immediate post-crisis period, but the pattern subsequently reversed – with higher income households faring best as a result of sharp cuts in mortgage rates for instance. Overall, working-age incomes were stagnant across the entire distribution in this period.

Figure 6: Recent income growth has been evenly distributed, but much weaker than we’ve been used to

Notes: Incomes are measured after housing costs. CPI-AHC deflator is a variant of the CPI that removes housing elements in order to match the after-housing-cost nature of the income measures shown here.

Source: Resolution Foundation analysis of DWP, Family Resources Survey

To describe these four ‘episodes’ another way, the opening up of income inequality in the 1980s can be thought of as a crack in the roof of society that has gone unmended in subsequent decades. This mattered less when the sun was shining as the economy grew through the 1990s and early-2000s, but the rain has been falling for the last decade and people are getting very wet. The EU referendum is likely to have come along at just the right time for the doused to express their discontent.

Looking forward, the UK faces the prospect of an unprecedented combination of falling incomes and rising inequality

There is of course some irony in the fact that any economic discontent is likely to have been made worse in the near-term by the Brexit result itself. The sharp spike in inflation recorded in 2017 was driven primarily by the post-referendum collapse in Sterling, and led to a return of the wage squeeze that had prevailed for so long after the financial crisis. Stronger-than-expected global growth has supported the UK economy to some degree over the past year, but the country has dropped to the bottom of the G7 growth league and household incomes have been falling in real-terms.

But this would appear to be a price deemed worth paying by many Brexit supporters. Figure 7 shows that, while the population remained very evenly split in terms of its overall attitude to Brexit towards the end of 2017, significantly more households (a balance of 33 per cent) expected it to make the economy worse over the coming 12 months than expected it to make things better. Even among those households who declared themselves “a bit positive” about the referendum result, a net balance of 8 per cent displayed pessimism about its near-term effects.

Figure 7: Even many Brexit supporters expect it to have a negative economic impact in the near-term

Source: Bank of England, NMG Survey 2017

Many of those whose referendum vote was influenced by economic concerns may therefore choose to ‘look through’ any near-term difficulties. And 2018 is likely to feel better than 2017 for many: with the Sterling effect now dropping out of the inflation figures and some signs of wage pressures building in the economy, the pay squeeze at least should come to an end in the coming months.

What comes next is potentially more challenging, however. That’s because, even in the absence of any ongoing ‘Brexit headwind’, Britain looks on course for a period in which income inequality starts to rise again. And crucially, unlike the ‘classic inequality’ of the 1980s, it looks set to be driven by those at the bottom falling behind rather than those at the top racing ahead. Such an outcome follows from the confluence of three factors.

First, it looks likely that the living standards boost associated with rapid employment gains in recent years has finally run out of road. Post-crisis, employment growth has been both remarkable and distinctly pro-poor. Figure 8 details the change in employment rates between 2009 and 2015 in each decile of the household income distribution, and shows that growth was concentrated in the bottom half of the distribution. Such a pattern did not hold in earlier periods of strong employment such as the 1990s, when employment growth was much more evenly distributed. But employment growth has plateaued in recent months and, while there may be marginal further gains to be made, it is hard to envisage jobs growing at anything like their previous pace over the next few years.

Figure 8: Post-crisis employment growth has been strongly pro-poor

Source: DWP, Family Resources Survey, various

Secondly, the pace of economic – and therefore earnings and incomes – growth is expected to remain constrained by disappointing productivity gains. Figure 9 sets out the path followed by the UK’s output per hour over the last 14 years, along with successive projections from the Office for Budget Responsibility (OBR) for its forward trajectory. Having consistently forecast a return to a post-crisis pace of growth that hasn’t materialised, the OBR has significantly lowered its projection for trend growth in its most recent outlooks.

Figure 9: Following a decade of productivity stagnation, the OBR has nearly halved its projection for trend growth

Source: OBR, Economic and Fiscal Outlook, various

This new projection may of course prove too pessimistic, but it is in line with the approach taken by many other forecasters including the Bank of England.[3] If correct, it implies that the UK economy is already growing at close to capacity and that nominal pay growth is unlikely to return to its pre-crisis average of roughly 4 per cent a year – leaving real wage growth in the coming years some way short of the level we had grown used to in the pre-crisis era. The last decade has already been the weakest for pay growth in nearly 200 years;[4] the latest projections suggest that this disappointment will persist, with the pre-crisis peak in average pay not now restored until 2025 – completing a 17-year period of ‘lost’ growth.[5]

The third force at play in the coming years – and the one with the most direct bearing on inequality – is the continued roll out of the £14 billion of working-age welfare cuts announced at the Summer Budget of 2015. While long in-train, only around one-fifth of these cuts have so far been delivered. Further cash-terms freezes and other restrictions on benefit uprating mean that some key benefits will be less valuable in 2020 than in the 1980s, while families will also lose out from large cuts for new claimants. The continued roll-out of the new Universal Credit system will have mixed impacts, but cuts to its ‘work allowances’ mean it is now set to be less generous overall than the benefits it will replace.

Even allowing for the recent introduction of the higher wage floor in the form of the National Living Wage and a series of income tax cuts, the overall effect of policies introduced after the 2015 General Election is to significantly lower incomes in the bottom half of the distribution. By the end of the current parliament, the poorest third of households are expected to be £745 a year worse off than they would have been had no policy changes been made after March 2015.[6]

So, with employment growth slowing, wages rising only modestly and welfare cuts biting, there appears to be a very real chance that income growth in the coming years will continue to disappoint. Figure 10 sets out the OBR’s projection, showing real-terms household disposable income per person flat-lining over the next three years and growing only modestly thereafter. Having peaked in 2016, household income per person looks to be in the midst of a longer (at 19 quarters) though gentler squeeze than the one recorded immediately after the financial crisis (17 quarters between Q4 2007 and Q4 2011).

Figure 10:     Post-crisis income stagnation is set to persist over the coming years

Source:        OBR, Economic and Fiscal Outlook, various

We are on course then for the sharpest increase in income inequality since the 1980s, combined with an overall slowdown in income growth similar in magnitude to the one endured after 2008. Returning to the crack in the roof analogy, it would appear that we are about to see the aperture widening even as the rain continues to pour.

As always, these projections are far from certain. Nor are they unalterable: we have the capacity to change the economic outlook and government can directly influence inequality via its approach to taxes and benefits. But the outlook painted is a troubling one. Economic dissatisfaction was not the only driver of Brexit and the subsequent discord that appears to have arisen in the country, but it played its part. If there is more disappointment to come – and this time of a more straightforwardly regressive nature – we may yet face more political turmoil in the coming years.

From this perspective, the Prime Minister is right to say that we need to focus on growth that “works for everyone”: inequality is a source of ongoing disillusionment that has been overlooked for too long. But nor must we ignore the other important element of the living standards equation: growth itself. Tackling one without the other will do nothing to heal the divide that has long been apparent in Britain.

This article was shortlisted for the 2017/18 Rybczynski Prize


[1] T May, speech to the Conservative Party Conference 2017, The Conservative Party, 10 October 2017

[2] Technically, a clustered standard errors approach. See S Clarke & M Whittaker, The importance of place, Resolution Foundation, July 2016

[3] See for example, Bank of England, Inflation Report, February 2018

[4] Bank of England, A millennium of macroeconomic data

[5] M Whittaker et al, Sugar Rush: Spring Statement response, Resolution Foundation, March 2018

[6] A Corlett, G Bangham & D Finch, The Living Standards Outlook 2018, Resolution Foundation, February 2018