Can Income Transfers Help Explain Brexit

BREXIT EFFECT Can Income Transfers Help Explain Brexit? Rahul Menon Notwithstanding the conventional narratives of the anti-immigrant sentiments holding sway during the referendum for Brexit, an empirical investigation of the possible factors leading to Brexit does not show any significant correlation between the share of migrants from European Union countries and the regions that voted for Brexit. Share of social security transfers in income was most important in determining if a region voted leave. Decreased employment opportunities and increased hardships that necessitate reliance on welfare payments could have fuelled the anti-immigrant and anti-European sentiment. Rahul Menon (menon.rahul@gmail.com) teaches Economics at St Xavier’s College, Mumbai. Economic & Political Weekly EPW AUGUST 13, 2016 T he United Kingdom’s (UK) decision to exit the European Union (EU)— dubbed Brexit—has dealt another hard blow to the global economy. The effects on trade and investment flows as a result of this decision cannot be foreseen with any precision. How will short-term and long-term growth in Britain be affected if the decision to withdraw from the EU is institutionalised and operationalised? What effects will the depreciating pound and the flight to safe assets have for future monetary policy decisions of the United States (US), and by extension the world economy? While fervent forecasts are being made about the future, just as much debate has been directed towards the causes motivating voters’ decisions to leave the EU. Around 52% of voters in the UK chose to exit the EU. Many pollsters and writers have pointed out the fact that a majority of older voters preferred to leave. While the young were largely in favour of staying in the EU, the relatively smaller turnout amongst younger voters saw to it that the leave vote carried the day. Two of the major factors influencing the vote—as reported by many online publications—were education and incomes; regions with higher incomes and a higher percentage of voters with advanced degrees voted remain (McGill 2016). Thus, while some theories highlight the influence of racist and xenophobic rhetoric of parties such as the UK Independence Party (UKIP) in motivating the decision to leave the EU, others hold that the stagnation of incomes and living standards are much to blame for the fact that a politically alienated populace saw no benefits to staying in the EU (Davies 2016). The data is still being sifted through, and this event will no doubt generate a large amount of academic discussion. While several studies look at current data and attempt a correlation with the vol lI no 33 results of the referendum, this article looks at the basic data in household incomes over the last 17 years to study its possible impact on the vote. The results show that one of the major factors determining whether a region voted leave— unreported by many studies so far—is the share of social security transfers per household. This brief study is confined to England and not Scotland, Northern Ireland or Wales. Data on voting percentages from over 300 voting constituencies in England were gathered from the British Broadcasting Corporation (BBC) website, and then organised by areas conforming to the Nomenclature of Territorial Units for Statistics (NUTS) codes so as to be in concordance with income and migration data from the Office for National Statistics (ONS). The data set consists of 122 regions of England, classified according to the NUTS 3 scheme. Rhetoric against Immigration The involvement of the UKIP and the uptick in racist events following the vote led many to conclude that the vote was largely a referendum on immigration, and that a violent conservative backlash against the ideals of modern cosmopolitanism was what explained the vote. The UKIP’s campaign urged voters to leave the EU, displaying photographs of Syrian refugees who according to them would now be let into Britain given that free movement of labour is allowed through Europe. The rhetoric may have been tilted against immigrants, but do we find evidence that supports the claim that leave voters were motivated by antiimmigrant rhetoric? On the face of it, there is nothing to support this assertion. When data from 122 regions of England were considered, one notices a strong inverse relationship between the number of EU-born residents and the probability of leave votes being cast (Figure 1, p 40); regions with a high number of foreignborn residents were more likely to vote remain. This does not completely negate the possibility of a strong antiimmigrant sentiment, as we shall hypothesise later, but it should question the narrative of anti-immigration sentiments 39 BREXIT EFFECT Figure 1: Correlation between Leave Vote Percentage and EU to UK Population Ratio by NUTS 3 Regions 40 relationship is not very strong, statistically speaking. 35 Social Security Benefits R2 ==0.4115 R² . 30 25 20 15 10 5 0 10 20 30 40 50 60 70 80 -5 X-axis indicates percentage of leave votes in each NUTS 3 region, y-axis indicates number of EU-born residents per 100 UK-born residents in each region. Figure 2: Correlation between Leave Vote Percentage and CAGR of Disposable Income Per Head, 1997–2014 5.5 5 2 ==0.2311 . RR² 4.5 4 3.5 3 A Tale of Two Englands? 2.5 2 1.5 10 The one variable that does tend to have a significant correlation with whether a region votes leave or not is the share of social security benefits and transfers received by a household. The share of benefits received (as a share of gross disposable income) by households rose from 25% in 1997 in both the UK and England to 28.4% and 27.8%, respectively. As Figure 3 and Figure 4 (p 41) indicate, there is a significant correlation between the share of leave votes in a region and the share of social security transfers in gross disposable income per household in 2014, as well as the difference in the share of social security per household between 1997 and 2014. 20 30 40 50 60 70 80 X-axis indicates percentage of leave votes in each NUTS 3 region, y-axis indicates the CAGR disposable income per head for each region over the period 1997–2014. Figure 3: Correlation between Leave Vote Percentage and Social Benefits Received Per Household as a Share of Gross Disposable Income, 2014 50 2 = 0.5419 RR² = . 40 30 20 What the Brexit vote reveals is a sharply divided England. A look at the regional share of votes reveals that the bulk of the remain votes were concentrated in areas whose residents enjoyed higher incomes and educational prospects; the cities of Manchester, Liverpool, London, Oxford and Cambridge all voted remain, while former industrial areas in the north polled heavily in favour of leaving. In statistical terms, there exist significant differences between the regions that voted remain and those that chose to stay, as Table 1 indicates. Table 1: t-Tests by Voting Region 10 Share of Difference in CAGR of Benefits, Benefits Share, GDHI 2014 1997–2014 Per Head 0 10 20 30 40 50 60 70 80 X-axis indicates percentage of leave votes in each NUTS 3 region, y-axis indicates the share of social benefits received per household in gross disposable income of each region in 2014. as the major cause of the high number of leave vote. Incomes and Income Growth According to the ONS, only two regions of the UK recorded positive income growth relative to average gross disposable household income per head; these regions were London and Scotland, both regions that voted remain. If the average disposable income per head of London was 22% higher than the UK average in 1997, it was 31.4% higher in 2014. Regions that 40 were poor tended to vote leave, but the relationship between leave votes and income growth has not been discussed. When applied to the sample, the relationship between income growth and leave votes tends to be weak. Figure 2 outlines the relationship of leave votes with the compound annual growth rate (CAGR) of disposable income per head. While there does seem to be a negative relationship, implying that regions with low income growth tended to vote leave, the low R-squared indicates that the Remain (35 regions) Leave (87 regions) Difference t-statistic Degrees of freedom p-value 23.29 31.26 -7.98 -8.10 120 0.000 -0.15 4.84 -4.99 -9.26 120 0.000 3.32% 3.07% 0.25% 3.02 120 0.0015 Regions divided according to whether a majority of votes were for remain or leave. GDHI stands for gross disposable household income. Regions that voted remain were, on an average, more likely to have a lower share of benefits per household to gross income in 2014, than to have a higher rate of growth of income per head, and have a reduction in the share of benefits over the period 1997–2014. All these differences are also statistically significant. AUGUST 13, 2016 vol lI no 33 EPW Economic & Political Weekly BREXIT EFFECT the odds of voting leave by 2%. This result has to be treated carefully, for these are nominal figures and do not take into account real changes. Nevertheless, this is a finding that deserves a lot more scrutiny and investigation. Figure 4: Correlation between Leave Vote Percentage and Change in the Share of Social Benefits Received Per Household to Gross Disposable Income, 1997–2014 15 R² = 0.5613 10 5 So Why Did Britain Leave? 0 -5 -10 10 20 30 40 50 60 70 80 X-axis indicates percentage of leave votes in each NUTS 3 region, y-axis indicates the change in the share of social benefits received per household in gross disposable income of each region over the period 1997–2014. What is important to note is that on an average, the regions that voted leave saw the share of social security transfers increase by nearly 4.84% or 484 basis points over the last 17 years. The importance of social security transfers in determining the outcome of the referendum is seen by a logistic regression. The dependent variable was coded “0” if a region decided to stay in the EU, and “1” if a region decided to leave. The independent variables considered were (i) the number of EU-born residents per 100 UK-born residents, (ii) the CAGR of gross disposable income per head, and (iii) the share of social security transfers per household in disposable income for 2014. The CAGR of income per head was converted into terms of basis points for a more meaningful interpretation of the results. Table 2: Logistic Regression Dependent variable: Whether region displayed majority vote for leave Odds Ratio p-Value Social security share, 2014 EU-born per 100 UK-born CAGR of disposable income per head Constant 1.30358 0.8874 0.001 0.060 1.0118 0.0000974 0.173 0.042 Number of observations: 122; Chi-squared: 54.67. What one notices immediately is that the share of social security transfers seems to be the main indicator in determining whether or not a region displayed a majority vote for leave (Table 2). Regions that had a higher share of social security benefits per household (higher by 100 basis points) displayed a 30% increase in the odds of voting leave; this is highly significant at the 1%. Somewhat Economic & Political Weekly EPW AUGUST 13, 2016 counter-intuitively, those regions that had a higher number of EU-born residents showed a reduction in the odds of voting leave, yet this is not significant at the 5%. What is surprising is that there is no statistically significant relationship between the CAGR of income per head and the decision to vote leave. The regression (Table 2) checked whether regions that had a high level of social security benefits had higher odds of voting leave. The same regression was now run to see whether the growth in the share of social security transfers could have had a role to play in voting leave. (Table 3). Table 3: Logistic Regression Dependent variable: Whether region displayed majority vote for leave Odds Ratio p-Value Difference in social security share, 1997–2014 EU-born per 100 UK-born CAGR of disposable income per head Constant 1.9715 0.000 0.8808 0.036 1.0216 0.037 0.00111 0.050 Number of observations: 122; Chi-squared: 69.51. This regression shows that regions with a higher increase in the share of benefits per household in income over the last 18 years displayed a much higher increase in the chances of voting leave; the odds ratio now registers a 97% increase in favour of voting leave. This also shows strong statistical significance. What is surprising—and more than a little confusing—is to note that now the CAGR of disposable income per head is significant, but in the opposite direction to what one would have originally theorised. An increase in the CAGR of disposable income by one basis point increases vol lI no 33 What is undeniable, though, is the fact that the major determinant of the decision to vote leave amongst the British populace is the share of social security transfers, both as a share in 2014 and in terms of its increase over the last 18 years. Without access to data as to the precise components of social security benefits, one can only hazard a guess. It might possibly be that an increase in social security payments reflects growing dependence of residents on the state owing to a disappearance of well-paying, sustainable employment alternatives. The collapse of manufacturing in the northern regions might have adversely affected several regions, leading to a possible increase in reliance on benefits and income transfers such as Employment Assistance Programmes (EAPs). The entry into the EU has seen no revival of manufacturing, for much of the growth since then has been in services, particularly financial services. It is no wonder then that major financial centres like London have voted remain, simply because they have seen tangible benefits to an EU membership. Avenues for further study include a mapping of the benefits of EU membership, and the regional concentration of incomes and growth since opening up to the EU, in order to better understand the factors causing the leave vote. It must be emphasised that this article does not believe that a spread of benefits could have contributed to making the population less inclined to seek work. The spread of benefits such as EAPs, etc, must be seen as a symptom, not a cause, of economic distress. The increase in transfers might also reflect pensions and old age benefits, in which case the results above can be interpreted as a backlash of older voters in favour of an imagined notion of sovereignty and prosperity. In that case, the role of structural transformation, the loss of manufacturing, etc, and the 41 BREXIT EFFECT reduction in gainful employment cease to be meaningful explanations. How does one explain anti-immigrant rhetoric in this scenario? Just because immigration does not turn out to be significant in a statistical sense in these exercises, it cannot be assumed that it did not play a role. If the former hypothesis is correct, and the increase in social security transfers does indicate a slowing down of employment and livelihood opportunities, then migrants become more visible as an explanation—however wrong—of one’s economic distress. When incomes are falling across the 42 board, domestic populations might turn on migrants and hold them responsible for reduced access to jobs, instead of questioning the objective factors leading to a loss of jobs in the first place.1 The point of this essay, however, is to point to future avenues of research, having identified an important factor determining the leave vote. Discussions so far have pointed to falling incomes, education, and the age of voters as possible explanatory factors. The role of benefits transfers is something that must now be considered. The avenue for future research lies in examining the precise mechanisms by which benefits transfers influenced a nation into making the decision to turn its back on the European ideal. Note 1 I am indebted to Gayatri Nair for making this point clear to me. References Davies, Will (2016): “Thoughts on the Sociology of Brexit,” Political Economy Research Centre, 24 June, http://www.perc.org.uk/project_ posts/thoughts-on-the-sociology-of-brexit/. McGill, Andrew (2016): “Who Voted for the Brexit?,” The Atlantic, 25 June, http://www.theatlantic. com/international/archive/2016/06/brexit-votestatistics-united-kingdom-european-union/ 488780/. AUGUST 13, 2016 vol lI no 33 EPW Economic & Political Weekly