Voting to leave: economic insecurity and the Brexit vote Book or Report Section 

Accepted Version 

Halikiopoulou, D. and Vlandas, T. (2018) Voting to leave: economic insecurity and the Brexit vote. In: Leruth, B., Startin, N. and Usherwood, S. (eds.) The Routledge handbook of Euroscepticism. Routledge Handbooks. Routledge, Abindon, pp. 444­455. ISBN 9781138784741 Available at http://centaur.reading.ac.uk/68822/ 

It is advisable to refer to the publisher’s version if you intend to cite from the work.  See Guidance on citing  .

Publisher: Routledge 

All outputs in CentAUR are protected by Intellectual Property Rights law, including copyright law. Copyright and IPR is retained by the creators or other copyright holders. Terms and conditions for use of this material are defined in the End User Agreement  . 

www.reading.ac.uk/centaur   

CentAUR 

Central Archive at the University of Reading 

Reading’s research outputs online

1

VotingtoLeave:EconomicinsecurityandtheBrexitvote

DaphneHalikiopoulou,UniversityofReadingTimVlandas,UniversityofReadingIntroductionThevictoryoftheBrexitcampintherecentUKreferendumonwhethertostayorleavetheEuropean Union (EU) represents a turning point in the history of European integration.Leavewonwith51.9percentvotingtoLeaveand48.1percentvotingtoRemain.Englandand Wales supported Leave with 53.4 per cent and 52.5 per cent respectively, whileScotlandandNorthern IrelandhadaRemainmajoritywith62percentand55.8percentrespectively. In England, London voted overwhelmingly to Remain but most other areasvotedtoLeaveincludingtheWestMidlands(59percent),EastMidlands(59percent)andtheNorthEast(58percent).LeavemadegainsfrombothLabourandConservativevoters,cuttingacrosspartylines.In many ways this was a surprising result. While the UK is indeed amongst the mostEurosceptic countries in the EU, the consensus expectation was nonetheless for a tightoutcomeinfavourofRemain:Remainwasthechoiceofbothpollstersandbookmakers.Sohowmayweexplainthisresult?ThischapterfocusesonthepoliticaleconomyoftheBrexitvote. It contributes to the debate on British Euroscepticism by examining the impact ofeconomicinsecurityontheBrexitreferendumvote.Drawingonresearchthathasexaminedthe role of economic insecurity and labourmarket institutions on far right party support(Halikiopoulou and Vlandas 2016; Vlandas and Halikiopoulou 2016), we explore thedeterminants of individual support for Brexit. Our focus is on several factors relating toeconomicinsecuritythathaveallreceivedsignificantattentioninbroaderpoliticaleconomydebates including income, occupation, unemployment, poverty and the level and type ofeducation.Ouranalysispaysparticularattentiontothelabourmarketpositionofindividualsandthelabourmarketriskstheyfaceaswewanttoexaminewhether-andifsohow-therisk of being unemployed, the risk of being in poverty and housing risk (whether or notindividuals rent their accommodation) have influenced support for Brexit. We alsoinvestigatetheimpactof incomelevelandsource,ofbeing in lowskilloccupations,whichcompetethemostwithimmigration,andofhavingstudiedahighereducationdegreethatis‘protective’(medicineorlaw).Our findings overall support the economic insecurity thesis. Our results from a logisticregression analysis of the British Election Study suggest thatwhite respondents, those athigherriskofpoverty,belowthemedianincome,withnoformaleducationaremorelikelyto vote for Brexit. Surprisingly, those that self-identify as being at greater risk ofunemploymentandthosethat rent theiraccommodationare less likely tovote forBrexit.Finally,workersinroutineorlowskilloccupations,whichweshowhavebeenmoreexposedto immigration, are more likely to vote for Brexit, while respondents that have studied‘protected’disciplinessuchaslawandmedicinearelesslikelytosupportBrexit.Genderhasno effect but agematters in the anticipatedway: older respondentsweremore likely tosupportBrexit.

2

Our findings are consistentwith the state of the art in the emerging literature onBrexit.Mostexistingstudieshaveemphasisedtheroleofincome,employment,ageandeducation(e.g.Beckeretal2016;GoodwinandHeath2016b).Thesestudieshavealsoexaminedtheprevious voting preferences of Brexit supporters, assessing the extent to which BrexitsupportcorrelateswithapreviousvoteforUKIP(ibid).Overall,accordingtoGoodwinandHeath‘the2016referendumgavefullexpressiontodeeperdividesinBritainthatcutacrossgenerational,educationalandclasslines’(GoodwinandHeath2016a).The chapter is structured as follows. We first provide a theoretical overview thatcontextualises party-based and public Euroscepticism. Here we juxtapose economic andcultural factors and briefly discuss the left-right dimension and nationalism as drivers ofEuroscepticism.WeproceedwithafocusontheeconomyandpublicEuroscepticismintheUK:wetheoriseeconomicinsecurity,discusswhyitcouldpotentiallyleadtoaLeavevote.Inthenextsectionwedescribeourdataanddiscusshowoperationalizekeyfactorsrelatedtoeconomic insecurity and present our results.We proceed by locating our findings in thecontextofother studieson the topic, outlining theprofileof theBrexit voter in termsoftheireconomicandculturalbackground,aswellashis/herattitudesonvarioussocialissues.Thefinalsectionconcludes.Party-basedandPublicEuroscepticismEuroscepticism is a broad umbrella term that encompasses various types and levels ofoppositiontotheEU.Taggart(1998:365)definesEuroscepticismas‘theideaofcontingentorqualifiedopposition,aswellas incorporatingoutrightandunqualifiedoppositiontotheprocessofEuropeanintegration’.Thedifferentlevelsofoppositionareusuallycapturedbythehard/softEuroscepticismtypology.Ontheonehand,theformerreferstoaprincipledoppositiontotheEUandtheideaofEuropeanintegration.ItrejectsthewholeEUprojectand supports withdrawal. Soft Euroscepticism, on the other hand, is not a principledobjection to European integration or EUmembership. Rather, opposition targets specificpolicies.Thereis,therefore,anexpressionofaqualifiedoppositiontotheEU.Withregardstobothpoliticalpartiesandvoters,Euroscepticismtendstobeunderstoodinterms of positions on the left-right dimension (Hooghe et al 2004; Van Elsas and Van deBrug,2015).PartiesopposeEuropeonbothstrategicand ideologicalgrounds. In termsofstrategy,Euroscepticismtendstobeassociatedwithperipheralparties(Taggart1998;Sitter2001; Halikiopoulou et al 2012; De Vries and Edwards 2009) driven by protest andincentivised to oppose Europe as a means of differentiating themselves from theirmainstreamcompetitors. Fringepartiesof both the right and the left are thereforemorelikely tobecriticalof theEUthan theirmainstreamrivals.The2014EuropeanParliamentelection results confirm this to a great extent. The intensification of political integrationwithinthecontextoftheEUeconomicandmigrantcriseshasbeenaccompaniedbytheriseoffarrightand-toa lesserextent- far leftEuroscepticpartiesacrossEurope includingtheFrench Front National (FN), the United Kingdom Independence Party (UKIP), the DanishPeople’sParty(DF),theGreekGoldenDawn(DN)andCoalitionoftheRadicalLeft(SYRIZA).

In terms of ideology, the EU project is underpinned by a rationale that is fundamentallyopposed to the core principles of the far left: a European union that favours neoliberal

3

policies and the freemarketandas such threatens radical left goals (Hoogheet al. 2004:128). Hence, opposition to the EU project originated mainly from far left parties, whichopposedfreemarketeconomics.FarrightEuroscepticismontheotherhandisunderpinnedbynationalism:a justificationofopposition toEU integrationon thebasisof identityandnational sovereignty. FollowingMaastricht and the intensification of political integration,Euroscepticism became increasingly widespread among far right parties whoseprogrammaticagendasemphasisetheneedtosafeguardnationalidentityandculturefromEUerosion.

ItisthereforeconventionalwisdominthefieldthatfarleftandfarrightEuroscepticismaretwo different phenomena: the former is associated with an opposition to neo-liberaleconomicswhilethelatterisseenasaproductofnationalism(DeVriesandEdwards2009).Otherwork,however,hasshownthatEuroscepticpositionsmaybeunderstoodintermsofa cultural cleavage dimension. Halikiopoulou et al (2012), for example, have shown thatboth far left and far rightof Euroscepticismareassociatedwithnationalism: the far rightwithethnicandculturalnationalismandthefarleftwitheconomicandcivicnationalism.

If party based Euroscepticism is driven by strategic and ideological considerations, whatdrivespublicattitudestowardsEuroscepticism?Sørensen(2008)identifies4typesofpublicEU opposition drivers: (1) economic, as money-based calculations are central to one’sevaluation of the EU; (2) sovereignty-based, as people tend to perceive increasing co-operation as a challenge to national sovereignty; (3) democratic, as the intensification ofpolitical integration raisesquestionswith regards to theEU’s so-calleddemocraticdeficit;and (4) social Euroscepticism which refers to a disagreement with the EU’s politicalorientation.Tomakesenseofthesecategories,existingliteraturefocusesonsocietalcleavages,testingpositions on left and right and/or economy and culture, and/or attitudes versus socio-economic voter characteristics. Findings are often conflicting, varying across time andcountry.Marks(2004:239)hasreferredtotheobjectofEuroscepticismasa‘movingtarget’:‘the relationship between left/right orientations and the degree of support for Europeanintegrationdependsonwhenone isasking thequestion’ (2004:239). Increasingly studiesemphasise the importance of a socio-cultural dimension (Van der Brug and Van Spanje,2009;Kriesi et al., 2008;VanElsas andVandeBrug, 2015), linkingEuroscepticism less toleft-right positions (Van der Eijk and Franklin, 2004), and increasingly to public attitudestowards immigrants (De Vries, Hakhverdian and Lancee 2013). However, this does notnecessarilyimplythattheeconomyisirrelevant.Asweshowinthenextsection,therearegoodreasonstoexpecttheeconomytohaveaninfluenceonpublicEuroscepticism.TheBrexitvote:Theroleofeconomicinsecurity‘Why’andwhichUKcitizensvotedagainstEurope?Thischapter focuseson theeconomicdimension of the Leave vote.More specifically, it examines how and towhat extent thisEurosceptic vote may be interpreted in terms of economic insecurity. The connectionbetweeneconomicperformanceandEuroscepticattitudes-mainlyseen intermsofvotingfor Eurosceptic parties - can be understood as either: (1) a rejection of underperformingeconomicpolicyunderstoodasaresultofEUpolicyimplementation;and/or(2)areactionofthelosersofintegration.

4

Economic insecurity and perceived economic risks have been shown tomediate far rightvote (Arzheimer 2009; Halikiopoulou and Vlandas 2016). We conceptualise economicinsecuritybothintermsofunemploymentandriskofunemployment,aswellastheriskofbeing in poverty and the risk of uncertain housing conditions. A number of studies inpoliticaleconomyhavefoundunemploymenttoimpactnegativelyonvariousdimensionsofone’s welfare, such as well-being, life satisfaction, or other metrics (e.g. Jahoda 1988;Gerlach and Stephan1996). The risk of unemployment remains a sourceof insecurity forthosewhoreturntoworkafteraperiodofunemployment(Böckerman2004).Thisinturnisexpectedtoaffectpoliticalpreferencesandisthuslinkedtovotingbehaviour:thosewithouta job or in precarious employment have differing policy preferences from those whocurrentlyhaveapermanentjob(Rueda2005;Rueda2007;Emmenegger2012;Marx2014).Insum,deterioratingeconomicconditionsareexpectedtoimpactnegativelyvoters’socio-economicpositionsand/ortheirexpectations.Theeconomicallyinsecurearemorelikelytosee themselves as the ‘losers’ of European integration- as well as modernization morebroadly.Thisactualorperceiveddeprivation is likely todriveanti-EUpositionseitherasaformofprotestvote,asapunishmentoftheestablishment,oroppositiontofreemovementoflabourandimmigrantaccesstowelfareandjobs.Theeconomically insecureareoftenexpectedtobekeyconstituentsofthepopulistright,because of protest and anti-systemic attitudes, and potential linkages made withunfavourable out-group and authoritarian attitudes (Lubbers and Scheepers 2002: 134).Thereisalsotheissueofcompetitionwithimmigrantsandlabourmarketoutsidersforjobs,welfare,andmorebroadly,foraccesstothecollectivegoodsofthestate(Wimmer1997;deKosteretal2012).ThischaptertestswhetherthesehypothesesmayalsobeappliedtoaEuroscepticvote,andmorespecifically,Brexit.Ouranalysispaysparticularattentiontothelabourmarketpositionofindividualsandthelabourmarketriskstheyfaceaswewanttoexaminewhethertheriskof being unemployed, the risk of being in poverty and housing risk (whether or not anindividualrentstheiraccommodation)haveinfluencedsupportforBrexit.EmpiricalanalysisDataIn order to explore the determinants of the Brexit vote, we rely on the 6th wave of theBritish Election Study1 that asked respondents about their voting intentions for thereferendum before it took place. The question about the referendumwas as follows: “IftherewasareferendumonBritain'smembershipoftheEuropeanUnion’turnout,howdoyouthinkyouwouldvote?”2Wecreateadummyvariablecoded1iftherespondentchoseLeaveand0otherwise(‘don’tknows’arerecodedmissing).

1Fieldhouse,E.,Green,J.,Evans,G.,Schmitt,H.,vanderEijk,C.,Mellon,J.,Prosser,C.(2016).BritishElectionStudy,2015:Face-to-FacePost-ElectionSurvey.UKDataService.SN:7972,http://dx.doi.org/10.5255/UKDA-SN-7972-1.Thesurveyincludes30,027respondents.214,010respondents(46.66%)responded“stayintheEU”,10,357(34.5%)“LeavetheEU”,883(2.9%)saidthey“wouldnotvote”and4,775(15.9%)answered“Don’tKnow”(weightsapplied).

5

Wecreateafemaledummyvariabletocapturethegenderoftherespondentandadummyvariablethatcaptureswhethertherespondentanswered‘white’totheethnicityquestioninthe survey. Tomeasure the impact of age of the respondent,we create two variables: a‘young’dummyvariablecoded1ifrespondentisunder25and0otherwise;andan‘elderly’dummy variable coded 1 if respondent is above 65 and 0 otherwise. As discussed in theprevious section, young people are typically less Eurosceptic and the elderly are typicallymoreEurosceptic.Thesurveyalsoasksrespondentsabouttheirgrosshouseholdincome.Ratherthanincludingallpossiblecategoriesfromunder5,000toabove150,000poundsperyear,weuseacut-offpoint. Inmostpoliticaleconomymodels, thecrucial factor is thepositionofan individualrelativetothemedian incomesowecreateavariable ‘abovemedian income’.GiventhatthemedianhouseholddisposableincomeintheUKfor2014/15was£25,600,3wecodeallrespondentswithhouseholdincomeabove£25,000asbeingabovemedian.Inaddition,welookattheroleofeducation,which ismeanttoconstituteacleardividinglinebetweenthelosersandwinnersofEuropeanintegration.Theabsolutelosersarelikelytobethosewithnoeducationsowecreateadummythattakesvalue1ifrespondentshaveno formal qualification. Among those that have undertaken higher education, the surveyasks themwhat subject they have studied. This provides an opportunity to testwhetherrespondentsthathavestudieddifferentsubjectsfeeldifferentlyaboutBrexit.Inparticular,wehypothesise that respondents thatstudiedmedicineor lawaremore likely towant toremainbecause these subjectareas lead to jobs thatmaybemore ‘closedoff’ to foreigncompetitionthanjobsbuildingoneducationinthehumanities,naturalandsocialsciences.Next, we want to test the effect of various risks on support for leaving. The first risk iswhethertherespondentisathighriskofpoverty,understoodasbeing“fairlylikely”or“verylikely” (coded 1, 0 otherwise) to have experienced timeswhen the respondent does nothave enoughmoney to cover day to day living costs during the next 12months. To theextentthatthosethatareatgreaterriskofpovertyaremoredissatisfiedwiththesystem,theymaybemorelikelytovoteforBrexit.Onthecontrary,iftheybelieveforecastsabouttheeconomytheyshouldbelesslikelytovotetoleave.Thesecondriskthatwewishtocaptureisunemploymentrisk.Wecreateadummyvariablethattakesavalue1iftherespondentanswersthattheyarefairlyorverylikelytobeoutofwork in the next 12 months, and 0 otherwise. Again, one could expect that moreeconomicallyinsecurerespondentswouldbemoreEuroscepticandhencebemorelikelytosupport Brexit, or if these respondents aremore concerned about the effect that Brexitwouldhaveontheeconomyandinturnontheriskofbecomingunemployed,theymaybelesslikelytosupportBrexitgiventheirgreaterexposuretounemployment.The third risk concerns housing insecurity. We create a variable that is coded 1 ifrespondents are renting and 0 otherwise. Here we expect individuals that experience3Accessedat:http://www.ons.gov.uk/peoplepopulationandcommunity/personalandhouseholdfinances/incomeandwealth/bulletins/nowcastinghouseholdincomeintheuk/2015-10-28

6

greater housing insecurity to be more supportive of Brexit both because they are moreeconomically insecureandbecausetheymight linkEuropean integrationwith immigrationand‘pressures’onhousing(notethatthisdoesnotrequirethelinktobecorrect,merelytherespondenttoholdsuchviews).Thefourthriskvariablecapturesincomeinsecurityandrelianceonstatebenefits.Wecreatea variables that takes value 1 if the main income of the respondent comes fromemploymentand0 if it comes fromother sources includingbenefits. This variablewill benegativelyrelatedtoBrexitifgreaterincomesecurityandlowerrelianceonbenefitslowerssupportforBrexit.Finally,we investigate the impactofdifferent labourmarketoccupations.Thesurveyasksrespondentswhether theyare ‘employers in largeorganisationsand inhighermanagerialpositions’, ‘higher professional occupations’, ‘lower professional and managerialoccupations and higher supervisory’, ‘employers in small organisations and own accountworkers’, ‘intermediateoccupations’, ‘lowersupervisoryandtechnicaloccupations’, ‘semi-routineoccupationsor routineoccupations’. I create adummyvariable if the respondentchoosesoneof the last threeoccupationsbecauseall threeare routineand/or lowskills.Workerswith lowerandmoreroutineskillsaremore insecureas theycanbemoreeasilyreplaced by employers and also tend to have lower income potential and higherunemploymentrisk.Inaddition,theywerethemostlikelytocompeteinthelabourmarketwithlowskillimmigrants.ResultsfromLogisticanalysisononeindependentvariableatatimeTable 1 reports the results for a series of models where each independent variable isincludedonitsown.Foreaseofinterpretationitreportsthepredictedprobabilityofvotingleavewhenthedummyvariableis1andwhenitis0.Wherethe95%confidenceintervalofthepredictedprobabilityfordifferentvaluesofthevariableoverlap(lasttwocolumns)thevariabledoesnothaveastatisticallysignificanteffect.The predicted probabilities for male and female respondents are very similar and notstatistically different: men are therefore not more likely to support Brexit than women.Whiterespondentshavea40%probabilityofpreferringtoleavecomparedto31%fornon-white respondents (all probabilities in the discussion are rounded), elderly respondentshavea50%predictedprobabilityofpreferring leavecomparedtonon-elderlywhereasfortheyoungit is18%(comparedto42%fornon-youngrespondents).This isconsistentwithpreviousfindingsthatshowolderwhite individualsasbeingmoresupportiveofpro-BrexitpartiessuchasUKIP(GoodwinandHeath2016b).Incontrasttoaccountsthatdownplaytheroleofmaterialfactors,wefindthattheymattersignificantly.First,respondentsabovethemedianincomehavea33%predictedprobabilityof supportingBrexit compared to47% for thosebelowmedian.Havinga lower income isassociatedwith greater support for Brexit. Second, thosewith no formal education havepredictedprobabilitiesthatarealmosttwiceaslarge(65%)asthosewithformaleducation(38%).ThisconfirmsthelinkbetweenloweducationandEuroscepticism.Third,theeffectofthedisciplineamongthosewithhighereducationqualificationsisstriking:respondentsthat

7

have studied lawandmedicine,disciplines that lead to jobswhichmaybemore shieldedfrom immigration, have a 29% predicted probability compared to 40% for otherrespondentsofsupportingBrexit.Fourth,individualsinloworroutineoccupationsalsohavemuchhigher predictedprobabilities (55% versus 37%) to vote Leave. Being in a low- skilloccupation is indeed associatedwith greater support for Brexit. Finally, respondents thatderive their main earnings from employment have lower predicted probabilities ofsupporting Brexit (35% versus 41%).While the risk of poverty does seem tomatter, theeffect is small (39% versus 42%). More surprisingly, a greater risk of unemployment isactually associated with a lower support for Brexit: respondents facing high risks ofunemployment have a lower predicted probability (36%) of choosing Leave compared toother respondents (41%). This is consistent with the notion that those facing highunemploymentrisksmayhavebeenmorereceptiveofRemainargumentsemphasisingthepotentialadverseimpactofBrexitontheeconomy.Similarly,rentingone’saccommodationhas the opposite effect to the one we would expect: respondents who rent theiraccommodationhave lowerpredictedprobabilityofsupporting leave(35%versus41%). Itmay be that respondents who rent aremore concerned about the detrimental effect ofBrexitor thatthis iscapturingconfoundingfactorssuchasage. Inordertobecertain,weneedtotestforallfactorssimultaneously.Table1:Logisticregressionanalysisofoneindependentvariableatatime

Variable RespondentPredictedprobability

95% confidenceinterval

Gender Male 39.78% 38.95% 40.60%

Female 39.80% 38.94% 40.66%

Ethnicity Non-white 30.94% 28.55% 33.32%

White 40.32% 39.70% 40.93%

Elderly(65+) Notelderly 36.76% 36.09% 37.43%

Elderly 50.08% 48.79% 51.35%

Young(18-24) Notyoung 42.21% 41.57% 42.84% Young 17.79% 16.24% 19.34%Income Notabovemedian 46.58% 45.50% 47.66%

Abovemedian 33.18% 32.31% 34.04%

Riskofpoverty Nothigh 38.93% 38.24% 39.61%

High 42.43% 41.21% 43.64%

Riskofunemployment Nothigh 40.51% 39.86% 41.16% High 35.82% 34.32% 37.30%Education Formaleducation 37.93% 37.32% 38.55%

Noformaleducation 64.75% 62.54% 66.97%

Highereducationdiscipline Notprotected 40.03% 39.43% 40.63% ‘Protected’discipline 28.72% 24.96% 32.48%Housing Notrent 40.86% 40.19% 41.52%

Rent 35.05% 33.70% 36.41%

Earningsfromemployment No 40.55% 39.91% 41.18%

Yes 3.46% 32.97% 36.22%

8

Occupations Others 37.49% 36.86% 38.13%

Lower/routineoccupations 54.89% 53.22% 56.56%

Note:eachrowreports thepredictedprobabilityofadichotomousvariable for its2valuesusingalogistic analysis regressing a dichotomous variable ‘voting leave’ on that variable with robuststandarderrors.ResultsfromlogisticanalysisonseveralindependentvariablesFigure1reportstheresultsofdifferentmodelsthattesttheeffectsofseveralfactorsjointly.Itplotssemi-standardisedcoefficients–i.e.coefficientshavebeenrescaledbythestandarddeviation of the variable in the data. A positive coefficient suggests the factor underconsideration increases the probability of the respondent supporting Brexit. For eachvariable, the figure also displays the 95% confidence interval,which is shownby the linearound thepointestimate:when it intersects the0-line, the coefficient isnot statisticallydifferentfrom0andweconcludethatthefactorisnotassociatedwithsupportforBrexit.

Aswasthecasebefore,genderisnotstatisticallysignificant,butthisisnowalsothecaseforthevariablecapturingwhethertherespondentderiveshis/herearningsfromemployment.Theyoungaremuch less likely tosupportBrexitwhile theopposite is true for theelderly(relativetomiddleagedrespondents).Theotherresultsarethesameasbeforebutrunningthelogisticregressionwithalltheindependentvariablestogetherallowsustocomparethemagnitudeoftheeffects.Thisrevealsthatageisastrongpredictorofvotingleaveandthattheeffectisreducedasmorevariablesareincluded.Next,beingbelowthemedianincome,havingno formaleducation,andbeing in lowor routineoccupationsalsohavevery largesignificanteffectsonvotingleave.These findings are consistent with the fact that immigration from EU countries wasparticularly acute for certain occupations. Thus for instance, ONS data4 suggest thatbetween 2010 and 2014, 164,000 immigrants with a professional or managerialoccupational background immigrated to the UK from the EU, while 169,000 immigratedfromoutside theEU. Inotherwords, theseoccupationsareunder similar ‘pressure’ fromnon-EUandEUcountries.Bycontrast,amuchlarger-277,000–numberofimmigrantswithamanualandclericaloccupationalbackgroundcamefromEU,whereasonly64,000camefromnon-EUcountries.Asa result,differences in support forBrexitbetweenoccupationsmayreflectdifferentdegreesofexposuretoimmigrationfromtheEU.Onewaytovisualisethepracticalimplicationsoftheseresultsistocalculatethepredictedprobabilitiesofvotingleavefortwohypotheticalindividuals.Awhitemalebetween25and65yearsold,withnoformaleducation,derivingmostofhisearningsfromemploymentinalowerorroutineoccupation,rentinghishouse,andreportingahighriskofunemploymentand poverty would have a 66% probability of supporting Brexit. Contrast this with anindividualthathasthesamecharacteristicsinallrespectsexceptthathehasatleastsomeformaleducation,isnotathighriskofpovertyorunemployment,norinalowerorroutine

4Long-TermInternationalMigration,estimatesfromtheInternationalPassengerSurvey-Annualdata.Source:OfficeforNationalStatistics(ONS).

9

occupation,andhasstudieda‘protecteddiscipline’:thisindividualwouldhaveapredictedprobabilityofonly25%ofsupportingBrexit.Figure1:materialdeterminantsofvotingleave

Note:Robuststandarderrorsclusteredbycountry.Effectsarerescaledbythestandarddeviationsofthepredictors.DiscussionOur findingswith regards toeconomic insecurity are consistentwith the findingsofmostexistingstudiesonBrexit.Focusingonthesocio-economiccharacteristicsoftheBrexitvote,Becker et al (2016) emphasise the importance of variables “not malleable to politicalchoices”, including education, industry structure and demography (2016:2). They furtherarguethat“policychoicesrelatedtopressurefromimmigration,fiscalcutsandthehousingmarket are linked to a higher Vote Leave share especially when socio-economicfundamentalsare‘weak’(lowincomes,highunemployment),andwhenthelocalpopulationis lessabletoadapttoadverseshocks(dueto lowqualificationsandarisingageprofile)”(Becker et al 2016: 4). Goodwin and Heath (2016b) also focus on economic inequality,drawingonananalysisofthe‘leftbehindthesis’.Throughanexaminationofbothindividualand area-level data they find that “the poorest households, with incomes of less than£20,000per year, aswell as the unemployed, low-skilled andmanualworkers, andmorebroadly economically deprived groups vulnerable to povertyweremore likely to supportBrexit”(GoodwinandHeath2016conline).Studiesalsoagreethateducationwasastrongpredictor(GoodwinandHeath2016b;Kaufmann2016;Beckeretal2016):supportforBrexitwasmuchhigher in areaswhere largenumbers of peopledonot hold anyqualifications.Finally, another strong predictor was age, suggesting that- in line with the ‘left behind’thesis andprevious analysesonUKIP support- there is a linkbetween Leave support andsocietalmarginalizationespeciallyamongpensionersandoldervoters.Beckeretal(2016)as

10

wellasGoodwinandHeath(2016)alsofindalinkbetweensupportforBrexitandavoteforUKIP: ‘Public support for Leave closely mapped past support for UKIP but was morepolarisedalongeducationlinesthansupportforUKIP’(GoodwinandHeath2016b:1).OurfindingsareinlinewithBeckeretal’s(2016)argumentthattheBrexitvote‘wasatleastpartiallyrelatedtodistributional issues’ (Beckeretal2016:6)andthatthere isaclear linkbetweenBrexitandausterity:‘justaslightlylessharshregimeofausterityaimedatcuttingbenefits could have substantially reduced support for the Vote Leave campaign andoverturnedtheresultoftheEUreferendum’(Beckeretal2016:4).Thisisalsoinlinewithpreviousworkthatexaminestheimpactofeconomicinsecurityonfarrightpartysupport.Analyses focusing on the mediating role of welfare state institutions- for exampleunemploymentbenefitsandEmploymentProtectionLegislation(EPL)-havefoundthatthefarrightismorelikelytoexperienceanincreaseinsupportincaseswheretheseinstitutionsareweakanddonotcontaineconomicinsecurity(HalikiopoulouandVlandas2016;VlandasandHalikiopoulou2016).Theadoptionofausteritypoliciesthatincreasetherisksandcostsofunemploymentandexacerbateeconomic insecurity thereforemakes theriseof the farrightmorelikely.In terms of the relationship between immigration and the Brexit vote, findings aremoremixed. While Goodwin and Heath (2016c) did not find a positive relationship betweensupport for Brexit and immigration levels, they did find a positive relationship betweenBrexit and change in immigration levels- in other words, in areas where the size of EUmigrantpopulationsincreasedrapidlyweremorelikelytosupportBrexit.Areas,therefore,“that had experienced a sudden influx of EUmigrants over the last 10 yearswere oftenmorepro-leave”(GoodwinandHeath2016c).ThiscoincidessomewhatwithBeckeretal’s(2016) finding that in termsofmigration growth, onlymigration from themainly EasternEuropeanEUaccessioncountriespositivelycorrelateswiththeVoteLeaveshare(Becketetal2016:23).Beyondthishowever,“relativelylittlevariationintheVoteLeavesharecanbeexplainedbymeasuresofalocalauthorityarea’sexposuretotheEuropeanUnion(e.g.,duetoimmigration”(2016:3).Butimmigrationwasn’tthekeydriverofBrexitaccordingtothisstudy.Evenmajorchangesinimmigrationwouldhavebeenunlikelytoswaythevote.Thereferendum result, they argue instead, “was driven by long-standing fundamentaldeterminants,most importantly those thatmake it harder to dealwith the challenges ofeconomicandsocialchange”(Becketet2016:13).Ouranalysistherefore,aswellasotherinitialstudiesonBrexit,pointtotheimportanceoftheeconomicdimension.Themore likelyBrexitvotersarethe ‘leftbehind’ (GoodwinandHeath 2016) frommodernization and globalization, older, from poorer households, withlowereducationlevels,whoperceiveimmigrantsasathreateithertotheculturalwayoflifeortheireconomicwellbeingandbelievetheyarecompetingwiththemfor jobs,accesstosocial services andmore broadly the collective goods of the state. There is an additionaldimensiontothis,whichreferstoattitudesandvalues.Thekeytothisissovereignty-basedEuroscepticism, nationalism and conservative social values. There is a strong correlationbetweenvotingtoLeaveforexampleandfeeling‘verystrongly’English(GoodwinandHeath2016c). Kaufmann (2016) also finds a correlation between Brexit support and sociallyconservative attitudes: those who support authoritarian positions such as harsh prison

11

sentences and the death penalty and oppose equal gender opportunities aremore likelyBrexitsupporters.ConclusionThischapterhasfocusedonthepoliticaleconomyoftheBrexitvote,andmorespecifically,ontheextenttowhicheconomicinsecurityplayedaroleinthevotetoLeavetheEU.Thelinkbetween theeconomyandEuroscepticism is often theorised in termsof thewinnersand losers of European integration. In other words, a vote against the EU is oftenunderstoodasareactionofthelosersofintegration.Wetheorisethisintermsofeconomicinsecurity,inlinewithstudiesthathaveattemptedtotesttheeffectofeconomicinsecurityon far right party support (Halikiopoulou and Vlandas 2016; Vlandas and Halikiopoulou2016).OurexaminationofBESdatahasshownthatlabourmarketfactorshaveaprofoundeffectonsupportforBrexit:lowincomerespondentswithnoformaleducationandinlowskilledorroutineoccupationsthatareatriskofpovertyandhavefacedthebruntofimmigrationfromtheEUareespeciallylikelytosupportBrexit.Ageandethnicitywereimportant,whilegenderdidnot play a role. This confirms thehypothesis that economic insecuritywas animportantdriverofBrexitand is in linewithstudies that findthat there isa linkbetweenausteritypoliciesandBrexitsupport.Our findingsareconsistentwithexisting literatureonBrexit todate.Anumberof studiessincetheUKEUreferendumhasfoundsimilarfindingsrelatingtotheeconomy:Beckeretal(2016)andGoodwinandHeath(2016a;2016b;2016c)agreethattheleftbehind,older,lesseducatedvotersinmanualemploymentaremorelikelyBrexitsupporters.ForBeckeretal(2016) this is because long-term conditions exacerbate one’s socio-economic standing,preventing opportunities. Austerity in the UK exacerbated the effect of economicdeprivation,thusincreasingthelikelihoodofaBrexitvote.We know, therefore, that the typical Brexit voter was more likely to be- but wasn’texclusively- older, low skilled, in manual employment and socially and economicallymarginalised.ThisvotercouldhavepreviouslybeenaUKIPsupporter,andalsowasmorelikely to oppose immigration, feel strongly about his/her English identity and adoptconservative – even authoritarian- values about social issues. The question, to a greatextent,ishowtointerpretthisdata.Studiesdifferintermsofwhichfactorstheyplacetheiremphasis on, suggesting these characteristics were important either because of theeconomic insecurities that these voters were facing per se, or because of a breach of abroader values consensus at the helm of which lies identity. In this latter view, socio-economicfactorsinthemselvesarelessimportantthanattitudinalfactors.Futureresearchcould focusmoreonthecausal linksanddeterminewhyandhowtheeconomic,cultural,sovereignty-baseddeterminantsofEuroscepticismaffectvotingbehaviour.References

12

Arzheimer,K.(2009),‘ContextualFactorsandtheExtremeRightVoteinWesternEurope’,1980–2002.AmericanJournalofPoliticalScience,53,259–275.doi:10.1111/j.1540-5907.2009.00369.xBecker,S,O.,Fetzer,TandNovy,D(2016),WhoVotedforBrexit?AComprehensiveDistrict-LevelAnalysis,UniversityofWarwick,WorkingpaperSeries,No.305BöckermanP.(2004)PerceptionofjobinstabilityinEurope,SocialIndicatorsResearch,67(3),283-314.DeKoster,W.P.AchterbergandJ.vanderWaal(2012)Thenewrightandthewelfarestate:TheelectoralrelevanceofwelfarechauvinismandwelfarepopulismintheNetherlands,InternationalPoliticalScienceReview,34(1):3–20.DeVries,C.E.&Edwards,E.E.(2009).TakingEuropetoitsextremes:ExtremistpartiesandpublicEuroscepticism.PartyPolitics15(1):5–28.DeVries,CE,Hakhverdian,AandLancee,B(2013)Thedynamicsofvoters’left/rightidentification:Theroleofeconomicandculturalattitudes,socialScienceResearchandMethods,1:2,pp.223-238EmmeneggerP.(2012)JobsecurityregulationsinWesterndemocracies:afuzzy-setanalysis,EuropeanJournalofPoliticalResearch,50(3),336-364.GerlachK.andStephanG.(1996)ApaperonunhappinessandunemploymentinGermany,EconomicsLetters,52(3),325-330.Goodwin,M.andHeath,O.,(2016a)‘Ataleoftwocountries:Brexitandthe‘leftbehind’thesis’,LSEBritishPoliticsandPolicyBlog,http://blogs.lse.ac.uk/politicsandpolicy/brexit-and-the-left-behind-thesis/[accessedNovember4th2016]Goodwin,M.J.andHeath,O.(2016b),The2016Referendum,BrexitandtheLeftBehind:AnAggregate-levelAnalysisoftheResult.ThePoliticalQuarterly,87:323–332.doi:10.1111/1467-923X.12285Goodwin,M.JandHeath,O.(2016c),Brexitvoteexplained:poverty,lowskillsandlackofopportunities,JosephRowntreeFoundation,https://www.jrf.org.uk/report/brexit-vote-explained-poverty-low-skills-and-lack-opportunities[accessedDecember5th2016)Halikiopoulou,D,Nanou,K,andVasilopoulou,S(2012)Theparadoxofnationalism:Thecommondenominatorofradicalleftandradicalrighteuroscepticism.EuropeanJournalofPoliticalResearch51(4):504-539.

HalikiopoulouDandVlandasT(2016)‘Risks,costsandlabourmarkets:explainingcross-nationalpatternsoffarrightpartysuccessinEuropeanParliamentelections’.JournalofCommonMarketStudies,doi:10.1111/jcms.12310.

13

Hooghe,L.,Marks,G.&Wilson,C.(2004).Doesleft–rightstructurepartypositionsonEuropeanintegration?InG.Marks&M.Steenbergen(eds),Europeanintegrationandpoliticalconflict.Cambridge:CambridgeUniversityPress.

JahodaM.(1988)Economicrecessionandmentalhealth:someconceptualissues,JournalofSocialIssues,44(4),13-23.

Kaufmann,E.,(2016)‘It’sNOTtheeconomystupid:brexitasastoryofpersonalvalues’,LSEBritishPoliticsandpolicyBlog,July7th2016,http://blogs.lse.ac.uk/politicsandpolicy/personal-values-brexit-vote/[accessedNovember4th2016]Lubbers,M.andScheepers,P.(2002)FrenchFrontNationalVoting:amicroandmacroperspective,EthnicandRacialStudies,25(1):120-149.Marks,G(2004)Conclusion:Europeanintegrationandpoliticalconflict.InGMarksandMRSteenbergen(eds),Europeanintegrationandpoliticalconflict.Cambridge:CambridgeUniversityPress,pp.235-259.MarxP.(2014)Labourmarketrisksandpoliticalpreferences:thecaseoftemporaryemployment,EuropeanJournalofPoliticalResearch,53(1),136–159.McLaren,L(2007)Explainingmass-leveleuroscepticism:Identity,interests,andinstitutionaldistrust.ActaPolitica42:233-251.RuedaD.(2005)Insider-outsiderpoliticsinindustrializeddemocracies:thechallengetosocialdemocraticparties,AmericanPoliticalScienceReview,99(1),61-74.RuedaD.(2007)Socialdemocracyinsideout:partisanshipandlabormarketpolicyinindustrializeddemocracies,Oxford,OxfordUniversityPress.Sitter,N.(2001).ThepoliticsofoppositionandEuropeanintegrationinScandinavia:Iseuro-scepticismagovernment–oppositiondynamic?WestEuropeanPolitics24(4):22–39.Sørensen,C(2008)Loveme,lovemenot:AtypologyofpublicEuroscepticism.EuropeanPartiesElectionsandReferendumsNetwork(EPERN)workingpaper,No.19.

Spiering,M(2004).BritishEuroscepticism.EuropeanStudies20(2004):127-149.

Taggart,P.(1998),ATouchstoneofDissent:EuroscepticisminContemporaryWesternEuropeanPartySystems.EuropeanJournalofPoliticalResearch,33:363–388.doi:10.1111/1475-6765.00387VanderBrug,WandVanSpanje,J(2009)Immigration,Europeandthe‘new’culturaldimension.EuropeanJournalofPoliticalResearch48:309–334

14

VanElsas,E.andVandeBrug,W,.(2015)Thechangingrelationshipbetweenleft–rightideologyandeuroscepticism,1973–2010,EuropeanUnionPolitics,16:2,pp.194-215Vasilopoulou,S,.(2016)UKEuroscepticismandtheBrexitReferendum,PoliticalQuarterly,87:2,pp.219-227

Vlandas,T.andHalikiopoulou,D.,(2016)Whyfarrightpartiesdowellattimesofcrisis:theroleoflabourmarketinstitutions.WorkingPaper.EuropeanTradeUnionInstituteISSN1994-4446

Wimmer,A.(1997).Explainingxenophobiaandracism:acriticalreviewofcurrentresearchapproaches,EthnicandRacialStudies,20(1):17-41.