Research ArticlePredicting the Rise of EU Right-Wing Populism in Response toUnbalanced Immigration

Boris Podobnik12345 Marko Jusup6 Dejan Kovac789 and H E Stanley1

1Boston University Boston MA 02215 USA2Faculty of Information Studies SI-8000 Novo Mesto Slovenia3University of Rijeka 51000 Rijeka Croatia4Zagreb School of Economics and Management 10 000 Zagreb Croatia5Luxembourg School of Business 2453 Luxembourg Luxembourg6Center of Mathematics for Social Creativity Hokkaido University Sapporo 060-0812 Japan7CERGE-EI Praha 1 11000 Nove Mesto Czech Republic8Woodrow Wilson School of Public and International Affairs Princeton University Princeton NJ 08544 USA9Adriatic Economic Association 10000 Zagreb Croatia

Correspondence should be addressed to Boris Podobnik bpphyhr

Received 19 February 2017 Revised 15 May 2017 Accepted 15 June 2017 Published 7 August 2017

Academic Editor Fabio Caccioli

Copyright copy 2017 Boris Podobnik et alThis is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Among the central tenets of globalization is the free migration of labor Although much has been written about the benefitsof globalization little is known about its limitations and how antiglobalist sentiment can be strongly affected by high levels ofimmigration Analyzing poll data from a group of EU countries affected by the recent migrant crisis we find that over the lastthree years the percentage of right-wing (RW) populist voters in a given country depends on the prevalence of immigrants in thiscountryrsquos population and the total immigration inflow into the entire EU The latter is likely due to the perception that the EUfunctions as a supranational state in which a lack of inner borders means that ldquosomeone elsersquos problemrdquo can easily become ldquomyproblemrdquo We find that the increase in the percentage of RW voters substantially surpasses the percentage of immigration inflowimplying that if this process continues ongoing democratic processes will cause RWpopulism to prevail and globalization to rapidlydecrease We locate tipping points between the fraction of immigrants and the rise of RW populism and we model our empiricalfindings using a complex network framework in which the success of globalization rests on a balance between immigration andimmigrant integration

1 Introduction

An important goal of globalization is to allow both capital andlabor to move freely across national borders [1ndash4] Althoughone EU country that lacks a sufficient labor force can drawfrom another EU country that has an overabundance oflabor this economic consideration neglects how movementof people can affect public opinion and alter the outcomeof subsequent elections A volatile situation can arise wheneither the native majority or the migrant minority sense thattheir national ethnic or religious identity is being threatenedThus the unprecedented inflow of immigrants into the EUduring the recent migrant crisis allows fresh insights into

the relationship between immigration and the popular voteandmdashby extensionmdashthe factors that directly affect the successof further globalization

Although a large body of literature is dedicated to theanalysis of how migration affects the global economy [1ndash3 5ndash10] and right-wing (RW) populism [11ndash17] much less isknown about the limitations of globalization [4 18] especiallyhow large-scale migrations sway the popular vote and whatthe economic consequences may be For example analyzingnational elections in 16 European countries from 1981 to1998 Swank and Betz reported that the welfare state directlydepresses RWpopulism [9] Smith reported that RWpopulistparties benefit from higher levels of crime by linking crime

HindawiComplexityVolume 2017 Article ID 1580526 12 pageshttpsdoiorg10115520171580526

2 Complexity

with higher levels of immigration [16] Borjas reported thatthe integration of immigrants into the USwas slow and that ittook four generations for the earnings of immigrants to equalthe earnings of natives not one or two as commonly believed[19]

Although the recent migrant crisis in the EU was causedby political turmoil and armed confrontation not global-ization some of the central tenets of globalization havenevertheless been tested The surge in the population ofimmigrants has beenmatched by a surge in voters supportingRW populist parties The growing number of RW votersacross the EU suggests that tolerance towards immigrantsis conditional [20] and that members of the general publicpreviously not identifying with RW populism have becomesupporters If globalization is to succeed wemust understandthis ldquochange of heartrdquoWhen immigration is more rapid thanintegration over a prolonged period of time [20] RWpopulistmovements canwin elections and this can cause globalizationto decline Apart from concerns about globalization expectedglobal climate change strongly indicates possible massivedisplacements in the global population [21]

Why immigration in the past decades did not stir asmuch RW populism across the EU as the most recent inflowdid is perhaps because the EU societies are approaching atipping point characterized by drastic political and economicupheaval The term tipping point is used (i) to denote apossible change during which RW populism becomes theruling political option and (ii) to emphasize a potentiallysudden (nonlinear) aspect of such a change thatmay not evenbe visible in the available data It may not be globalization andimmigration themselves that are the root of the problem butrather the high speed at which these processes are occurringright now Herein we find that during the last three yearsthe rise of RW populism in a set of EU countries has beensubstantially more rapid than the immigrant inflow into theEU We report that the percentage of RW voters in a countrysignificantly depends on the percentage of immigrants withinthis country and the total immigration inflow into the entireEU We also find that the occurrence of violent incidentsis unrelated to the rate of immigrant inflow and has littleeffect on the rising percentage of RW voters We identifytipping points connecting the percentage of immigrantsand the popularity of RW populist parties Globalizationis meant to be a tolerant form of democracy in whichcooperation between nations supersedes individual nationalinterests but there are circumstances under which growingRW populist movements can overthrow this tolerant formof democracy We analyze whether the interconnectednessbetween countries characteristic of globalization can facilitatecascades (ie domino effects) in which growing populistmovements within one country trigger similar movements inother countries

2 Results

It seems reasonable that the causes of populist behavior areat least partly nation-dependent Generally when a society isrelatively far from a tipping point the nationrsquos interests likelyspan a multidimensional space For example according to

Eurobarometer 65 published in 2006 the main concerns ofEuropean citizens were unemployment (49) crime (24)economic health (23) immigration (14) and terrorism(10) In a concurrent survey in the UK however 38 ofthe respondents listed race and immigration as the top issueBecause RW populism in the UK eventually led to a winover the BREXIT vote previous surveys suggest an intriguingpossibility that as a society approaches a tipping point itsmultidimensional space starts to shrink finally reducing toa one-dimensional space in which a single issue dominatesthe current affairs Inspired by the developments in the UKwe analyze the rise of RW populism in the EU wherein therecent migrant crisis suggests that the Union is approachinga tipping point

RW populism often embraces intolerance which is awidespread social phenomenon that produces conflicts andgenerates segregation [22ndash28] Intolerance combined withradicalization is the main cause of violence and terrorism[29ndash33] However RW populism often shares certain valuesfor example antiglobalization protectionism and Euroscep-ticism with left wing (LW) populism In principle peoplenot satisfied by the current government can swing betweenRW and LW populism Such a cyclic dynamics may arise asthe prevailing issues (eg economic versus ethnic) changeat a faster-than-generational time scale Here we focus onlyon RW populism and the fraction of RW populist voters inresponse to unbalanced immigration as the driving factor

For each country affected by the recent migrant crisiswe calculate the percentage of immigrants from September2013 to September 2016 by combining the official value for2013 with the number of visa applicants recorded monthly[34 35] We then collect the available election poll dataand election results for the same range of months [36]Figure 1(a) shows that in June 2016 in this set of EUcountries there is a rising trend in the percentage of RWpopulist supporters in response to the increasing percentageof immigrants in the general population We calculated thebest linear fit that can be interpreted as the cumulativeprobability function of a uniform distribution From this fitwe estimate that when the fraction of immigrants reachesapproximately 30 RW populism attains the majority Theslope value of 180 is highly statistically significant (119905-statistic= 262) Besides the linear fit we relate the fractions ofimmigrants and RW votes using the cumulative exponentialprobability function thus finding that as the percentage ofimmigrants approaches approximately 22 the percentage ofRWpopulist voters exceeds 50 which is again the thresholdat which in democratic societies a party can take over thegovernment The considerable scattering of the data suggeststhat tolerance levels may differ among countries and thelower the percentage of immigrants needed to trigger highlevels of RW populism in a country the lower the level oftolerance in that country Figure 1(b) shows for examplethat in Austria the 50 threshold is reached even when thepercentage of immigrants is below 20

The current fraction of immigrants in the general popula-tion is not the only factor that affects voter sentiment Figure 2uses the data for Austria and Germany for 2013ndash2016 andshows that the growing percentage of RWpopulist supporters

Complexity 3

40

30

10

20

20 25

AUT

FRA

NL

NOR

SWE

UK

50

0

0 5

FINGRC

10

BEL

DEU

ITA15

R

W v

oter

s

DK

immigrants

(a)

40

30

20

50

16

immigrants17 18 191514

R

W v

oter

s

Austria

(b)

Figure 1 Immigration affects the support for right-wing populism I (a) Among the EU countries involved in the recent migrant crisis supportfor RW populism is generally higher in those countries that accepted a larger number of immigrants relative to the countryrsquos populationsize Shown is June 2016 Seeing democracy as the majority rule principle we presume that RW populism becomes a dominant politicaloption when the percentage of RW voters exceeds 50 Judging based on a cumulative exponential function that fits the data reasonablywellmdash119910 = 443 exp(0097119909)mdashRW populism in the examined EU countries may take over if the percentage of immigrants in the totalpopulation approaches 22 Shown is also a linear fit 119910 = minus367 + 180119909 roughly giving that the RW populism reaches the majority forthe percentage of immigrants equal to approximately 30 Coefficients of determination (1198772) for the two models are 03866 and 04077respectively Akaike weights indicate that given the dataset at hand the exponential model is better with probability 443 while the linearmodel is better with the remaining probability 557 (b) Similar as in the other EU countries Austrian data reveal that the increase inthe percentage of immigrants is accompanied with an increase in the percentage of RW populist voters Here too a cumulative exponentialfunction fits the data well This function predicts the rise of RW populism in Austria when the percentage of immigrants is slightly below the20 mark

is responding to the inflow of immigrants For example inAustria the far-right party won 205 of the popular votein 2013 in response to the percentage of immigrants livingin Austria at the time but in the second half of 2015 theinflow of immigrants increased sharply [34 35] and a localelection in Vienna saw the percentage of RW votes jump to33 This sudden change suggests the presence of a phasetransition tipping point or critical point [37 38] that is thenearer the percentage of immigrants comes to the tippingpoint the more quickly voters turn to extreme politicalalternatives Figures 2(b) and 2(c) show a qualitatively similarphenomenon occurring in Germany

In an attempt to probe deeper into the internal dynamicsof RW populism in the EU as a function of the inflow ofimmigrants next we analyze how the immigration rate affectsthe rise in RW populist voters In Figure 3 we qualitativelyrepresent the annualized increase in RW votes by taking thedifferences between the popularity of RW populist partiesin September 2016 and September 2013 and subsequentlyannualizing these differences by dividing them with thenumber of years in the specified period Surprisingly fora group of countries in which the annualized increase inthe percentage of RW voters exceeded 2 Figure 3 showsthat this increase is practically independent of the inflowof immigrants Why would countries with a relatively highand a relatively low inflow of immigrants exhibit about thesame increase in the percentage of RW voters This result

may be a consequence of the EUrsquos political organizationBecause the EU functions practically as a supranational statewith no internal borders if one country decides to acceptimmigrants this decision may have repercussions for all theother member states The increase in the percentage of RWpopulist voters may therefore more systematically depend onthe total inflow of immigrants into the entire EU expressedhere as a percentage of the total EU population than theinflow in any individual country

Anecdotal evidence to this effect can be seen in the caseof Sweden and Norway Sweden was among the countries hithard by the recent migrant crisis yet Norway had approx-imately the same annualized increase in the percentage ofRW voters A similar occurrence happened in Germany andPoland Germany experienced a high inflow of immigrantsand in Poland 53 of the population wanted their govern-ment to refuse asylum seekers from the Middle East andNorth Africa and only 33 thought the opposite If Polandhas already transitioned from the tolerant mode of democ-racy associated with globalization to a mode dominated byRW populism then the fraction of immigrants at whichthe Polish population is pushed beyond the tipping pointis much lower than in western EU countries Poland andHungary both share decades of socialist experience and areboth among the toughest opponents of immigration intothe EU Both strongly oppose EU quotas designed to evenlyspread the shock of the migrant crisis

4 Complexity

Presidential election

05

04

03

02

01

40

30

30

2016

2015

2014

10

10

20

20

50

0

R

W v

oter

s

VotersInflow

Imm

igra

tion

inflo

w (J

p

opul

atio

n)

Time (months)

(a)

10

6

2

14

Time (months)

2016201520142012

4812 24 360

1

2

3

4

5

6

7

0

R

W v

oter

s

Imm

igra

tion

inflo

w(J

times104)

VotersInflow

(b)

10

15

5

0

20

1 2 3 4 5 6

R

W v

oter

s

Immigration inflow (J times 104)

(c)

Figure 2 Immigration affects the support for right-wing populism II (a) An unprecedented inflow of immigrants into Austria coincided witha steady increase in the fraction of RW populist voters A solitary black dot represents the results of Austrian presidential election in May2016 in which an RW populist candidate secured almost 50 of votes This election shows that even after the record immigrant inflow atthe end of 2015 had subsided a decreasing trend in the number of immigrants that enter Austria did not automatically translate into lowersupport for the RWpopulist political option that is RWpopulism seems to bemore than just a craze (b) In Germany the increasing inflow ofimmigrants (monthly data [34 35]) rather clearly coincided with the increasing support for an RW populist party (c) A significant regressionemerges when the German case is presented as a scatter plot between the inflow of immigrants and the percentage of far-right voters

Figure 3 indicates that the interplay of factors influencingthe rising popularity of RW populism is more complex than asimple bivariate regression and thus we turn to econometricanalysis and multivariate regression Using the results of thesimple regression in Figure 1 we assume that the fractionof RW voters (response variable RW119905) in a given countryis determined by the fraction of immigrants (IM119871119905 ) livingin the country We further assume that the fraction of RWvoters depends on the overall inflow of immigrants into theEU (IMEU

119905 ) calculated relative to the total EU populationThis variable represents an ldquoimmigration shockrdquo in themodel To also take into account the possibility that violentincidents involving immigrants could affect the popular votewe include the total number of injuries (119868) and casualties

(119863) involving immigrants recorded across the EU [39] in themodel Additionally we take into account the unemploymentrate (119880) that might also affect the popular vote Finally weadd a variable119872119894119905 = (1minusMIPEX100) in whichMIPEX is themigrant integration policy index [40] a proxy for the integra-tion ratemdashthe larger the MIPEX the better the integration

We perform econometric analysis using a pooled time-series cross-section (TSCS) method that combines the cross-sectional data on multiple countries Here the number ofcountries is 119873 = 10 entirely consisting of the so-called olddemocracies Germany France Austria Netherlands Swe-den Norway Denmark Finland Greece and Italy Becausefor each country there are 119879 observations along the temporaldimension the entire dataset has119873 times 119879 = 370 observations

Complexity 5

FRA (116)

GRC (89)

HUN (06)

BGR (12)ESP

PRT

SWE (159)

DK (99)UK (124)

ITA (94)

FIN (54)

05 1

2

minus2

4

0

0

Annual immigration inflow ( population)

Ann

ual

chan

ge in

RW

vot

ers NOR (138) DEU (119)

BEL (104) AUT (157)

NL (117)

Figure 3 Immigrant inflows and the popularity of right-wing populistmovementsmdasha nonlinear threshold Shown is the annualized immi-grant inflow into a given country (horizontal axis) as a percentage ofthat countryrsquos population as well as the corresponding percentagechange in RW populist votes (vertical axis) In parentheses are thefractions of immigrants in the total population of the correspondingcountry For a group of countries in which the annualized increasein the percentage of RW voters exceeded 2 this increase isvirtually independent of the inflow of immigrants Such a result mayreflect the EUrsquos political organization that is the lack of internalborders whereby if one country decides to accept immigrants thedecision may have repercussions for all the other member statesWe also observe a threshold indicated by a dashed line at whichthe immigrant inflow into a given country is sufficiently high toinvariably provoke an increase in the percentage of RW populistvoters In the model construction this threshold suggests 120572 = 0004on an annual basis

We have an extra index 119894 = 1 2 119873 that refers to a cross-sectional unit giving

RW119894119905 = 1205730 + 120573119871IMIM119871119894119905 + 120573

EUIM IMEU119894119905 + 120573

ter119863 119863119894119905 + 120573

ter119868 119868119894119905

+ 120573119880119880119894119905 + 120573119872119872119894119905 + 119890119894119905(1)

where 119890119905 is the random errorTable 1 shows the results of the TSCS regression model

indicating that the fraction of immigrants in the generalpopulation and the immigration inflow into the entire EUare the significant explanatory variables In addition thecoefficient 120573119871IM is not only significant but also higher thanunity Perhaps surprisingly the response variable is notsignificantly affected by the number of injuries and casualtiesin violent incidents involving immigrants We also show thatunemployment insignificantly affects the popular vote

The aforementioned survey data suggest that not everyEU nation is equally tolerant to immigrants We believe thata proxy for this tolerance can be a fraction of RW votes justafter the Second World War (RW0) when the fractions ofimmigrants were considerably smaller than nowadays Notethat by far the largest fraction of RW votes was recordedin Austria 1167 Table 2 show the results of the TSCSregression model when RW0 is included We find that this

Table 1 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression as defined in (1) Test statistics Wald1205942(6) = 14787 and Prob gt 1205942 = 0000

Coeff Std err 119911 119875 gt |119911|120573119871IM 217 0527 412 0000120573EUIM 4311 532 810 0000120573119863ter minus37119890 minus 04 49119890 minus 04 minus078 0437120573119868ter 21119890 minus 04 16119890 minus 04 128 0202120573119872 0208 0212 098 0325120573119880 minus0167 0348 minus048 06301205730 minus0298 0103 minus289 0004

Table 2 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression as defined in (1) Test statistics Wald1205942(6) = 15154 and Prob gt 1205942 = 0000

Coeff Std err 119911 119875 gt |119911|120573119871IM 132 0445 296 0003120573EUIM 4559 535 852 0000120573ter119863 minus42119890 minus 04 49119890 minus 04 minus086 0392120573ter119868 22119890 minus 04 16119890 minus 04 138 0167120573119872 minus0019 02 minus009 0926120573119880 minus0015 0289 minus005 0959RW0 0904 0562 161 01081205730 minus0139 0093 minus150 0133

Table 3 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression Test statistics Wald 1205942(5) = 1424and Prob gt 1205942 = 0000

Coeff Std err 119905 stat 119875 gt 119905120573119871IM 217 0466 465 0000120573EUIM 4672 501 933 00001205730 minus0235 0056 minus415 0000

new regression insignificantly contributes to modern RWpopulism

Table 3 shows the results of the TSCS regression modelwithoutMIPEX unemployment and violent incidents Inter-estingly with the total inflow of immigrants into the EUof 100000 on a monthly basis coefficient values in Table 3suggest that around 30 of immigrants in the total popu-lation of a country are sufficient to cause larger than 50support for RW populist parties which corresponds to theresult obtained by a linear fit in Figure 1

The analyzed data do not indicate whether the rise of RWpopulism is a transient phenomenon or a longer-term changein political orientation One factor is the persistence of votermemory In the October 2015 local election in Vienna theRW populist party won 33 of the popular vote During thepresidential election a fewmonths later themigrant crisis hadreached a peak and the RW populist movement candidatesecured almost 50 of the votes narrowly losing to a leftistrival These results have been contested and a new electionin December 2016 brought around 48 to the RW populistcandidate indicating that this rise in RW populism is not

6 Complexity

short-term This phenomenon has been described in theliterature Betz [41] describes how a substantial increase inthe number of refugees and illegal immigrants in Europeancountries during the 1980s provoked a wave of radical RWpopulism Following these events in the early 1990s thereremained between 11 and 14 of Europeans who felt thepresence of other nationalities races and religions to beunsettling [41]

3 Model

Human interactions are often heterogeneous and prone toabrupt nonlinear responses Because this characterizes therise of the RW populist party and its RW candidate in theAustrian elections such linear approaches as the regressionin (1) yield only partial results A useful intuition is gained bythinking about the election system in a democratic country asa randomwalker Even in a bipolar political system not everyleft government is equally left and similarly not every rightgovernment is equally right Therefore as a result of electionthe randomwalker can move either left or right with the sizeof this move depending on the standard deviation Withoutlimitations such that democracy and the election processpossess an unlimited tolerance for every political option aftera long enough time the random walker is bound to endup either extremely right or extremely left Both of theselimits exhibit ideological rigidity likely to substantially reducethe level of democracy and tolerancemdashin agreement withPopper ldquounlimited tolerance must lead to the disappearanceof tolerancerdquo [42] Thus RW populism can be considered asjust one of the two randomwalk limits However the randomwalker intuition does not explain why a society would moveleft or right nor does it provide a microscopic interpretationof the societal processes at the individual agent level

To mechanistically characterize the rise of RW populismand account for the existence of tipping points in socialdynamics we use a complex network approach [20 43ndash45]Complex network science is able to emulate heterogeneity inhuman interactions and goes beyond capturing the dynam-ics near tipping points Heterogeneity is important whenconsidering immigration and integration issues becauseimmigrants in order to sustain their identity live in ldquohubsrdquothat make them more difficult to integrate than immigrantsmixed into the native population

We construct our model by setting a constant numberof native ldquoinsiderrdquo agents and arranging them in an Erdos-Renyi random network of business and personal contactsImmigrant ldquooutsiderrdquo agents are then added to the networkEach insider notices the percentage of outsiders in theirneighborhood and based on this percentage decides whetherto be supportive of globalization or RW populism Insideragents get information from their neighborhood and theinteraction is local and this data is essential in understandingtipping points [46] but there are also other relevant nonlocalinteractionsThere is furthermore the factor that informationcan be misperceived or misinterpreted In the following weformalize these concepts with three assumptions

Assumption i (media and economic influences) At eachone-month period of time 119905 insider agents are influencedby media at a probability rate 119901 and remain influencedfor a period 120591 We assume that this influence transformsinsiders into RW populist supporters Although media canaffect insiders in both directions we focus on the growthrate and disregard negative values of 119901 The effect of themedia is global and hearing that immigration is occurringcan transform some insiders into RW populist supportersirrespective of their local situation For example most likelyas the effect of media coverage of immigration to boththe EU and the UK during the BREXIT referendum mostUK districts with low immigration voted mainly for Leave[47] The media alongside a lower level of tolerance toimmigration can be an important reason why in ex-socialistcountries the large fraction of voters oppose receiving evena low overall fraction of immigrants The probability rate119901 can also reflect such economic factors as unemploymentThus we use equation [43] 119901lowast = 1 minus exp(minus119901120591) to calculatethe probability that a randomly chosen insider agent is beinginfluenced by the media

Assumption ii (local influence of outsiders) In local electionsin Greece in November 2010 although the far-right GoldenDawn party received only 53 of the vote in some neighbor-hoods of Athens with large immigrant communities the partywon nearly 20 [48] This suggests that contacts betweeninsiders and outsiders do matter Within our network modelwe maintain a constant number of 119873 insiders and then add119868(0) outsidersWe then increase the number of outsiders withan inflow 119869119905 at each moment 119905 (representing one month)We randomly place newly arriving outsiders between insideragents both of whom initially have an average number ofconnections (ie a degree) 119896 The total number of outsiders119868(119905) is obtained by summing the monthly 119869 values accordingto 119868(119905) = 119868(0) + sum119905119904=1 119869119904 At any given moment the fractionof outsiders will equal 119891119868(119905) = 119868(119905)(119873 + 119868(119905)) To accountfor the effect of contacts between insiders and outsiders weassume that any insider agent 119894with 119896119894 total connections turnsto RWpopulism at a rate 1199011015840 when this agent is surrounded byat least 119898119894 = 1198911015840119868119896119894 outsiders [43 46 49] where 0 lt 1198911015840119868 lt 1is a constant model parameter quantifying the toleration ofinsiders This assumption merits a few additional comments

First the probability that randomly chosen insider agent 119894with 119896119894 connections is surrounded by119898119894 outsiders and there-fore prone to RW populism is 1199011(119896119894 119898119894 119891119868) equiv sum119896119894119895=119898119894 119891

119895119868 (1 minus

119891119868)119896119894minus119895 ( 119896119894119896119894minus119895 ) In this formula 119891119868 is the true current state

of the network However the information may be biasedand cause insiders to think there are more outsiders than isactually the case If the bias is Δ119891119868 then 1199011(119896119894 119898119894 119891119868 + Δ119891119868) gt119901(119896119894 119898119894 119891119868) This increased probability 119901(119896119894 119898119894 119891119868 + Δ119891119868)implies that the tolerance parameter 1198911015840119868 must decrease byamount Δ1198911015840119868 which we estimate using condition 119901(119896119894 119898119894 minusΔ1198911015840119868119896119894 119891119868) = 119901(119896119894 119898119894 119891119868 + Δ119891119868) An implicit assumption hereis that all insider agents are equally tolerant to immigrantsbecause the tolerance parameter 1198911015840119868 is defined as a globalnetwork property rather than an individual agent property

Complexity 7

An alternative would be to assume a distribution of tolerancelevels in which case 1198911015840119868 would represent the mean [20]

We now expand assumption (ii) with extension (A) whenthe immigration inflow 119869 is below some threshold 1198691015840 thesociety becomes more tolerant When 119869 lt 1198691015840 at a giventime moment the tolerance parameter 1198911015840119868 increases by anamount Δ1198911015840119868 = 120575 gt 0 there is a balance between immigra-tion and immigrant integrationmdashoutsiders are successfullyintegratedmdashand insiders are able to acclimate to the changesin their society Figure 2(b) shows that the 1198691015840 value forGermany is approximately 10000 people per month Beingable to accurately determine the maximum 1198691015840 value is highlyrelevant to the success of globalization According to ourmodel immigration and integration can be the inevitableconsequences of globalization when 119869 lt 1198691015840 If this is notthe case globalization will be threatened by the rise of RWpopulism the democratic system will enter an intolerantmode and cooperation between nations will be downgradedon the list of political priorities

Empirical evidence suggests that we add an oppositeextension (B) as the inflow of outsiders 119869 crosses somethreshold 11986910158401015840 (which does not need to be equal to 1198691015840)society becomes less tolerant Mathematically extension (B)indicates that when 119869 gt 11986910158401015840 the tolerance parameter 1198911015840119868decreases by

Δ1198911015840119868 = minus120574119869 (2)

Using the econometric models in (1) we decrease thetolerance parameter proportional to inflow 119869 where 120574 isa proportionality coefficient expressing the sensitivity ofinsiders to high levels of outsider inflow Figure 3(b) showsthreshold 11986910158401015840 in terms of total population that is 11986910158401015840 = 120572119873where 120572 is another proportionality coefficient Figure 3(b)estimates the 120572 value when all of the EU countries are hitby the migrant crisis The dashed line is annual inflow belowwhich countries have a mixed response to immigration andabove which support for RW populism increases Because120572 = 11986910158401015840119873 and using Figure 3(b) 1211986910158401015840119873 asymp 0004 we obtain

120572 asymp 000033 (3)

Extensions (A) and (B) are opposite limiting cases onein which immigration is slow and the other in whichimmigration is rapid Apart from the empirical evidencethat these extensions are needed brain science for exampleoffers a physiological interpretation political attitudes have acounterpart in brain structure [50ndash52] If outsiders increaseat a rate and in a manner perceived as controllable byinsiders the prefrontal cortex of the human brain responsiblefor decision making and for moderating social behavioracclimates to the new circumstances but if the insidersperceive the outsiders to be invaders the prefrontal cortexis supplanted by the amygdala which induces a fightingreaction and tolerance is suppressed

Although with assumptions (i) and (ii) our modelaccounts for the processes that affect individual insider opin-ion the expansion of RW populism can become extremelyrapid when insiders are influenced by their peers This well-documented phenomenon in human interactions is further

accelerated when social media is added Thus the spread ofRW populism can be highly nonlinear much like the spreadof a highly contagious disease We include this nonlinear col-lective spreading mechanism in our third model assumption

Assumption iii (mutual insider contagion) At any givenmoment 119905 an insider agent 119894 with 119870119894 connections to otherinsiders turns to RW populism at rate 11990110158401015840 if at time 119905 minus 1this agent has at least119872119894 = 1198701198942 RW populist supporters intheir neighborhood Note that for simplicity factor 12 playsa role analogous to the tolerance parameter in assumption(ii) Because of connections between insider agents whenRW populism emerges anywhere in the network the populistmovement is able to spread like a contagion This collectivespreading indicates that insider agents can become RWpopulist supporters even when there are no outsiders inthe immediate neighborhood Thus some EU countries withalmost no immigration are opposed to accepting even a smallgroup of immigrants This may have affected the outcomeof the recent US presidential election in which the winningcandidate was often ridiculed in the mainstream mediamdashanattitude represented in our model by assumption (i)

The model here assumes imitative interactions withouttesting the validity of such an assumption However thisassumption could be tested by themethod of Agliari et al [53]if the required data were available The authors use the toolsfrom statistical mechanics to determine from data the natureof interactions in social customs such as local and mixedmarriages in Italy and neighboring European countries Thefraction of actualmarriages satisfying a given condition scalesdifferently with the fraction of all possible couples that satisfythe same condition depending on the nature of interactionsbetween agents If the interactions are independent (one-body model) the scaling is linear whereas if the interactionsare imitative (two-body model) the scaling is square-rooted

We now turn to the simulation results and their impli-cation Figure 4 shows how in a network of 5000 agents thefraction of RW populist supporters increases when there isa constant inflow of outsiders here 119869 = 2 per month Thisinflow is annually approximately 05 of the total populationwhich is slightly more than the threshold value impliedin Figure 3(b) Simulations with 119869 = 2 per month aredesigned to emulate the rapid limit in extension (B) describedabove After approximately 37 years of rapid globalization thenetwork reaches a tipping point and abruptly shifts to amodedominated by RW populism RW populism dominates whenmore than 50 of the network is made up of RW populistsupporters (ie 119875 gt 05 where 119875 denotes the fraction of RWpopulists) The threshold is 50 because in a democracy themajority rules

Figure 4 also shows how the simulated network underassumptions (i)ndash(iii) responds to shocks (red curve) At thisstage extension (B) does not yet operateThe constant annualinflow of outsiders 119869 = 2 is supplemented by two eventsat times 1199051 and 1199052 when 119869 = 200 The state of the networkcharacterized by the proportion of RWpopulists (119875) exhibitsa much stronger response at 1199052 than at 1199051 although the shockinflow (119869 = 200) is the same This occurs because at 1199052 the

8 Complexity

08

00

02

J = 2

t1

J = 2 except at

t2

Change in tolerance

500

04

06

Time

Prob

abili

ty o

f RW

pop

ulism

t1 amp t2

Figure 4 Nonlinearity a tipping point and the rise of right-wingpopulism Using a network of 119873 = 5000 agents each with anaverage of 15 connections we examine the effect of a constantinflow of outsiders at rate 119869 = 2 at each time step In this setupthe total number of outsiders at any moment in time is 119868(119905) =sum119894 119869119894 = ⟨119869⟩119905 As the fraction of outsiders 119891119868 = 119868(119873 + 119868)approaches the tolerance parameter 1198911015840119868 = 015 the presence of atipping point causes the fraction of RW populist supporters to startincreasing nonlinearly and eventually undergo a sudden jump (iea discontinuous change) at about 37 years (450 months) into thesimulation (black curve) The sudden jump happens much earlierif the inflow of outsiders experiences shocks at times 1199051 and 1199052at which 119869 = 200 outsiders enter the network In particular asthe network approaches the tipping point the effect of exactlythe same shock becomes disproportionately higher (red curve) Inthis case however the tolerance parameter is still kept constantFinally we also examine the case in which shocks at times 1199051 and1199052 affect the tolerance parameter where responsiveness is controlledby parameter 120574 = 00001 Here the second shock at 1199052 is sufficientto instantly tip the network into RW populism (green curve) Otherparameters are 119901 = 0007 120591 = 15 1199011015840 = 05 11990110158401015840 = 05 and 120572 = 0001

system is closer to the tipping point and consequently moreunstable than at 1199051 Because in real-world data the value of119869 can be biased due to estimation errors or misinterpretedinformation our results suggest that approaching the tippingpoint can be concurrent with strong nonlinear effects suchthat even a small shock can trigger a transition to a modedominated by RWpopulismThis can be evenmore explosive(in terms of 119875) if extension (B) is allowed to operate that isif the tolerance parameter 1198911015840119868 changes with 119869

Figure 4 shows a third simulation (green curve) inwhich the dynamics operate under assumptions (i)ndash(iii) withextension (B) The tolerance parameter 119891119868 thus changes with119869 as prescribed by (2) Because of the decreasing tolerancethe second shock at 1199052 can now push the system beyondthe tipping point and thus the dominance of RW populismoccurs earlier than in the two previous simulations

From the simulations in Figure 4 alone it is unclearhow much the local influence of outsiders (assumption (ii))contributes to the rise of RW populism relative to the insider

08

1

06

04

02

00

TotalLocal influence (ii)Contagion (iii)

250 500

Time

Prob

abili

ty o

f RW

pop

ulism

Figure 5 Breakdown of the causes of right-wing populism Figure 4shows that the probability of RW populism 119875 suddenly increasesas society approaches a tipping point but remains silent on theunderlying causes Here we discern between the contributionsof local outsider influence (assumption (ii)) and mutual insidercontagion (assumption (iii)) Far from the tipping point 119875 mainlyresponds to local outsider influence (ii) By contrast as the networkapproaches its tipping point mutual insider contagion (iii) takesover and accelerates the transition to RW populist dominanceParameter values are119873 = 5000 with an average degree of 15 119869 = 2119901 = 0007 120591 = 15 1199011015840 = 07 and 11990110158401015840 = 08

contagion (assumption (iii)) Figure 5 shows these contribu-tions After the initial transients fade the local influence ofoutsiders drives the increase in RW populists in the networkThe contribution of mutual insider contagion is relativelysmall until the system approaches a tipping point Near thetipping point contagion spreads rapidly and overtakes thelocal outsider influence as the main contributor to the riseof RW populism Thereafter the RW populist movement cansustain itself without support from the outside

In the regime of moderate immigration inflows (1198691015840 lt 119869 lt11986910158401015840) we can examine the dynamics of our complex networkusing themean-field theory (MFT) analytic techniqueWhenthe number of agent connections does not deviate greatlyfrom the network average the probability 119875 that a randomlychosen insider agent 119894 is an RWpopulist supporter due to anyof the processes underlying assumptions (i)ndash(iii) is

119875 = 119901lowast + 11990110158401199011 (119896119894 119898119894 119891119868) + 119901101584010158401199011 (119870119894119872119894 119875)

minus 119901lowast11990110158401199011 (119896119894 119898119894 119891119868) minus 119901lowast119901101584010158401199011 (119870119894119872119894 119875)

minus 11990110158401199011 (119896119894 119898119894 119891119868) 119901101584010158401199011 (119870119894119872119894 119875)

(4)

where the last three terms avoid double counting in accor-dance with the probability theory formula 119875(119860 cup 119861 cup 119862) =119875(119860) + 119875(119861) + 119875(119862) minus 119875(119860)119875(119861) minus 119875(119860)119875(119862) minus 119875(119861)119875(119862) forthree mutually independent events 119860 119861 and 119862 that cannotoccur simultaneously In the MFT approximation we candrop index 119894 because no single agent is markedly differentfrom the collective average Previously we set 119872 = 1198702

Complexity 9

417 882

88

200 600 800 1000 1200

Time (months)

13208

1

06

04

02

00

Prob

abili

ty o

f RW

pop

ulism

400

294

263185

555

J = 1 fI = 005

J = 1 fI = 01

J = 1 fIfI

= 015

J = 2 = 005

J = 3 fI = 005

J = 3 fI = 01

J = 3 fI = 015

J = 3 fI = 02

Figure 6 Predicting the timing of RW populism We find that for abroad range of outsider inflows (119869) and tolerance parameter values(1198911015840119868 ) (5) predicts the moment at which 119875 gt 05 in a manner thatagrees favorably with the simulation results Except for 120574 = 0 otherparameters are the same as in Figure 4

for simplicity but the peer pressure measured by the valueof the proportionality factor between 119872 and 119870 can differbetween countries or regions In addition parameters 1199011015840 and11990110158401015840 are constants only in theory The real social dynamicsare such that these parameters may change in responseto rumors political manipulations or outside shocks Ifparameters 1199011015840 and 11990110158401015840 substantially increase the value of 119875also increases thus further improving the prospects for RWpopulism dominance In our framework due to democracyrsquosmajority rule principle when 119875 approaches 05 the nonlinearprocesses embedded in assumptions (ii) and (iii) cause asudden transition to RW populist mode

Because a theoretical model is more useful if it haspredictive power [54 55] we show how a network of agentsunder assumptions (i)ndash(iii) leads to a simple formula for thetiming at which RW populism starts to dominate

119905119905ℎ =1198731198911015840119868

119869 (1 minus 1198911015840119868 )minus 119868 (0)

119869 (5)

Gaining this result involves three steps First if the immi-gration inflow is constant then the number of outsidersin the network after 119905 time steps is 119868(0) + 119869119905 Second thetotal population size thus equals 119873 + 119868(0) + 119869119905 Finally (5)follows if the current fraction of outsiders (119868(0) + 119869119905)(119873 +119868(0) + 119869119905) is equated with the critical parameter 1198911015840119868 Figure 6shows that for a number of immigration inflow-toleranceparameter pairs (119869 1198911015840119868 ) the simulated timing of the shift toRW populist mode (ie 119875 gt 05) fits theoretical predictionsIn conjunction with the empirical data on tolerance towardsimmigrants in the EU countries the formula in (5) could beused to provide an estimate of when a given country mightbe approaching a possible tipping point to RW populismdominance

Numerical simulations allow us to examine not only iso-lated single networks but also the interdependence betweentwo or more networks Note that there is a potential for a cas-cade effect when an RW populist movement in one networkcauses the rise of RW populist movements in other networksThis cascade effect is growing in relevance because expandingglobalization is causing countries to become increasinglysimilar to each another and this similarity increases in suchsupranational organizations as the EU in which bordersbetween nation states are rapidly fading

To examine how interdependence affects the rise of RWpopulism we set up two random economically equivalentER networks (equal 119901 in the model) with different tolerancelevels towards outsiders (different 1198911015840119868 in the model) Tothe usual intraconnections within each network we addinterconnecting agents that linkwith their counterparts in theother network Apart from this addition we retain the samemodel assumptions as previously held

To examine the effects of interconnectedness we first runnumerical simulations of two independent networks withoutinterconnections [see Figure 7(a)] As expected when theinflow level of outsiders is the same (119869 = 2) the networkwith a higher tolerance parameter (11989110158401198681 = 04) reaches thetipping point much later than the network with a lowertolerance parameter (11989110158401198682 = 02) When the networks areinterconnected and the more tolerant network experiencesan increased inflow (1198692 = 4) and a shock (119869 = 500) at time1199051 = 500 its susceptibility to RW populism increases andit also affects the other network and shortens the time oftransition to RW populism [see Figure 7(b)] It would appearthat countries do not want to be the first to cross the line butin an interconnected world being second is easier

4 Discussion and Conclusion

Why some countries (eg the ex-socialist EU countries)strongly oppose receiving immigrants while others (eg theUSA) have a long history of receiving immigrants is animportant topic in the social sciences and more recently amajor issue for the EU Perhaps receiving immigrants hasbeen an ongoing pattern in the USA because there is nosingle dominant ethnicity and thus a clear distinction ismadebetween USA national identity and the country of originof its citizens In addition because USA citizens are peoplefrom all over the world there is no single dominant religiousethnic or cultural group that can organize and threaten theestablished social order In France we find the oppositeThereis a large group of immigrants with different language andreligion from the French majority whose presence can instillfear among themajority Fear exacerbated by an inflow rate ofimmigrants that exceeds the rate of their integration can leadto a volatile situation one that is often resolved in one of twoways Either there is an immigrant uprising as exemplifiedby the Visigoth immigrants and their ex-Roman commanderAlaricwhoplunderedRome in 410 or themajority populationsuppresses the inflow of immigrants Currently this secondoption is often accomplished by supporting populist right-wing parties

10 Complexity

08

1

00

06

0402

J1 = 2

J2 = 2

200015001000500

Prob

abili

ty o

f RW

pop

ulism

Time

(a)

08

1

00

06

2000

0402

J1 = 2

J2 = 4

Time15001000500Pr

obab

ility

of R

W p

opul

ism

(b)

Figure 7 Interconnected networks or why ldquosomebody elsersquos problemrdquo easily turns into ldquomy problemrdquo In (a) we show the case when there areno interlinks between networks The tolerance parameters between the two networks differ 1198911198681 = 02 and 1198911198681 = 04 while the inflows intoboth networks are the same 1198691 = 1198692 = 2 (b) More tolerant network is now exposed to a higher inflow 1198692 = 4 and a shock at 1199051 = 500 Theaverage number of connections for intraconnections (interconnections) in both networks equals 15 (10) The other parameters are the sameas in Figure 4

Although globalization was conceived to allow capitaland labor to move freely across national borders real-worldglobalization is affected by a multitude of noneconomicfactors such as ethnicity culture and religion With so manyfactors at play globalization is an enormously complex pro-cess and often noneconomic factors do not align with purelyeconomic factors This misalignment can lead to frustrationsthat feed populist movements By opposing the collaborationthat comes with globalization populist movements act asa feedback mechanism that pushes the general populationtowards becoming more ideologically rigid and this inturn further accelerates the populism [56] A strengthenedpopulist movement can also trigger tectonic shifts in worldaffairs as exemplified by BREXIT in the UK and the recent2016 US presidential election

Problems in a globalized world rarely confine themselvesto one place Interdependencemakes the developed countriesmore alike and synchronizes their social dynamics Thissynchronization can cause political shifts in one country tospill over into other countries and this is what enables the riseof RW populism to spread across large regions of the worldAfter BREXIT and the 2016 US presidential election politicalelites should not expect to continue business as usual Beingthe first to adopt amajor political shift with a high probabilityof negative economic consequences is difficult but once thatline has been crossed the interdependence of globalizationmakes cascades (domino effects) an active possibility No onewants to be first but many are ready to be second

The tipping point that brings on the rise of RW populismmay be reached more quickly when voters face a binarychoice for example the ldquoyesrdquo or ldquonordquo choice in the UKBREXIT referendum and the two major-party candidates inthe recent US presidential election In both cases the populistoption secured a narrow victory As mentioned above thepolls indicate that the populist party in Austria can expect toreceive 34 of the votes but the populist party candidate inthe presidential race can expect close to 51 A simple wayto understand these percentages is to assume that politicalattitudes of voters are approximately symmetrical in their

distribution across the political spectrum from left to rightConsequently if leftist voters comprise 120595119871 of the totalpopulation then RW populist voters will maintain a similarpresence that is 120595119877 asymp 120595119871 Facing this binary choice thecentrist voters have no one representing their views and thusare likely to vote evenly between the two available optionsAn implication is that when 120595119871 asymp 120595119877 asymp 33 even aslight (statistically significant) imbalance in favor of120595119877over120595119871 can tip the society towards RW populism In Austria120595119877 asymp 36 seems to be sufficient to make the RW populistcandidate a front runner in the presidential race

The final outcome of the battle between the conflictingfactors surrounding globalization will almost surely havetremendous economic implications If a country approachesits tipping point how will the ensuing volatility affect itslong-term credit rating If a domino effect cascades acrossa large region of the EU or the entire EU how will thataffect the Euro and the common banking system Withouta proper resolution of the migrant crisis what will be theimpact on systemic risk In an attempt to shed light onsome of the factors and underlying processes that affectthe success of globalization we offer an empirically backedtheoretical model of the rise of RW populism in response tounsustainable immigration inflows

Our model emphasizes the need for controlled globaliza-tion in which immigration inflows into a society are balancedwith the ability of the society to integrate the immigrantsThis ability is arguably improved when immigrants mixwith the native population which is a principle practicedin Singapore where immigrant hubs are discouraged andtenants in government-built housing (comprising 88 ofall housing) must be of mixed ethnic origin [20] Becausetolerance towards immigrants is conditional when immigra-tion inflows overshadow integration rates the society can betipped into RWpopulism an intolerant mode of functioningThe rise of RW populism can occur because elections areby their nature stochastic and they resemble a mathematicalrandom walker Left to its own devices a random walkerwill eventually hit an absorbing barrier Here this barrier

Complexity 11

of course is a metaphor for the demise of globalizationironically at the very hands of the progressive system (iedemocracy) thatmade globalization possible in the first place

Disclosure

A version of this work was presented in Warsaw at ldquoEcono-physics Colloquium 2017rdquo

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

The authors are grateful to S Galam and T Lipic forhelpful suggestions Boris Podobnik received the supportfrom Slovenian Research Agency (ARRS) via Project J5-8236and from the University of Rijeka Boris Podobnik and H EStanley received support from the Defense Threat ReductionAgency (DTRA) theOffice ofNaval Research (ONR) and theNational Science Foundation (NSF) (Grant CMMI 1125290)Marko Jusup was supported by the Japan Science andTechnology Agency (JST) Program to Disseminate TenureTracking System

References

[1] H Grubel and A Scott ldquoThe international flow of humancapitalrdquo American Economic Review vol 56 pp 268ndash274 1966

[2] G C Hufbauer and K Suominen Globalization at Risk Chal-lenges to Finance and Trade Yale University Press 2010

[3] F Docquier and H Rapoport ldquoGlobalization Brain Drain andDevelopmentrdquo Journal of Economic Literature vol 50 pp 681ndash730 2012

[4] D Rodrik The Globalization Paradox Democracy and theFuture of the World Economy W W Norton amp Company NewYork USA 2011

[5] B R Chiswick ldquoThe Effect of Americanization on the Earningsof Foreign-born Menrdquo Journal of Political Economy vol 86 no5 pp 897ndash921 1978

[6] G J BorjasHeavenrsquosDoor Immigration Policy and theAmericanEconomy Princeton University Press 1999

[7] G J Borjas ldquoThe economics of immigrationrdquo Journal ofEconomic Literature vol 32 pp 1667ndash1717 1994

[8] G J Borjas ldquoAssimilation Changes in Cohort Quality and theEarnings of Immigrantsrdquo Journal of Labor Economics vol 3 no4 pp 463ndash489 1985

[9] D Swank and H G Betz ldquoGlobalization the welfare stateand right-wing populism in Western Europerdquo Socio-EconomicReview vol 1 no 2 pp 215ndash245 2003

[10] J Gibson and D McKenzie ldquoEight Questions about BrainDrainsrdquo Journal of Economic Perspectives vol 25 no 3 pp 107ndash128 2011

[11] H Betz ldquoPolitics of Resentment Right-Wing Radicalism inWest Germanyrdquo Comparative Politics vol 23 no 1 pp 45ndash601990

[12] R Knight ldquoHaider the freedom party and the extreme right inAustriardquo Parliamentary Affairs vol 45 no 3 pp 285ndash299 1992

[13] P Fysh and J Wolfreys ldquoLe pen the National Front and theextreme right in Francerdquo Parliamentary Affairs vol 45 no 3pp 309ndash326 1992

[14] G Voerman and P Lucardie ldquoThe extreme right in the Nether-lands The centrists and their radical rivalsrdquo European Journalof Political Research vol 22 no 1 pp 35ndash54 1992

[15] WD Chapin ldquoExplaining the electoral success of the new rightThe German caserdquoWest European Politics vol 20 no 2 pp 53ndash72 1997

[16] J M Smith ldquoDoes crime pay Issue ownership political oppor-tunity and the populist right in Western Europerdquo ComparativePolitical Studies vol 43 no 11 pp 1471ndash1498 2010

[17] B Bonikowski and PDiMaggio ldquoVarieties of American PopularNationalismrdquo American Sociological Review vol 81 no 5 pp949ndash980 2016

[18] D Rodrik ldquoHow far will international economic integrationgordquo Journal of Economic Perspectives vol 14 no 1 pp 177ndash1862000

[19] B Davis ldquoDespite his heritage prominent economist backs im-migration cutrdquo The Wall Street Journal 1996 httpswwwhksharvardedufsgborjaspublicationspopularWSJ042696pdf

[20] B Podobnik M Jusup Z Wang and H E Stanley ldquoHow fearof future outcomes affects social dynamicsrdquo Journal of StatisticalPhysics vol 167 no 3-4 pp 1007ndash1019 2017

[21] R Reuveny ldquoClimate change-induced migration and violentconflictrdquo Political Geography vol 26 no 6 pp 656ndash673 2007

[22] T C Schelling ldquoDynamic models of segregationrdquo Journal ofMathematical Sociology vol 1 pp 143ndash186 1971

[23] R Axelrod ldquoThe dissemination of culture a model withlocal convergence and global polarizationrdquo Journal of ConflictResolution vol 41 no 2 pp 203ndash226 1997

[24] M McPherson L Smith-Lovin and J M Cook ldquoBirds of aFeather Homophily in Social Networksrdquo Annual Review ofSociology vol 27 pp 415ndash444 2001

[25] T Antal P L Krapivsky and S Redner ldquoDynamics of socialbalance on networksrdquo Physical Review E vol 72 Article ID036121 2005

[26] S Marvel J Jon Kleinbergb D Robert R D Kleinberg andS Strogatz ldquoContinuous-Time Model of Structural BalancerdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 108 no 5 pp 1771ndash1776 2011

[27] C Gracia-Lazaro L M Flora and Y Moreno ldquoSelectiveAdvantage of Tolerant Cultural Traits in the Axelrod-SchellingModelrdquo Physical Review E vol 83 Article ID 056103 2011

[28] F Aguiar and A Parravano ldquoTolerating the IntolerantHomophily Intolerance and Segregation in Social BalancedNetworksrdquo Journal of Conflict Resolution vol 59 no 1 pp 29ndash50 2015

[29] M Lim R Metzler and Y Bar-Yam ldquoGlobal pattern formationand ethniccultural violencerdquo Science vol 317 no 5844 pp1540ndash1544 2007

[30] R M Dancygier Immigration and Conflict in Europe Studies inComparative Politics Cambridge University Press 2010

[31] R M Dancygier and D D Laitin ldquoImmigration into EuropeEconomic Discrimination Violence and Public Policyrdquo AnnualReview of Political Science vol 17 pp 43ndash64 2014

[32] S Galam and M A Javarone ldquoModeling Radicalization Phe-nomena in Heterogeneous Populationsrdquo PLoS ONE vol 11Article ID e0155407 2016

Complexity 3

40

30

10

20

20 25

AUT

FRA

NL

NOR

SWE

UK

50

0

0 5

FINGRC

10

BEL

DEU

ITA15

R

W v

oter

s

DK

immigrants

(a)

40

30

20

50

16

immigrants17 18 191514

R

W v

oter

s

Austria

(b)

Figure 1 Immigration affects the support for right-wing populism I (a) Among the EU countries involved in the recent migrant crisis supportfor RW populism is generally higher in those countries that accepted a larger number of immigrants relative to the countryrsquos populationsize Shown is June 2016 Seeing democracy as the majority rule principle we presume that RW populism becomes a dominant politicaloption when the percentage of RW voters exceeds 50 Judging based on a cumulative exponential function that fits the data reasonablywellmdash119910 = 443 exp(0097119909)mdashRW populism in the examined EU countries may take over if the percentage of immigrants in the totalpopulation approaches 22 Shown is also a linear fit 119910 = minus367 + 180119909 roughly giving that the RW populism reaches the majority forthe percentage of immigrants equal to approximately 30 Coefficients of determination (1198772) for the two models are 03866 and 04077respectively Akaike weights indicate that given the dataset at hand the exponential model is better with probability 443 while the linearmodel is better with the remaining probability 557 (b) Similar as in the other EU countries Austrian data reveal that the increase inthe percentage of immigrants is accompanied with an increase in the percentage of RW populist voters Here too a cumulative exponentialfunction fits the data well This function predicts the rise of RW populism in Austria when the percentage of immigrants is slightly below the20 mark

is responding to the inflow of immigrants For example inAustria the far-right party won 205 of the popular votein 2013 in response to the percentage of immigrants livingin Austria at the time but in the second half of 2015 theinflow of immigrants increased sharply [34 35] and a localelection in Vienna saw the percentage of RW votes jump to33 This sudden change suggests the presence of a phasetransition tipping point or critical point [37 38] that is thenearer the percentage of immigrants comes to the tippingpoint the more quickly voters turn to extreme politicalalternatives Figures 2(b) and 2(c) show a qualitatively similarphenomenon occurring in Germany

In an attempt to probe deeper into the internal dynamicsof RW populism in the EU as a function of the inflow ofimmigrants next we analyze how the immigration rate affectsthe rise in RW populist voters In Figure 3 we qualitativelyrepresent the annualized increase in RW votes by taking thedifferences between the popularity of RW populist partiesin September 2016 and September 2013 and subsequentlyannualizing these differences by dividing them with thenumber of years in the specified period Surprisingly fora group of countries in which the annualized increase inthe percentage of RW voters exceeded 2 Figure 3 showsthat this increase is practically independent of the inflowof immigrants Why would countries with a relatively highand a relatively low inflow of immigrants exhibit about thesame increase in the percentage of RW voters This result

may be a consequence of the EUrsquos political organizationBecause the EU functions practically as a supranational statewith no internal borders if one country decides to acceptimmigrants this decision may have repercussions for all theother member states The increase in the percentage of RWpopulist voters may therefore more systematically depend onthe total inflow of immigrants into the entire EU expressedhere as a percentage of the total EU population than theinflow in any individual country

Anecdotal evidence to this effect can be seen in the caseof Sweden and Norway Sweden was among the countries hithard by the recent migrant crisis yet Norway had approx-imately the same annualized increase in the percentage ofRW voters A similar occurrence happened in Germany andPoland Germany experienced a high inflow of immigrantsand in Poland 53 of the population wanted their govern-ment to refuse asylum seekers from the Middle East andNorth Africa and only 33 thought the opposite If Polandhas already transitioned from the tolerant mode of democ-racy associated with globalization to a mode dominated byRW populism then the fraction of immigrants at whichthe Polish population is pushed beyond the tipping pointis much lower than in western EU countries Poland andHungary both share decades of socialist experience and areboth among the toughest opponents of immigration intothe EU Both strongly oppose EU quotas designed to evenlyspread the shock of the migrant crisis

4 Complexity

Presidential election

05

04

03

02

01

40

30

30

2016

2015

2014

10

10

20

20

50

0

R

W v

oter

s

VotersInflow

Imm

igra

tion

inflo

w (J

p

opul

atio

n)

Time (months)

(a)

10

6

2

14

Time (months)

2016201520142012

4812 24 360

1

2

3

4

5

6

7

0

R

W v

oter

s

Imm

igra

tion

inflo

w(J

times104)

VotersInflow

(b)

10

15

5

0

20

1 2 3 4 5 6

R

W v

oter

s

Immigration inflow (J times 104)

(c)

Figure 2 Immigration affects the support for right-wing populism II (a) An unprecedented inflow of immigrants into Austria coincided witha steady increase in the fraction of RW populist voters A solitary black dot represents the results of Austrian presidential election in May2016 in which an RW populist candidate secured almost 50 of votes This election shows that even after the record immigrant inflow atthe end of 2015 had subsided a decreasing trend in the number of immigrants that enter Austria did not automatically translate into lowersupport for the RWpopulist political option that is RWpopulism seems to bemore than just a craze (b) In Germany the increasing inflow ofimmigrants (monthly data [34 35]) rather clearly coincided with the increasing support for an RW populist party (c) A significant regressionemerges when the German case is presented as a scatter plot between the inflow of immigrants and the percentage of far-right voters

Figure 3 indicates that the interplay of factors influencingthe rising popularity of RW populism is more complex than asimple bivariate regression and thus we turn to econometricanalysis and multivariate regression Using the results of thesimple regression in Figure 1 we assume that the fractionof RW voters (response variable RW119905) in a given countryis determined by the fraction of immigrants (IM119871119905 ) livingin the country We further assume that the fraction of RWvoters depends on the overall inflow of immigrants into theEU (IMEU

119905 ) calculated relative to the total EU populationThis variable represents an ldquoimmigration shockrdquo in themodel To also take into account the possibility that violentincidents involving immigrants could affect the popular votewe include the total number of injuries (119868) and casualties

(119863) involving immigrants recorded across the EU [39] in themodel Additionally we take into account the unemploymentrate (119880) that might also affect the popular vote Finally weadd a variable119872119894119905 = (1minusMIPEX100) in whichMIPEX is themigrant integration policy index [40] a proxy for the integra-tion ratemdashthe larger the MIPEX the better the integration

We perform econometric analysis using a pooled time-series cross-section (TSCS) method that combines the cross-sectional data on multiple countries Here the number ofcountries is 119873 = 10 entirely consisting of the so-called olddemocracies Germany France Austria Netherlands Swe-den Norway Denmark Finland Greece and Italy Becausefor each country there are 119879 observations along the temporaldimension the entire dataset has119873 times 119879 = 370 observations

Complexity 5

FRA (116)

GRC (89)

HUN (06)

BGR (12)ESP

PRT

SWE (159)

DK (99)UK (124)

ITA (94)

FIN (54)

05 1

2

minus2

4

0

0

Annual immigration inflow ( population)

Ann

ual

chan

ge in

RW

vot

ers NOR (138) DEU (119)

BEL (104) AUT (157)

NL (117)

Figure 3 Immigrant inflows and the popularity of right-wing populistmovementsmdasha nonlinear threshold Shown is the annualized immi-grant inflow into a given country (horizontal axis) as a percentage ofthat countryrsquos population as well as the corresponding percentagechange in RW populist votes (vertical axis) In parentheses are thefractions of immigrants in the total population of the correspondingcountry For a group of countries in which the annualized increasein the percentage of RW voters exceeded 2 this increase isvirtually independent of the inflow of immigrants Such a result mayreflect the EUrsquos political organization that is the lack of internalborders whereby if one country decides to accept immigrants thedecision may have repercussions for all the other member statesWe also observe a threshold indicated by a dashed line at whichthe immigrant inflow into a given country is sufficiently high toinvariably provoke an increase in the percentage of RW populistvoters In the model construction this threshold suggests 120572 = 0004on an annual basis

We have an extra index 119894 = 1 2 119873 that refers to a cross-sectional unit giving

RW119894119905 = 1205730 + 120573119871IMIM119871119894119905 + 120573

EUIM IMEU119894119905 + 120573

ter119863 119863119894119905 + 120573

ter119868 119868119894119905

+ 120573119880119880119894119905 + 120573119872119872119894119905 + 119890119894119905(1)

where 119890119905 is the random errorTable 1 shows the results of the TSCS regression model

indicating that the fraction of immigrants in the generalpopulation and the immigration inflow into the entire EUare the significant explanatory variables In addition thecoefficient 120573119871IM is not only significant but also higher thanunity Perhaps surprisingly the response variable is notsignificantly affected by the number of injuries and casualtiesin violent incidents involving immigrants We also show thatunemployment insignificantly affects the popular vote

The aforementioned survey data suggest that not everyEU nation is equally tolerant to immigrants We believe thata proxy for this tolerance can be a fraction of RW votes justafter the Second World War (RW0) when the fractions ofimmigrants were considerably smaller than nowadays Notethat by far the largest fraction of RW votes was recordedin Austria 1167 Table 2 show the results of the TSCSregression model when RW0 is included We find that this

Table 1 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression as defined in (1) Test statistics Wald1205942(6) = 14787 and Prob gt 1205942 = 0000

Coeff Std err 119911 119875 gt |119911|120573119871IM 217 0527 412 0000120573EUIM 4311 532 810 0000120573119863ter minus37119890 minus 04 49119890 minus 04 minus078 0437120573119868ter 21119890 minus 04 16119890 minus 04 128 0202120573119872 0208 0212 098 0325120573119880 minus0167 0348 minus048 06301205730 minus0298 0103 minus289 0004

Table 2 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression as defined in (1) Test statistics Wald1205942(6) = 15154 and Prob gt 1205942 = 0000

Coeff Std err 119911 119875 gt |119911|120573119871IM 132 0445 296 0003120573EUIM 4559 535 852 0000120573ter119863 minus42119890 minus 04 49119890 minus 04 minus086 0392120573ter119868 22119890 minus 04 16119890 minus 04 138 0167120573119872 minus0019 02 minus009 0926120573119880 minus0015 0289 minus005 0959RW0 0904 0562 161 01081205730 minus0139 0093 minus150 0133

Table 3 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression Test statistics Wald 1205942(5) = 1424and Prob gt 1205942 = 0000

Coeff Std err 119905 stat 119875 gt 119905120573119871IM 217 0466 465 0000120573EUIM 4672 501 933 00001205730 minus0235 0056 minus415 0000

new regression insignificantly contributes to modern RWpopulism

Table 3 shows the results of the TSCS regression modelwithoutMIPEX unemployment and violent incidents Inter-estingly with the total inflow of immigrants into the EUof 100000 on a monthly basis coefficient values in Table 3suggest that around 30 of immigrants in the total popu-lation of a country are sufficient to cause larger than 50support for RW populist parties which corresponds to theresult obtained by a linear fit in Figure 1

The analyzed data do not indicate whether the rise of RWpopulism is a transient phenomenon or a longer-term changein political orientation One factor is the persistence of votermemory In the October 2015 local election in Vienna theRW populist party won 33 of the popular vote During thepresidential election a fewmonths later themigrant crisis hadreached a peak and the RW populist movement candidatesecured almost 50 of the votes narrowly losing to a leftistrival These results have been contested and a new electionin December 2016 brought around 48 to the RW populistcandidate indicating that this rise in RW populism is not

6 Complexity

short-term This phenomenon has been described in theliterature Betz [41] describes how a substantial increase inthe number of refugees and illegal immigrants in Europeancountries during the 1980s provoked a wave of radical RWpopulism Following these events in the early 1990s thereremained between 11 and 14 of Europeans who felt thepresence of other nationalities races and religions to beunsettling [41]

3 Model

Human interactions are often heterogeneous and prone toabrupt nonlinear responses Because this characterizes therise of the RW populist party and its RW candidate in theAustrian elections such linear approaches as the regressionin (1) yield only partial results A useful intuition is gained bythinking about the election system in a democratic country asa randomwalker Even in a bipolar political system not everyleft government is equally left and similarly not every rightgovernment is equally right Therefore as a result of electionthe randomwalker can move either left or right with the sizeof this move depending on the standard deviation Withoutlimitations such that democracy and the election processpossess an unlimited tolerance for every political option aftera long enough time the random walker is bound to endup either extremely right or extremely left Both of theselimits exhibit ideological rigidity likely to substantially reducethe level of democracy and tolerancemdashin agreement withPopper ldquounlimited tolerance must lead to the disappearanceof tolerancerdquo [42] Thus RW populism can be considered asjust one of the two randomwalk limits However the randomwalker intuition does not explain why a society would moveleft or right nor does it provide a microscopic interpretationof the societal processes at the individual agent level

To mechanistically characterize the rise of RW populismand account for the existence of tipping points in socialdynamics we use a complex network approach [20 43ndash45]Complex network science is able to emulate heterogeneity inhuman interactions and goes beyond capturing the dynam-ics near tipping points Heterogeneity is important whenconsidering immigration and integration issues becauseimmigrants in order to sustain their identity live in ldquohubsrdquothat make them more difficult to integrate than immigrantsmixed into the native population

We construct our model by setting a constant numberof native ldquoinsiderrdquo agents and arranging them in an Erdos-Renyi random network of business and personal contactsImmigrant ldquooutsiderrdquo agents are then added to the networkEach insider notices the percentage of outsiders in theirneighborhood and based on this percentage decides whetherto be supportive of globalization or RW populism Insideragents get information from their neighborhood and theinteraction is local and this data is essential in understandingtipping points [46] but there are also other relevant nonlocalinteractionsThere is furthermore the factor that informationcan be misperceived or misinterpreted In the following weformalize these concepts with three assumptions

Assumption i (media and economic influences) At eachone-month period of time 119905 insider agents are influencedby media at a probability rate 119901 and remain influencedfor a period 120591 We assume that this influence transformsinsiders into RW populist supporters Although media canaffect insiders in both directions we focus on the growthrate and disregard negative values of 119901 The effect of themedia is global and hearing that immigration is occurringcan transform some insiders into RW populist supportersirrespective of their local situation For example most likelyas the effect of media coverage of immigration to boththe EU and the UK during the BREXIT referendum mostUK districts with low immigration voted mainly for Leave[47] The media alongside a lower level of tolerance toimmigration can be an important reason why in ex-socialistcountries the large fraction of voters oppose receiving evena low overall fraction of immigrants The probability rate119901 can also reflect such economic factors as unemploymentThus we use equation [43] 119901lowast = 1 minus exp(minus119901120591) to calculatethe probability that a randomly chosen insider agent is beinginfluenced by the media

Assumption ii (local influence of outsiders) In local electionsin Greece in November 2010 although the far-right GoldenDawn party received only 53 of the vote in some neighbor-hoods of Athens with large immigrant communities the partywon nearly 20 [48] This suggests that contacts betweeninsiders and outsiders do matter Within our network modelwe maintain a constant number of 119873 insiders and then add119868(0) outsidersWe then increase the number of outsiders withan inflow 119869119905 at each moment 119905 (representing one month)We randomly place newly arriving outsiders between insideragents both of whom initially have an average number ofconnections (ie a degree) 119896 The total number of outsiders119868(119905) is obtained by summing the monthly 119869 values accordingto 119868(119905) = 119868(0) + sum119905119904=1 119869119904 At any given moment the fractionof outsiders will equal 119891119868(119905) = 119868(119905)(119873 + 119868(119905)) To accountfor the effect of contacts between insiders and outsiders weassume that any insider agent 119894with 119896119894 total connections turnsto RWpopulism at a rate 1199011015840 when this agent is surrounded byat least 119898119894 = 1198911015840119868119896119894 outsiders [43 46 49] where 0 lt 1198911015840119868 lt 1is a constant model parameter quantifying the toleration ofinsiders This assumption merits a few additional comments

First the probability that randomly chosen insider agent 119894with 119896119894 connections is surrounded by119898119894 outsiders and there-fore prone to RW populism is 1199011(119896119894 119898119894 119891119868) equiv sum119896119894119895=119898119894 119891

119895119868 (1 minus

119891119868)119896119894minus119895 ( 119896119894119896119894minus119895 ) In this formula 119891119868 is the true current state

of the network However the information may be biasedand cause insiders to think there are more outsiders than isactually the case If the bias is Δ119891119868 then 1199011(119896119894 119898119894 119891119868 + Δ119891119868) gt119901(119896119894 119898119894 119891119868) This increased probability 119901(119896119894 119898119894 119891119868 + Δ119891119868)implies that the tolerance parameter 1198911015840119868 must decrease byamount Δ1198911015840119868 which we estimate using condition 119901(119896119894 119898119894 minusΔ1198911015840119868119896119894 119891119868) = 119901(119896119894 119898119894 119891119868 + Δ119891119868) An implicit assumption hereis that all insider agents are equally tolerant to immigrantsbecause the tolerance parameter 1198911015840119868 is defined as a globalnetwork property rather than an individual agent property

Complexity 7

An alternative would be to assume a distribution of tolerancelevels in which case 1198911015840119868 would represent the mean [20]

We now expand assumption (ii) with extension (A) whenthe immigration inflow 119869 is below some threshold 1198691015840 thesociety becomes more tolerant When 119869 lt 1198691015840 at a giventime moment the tolerance parameter 1198911015840119868 increases by anamount Δ1198911015840119868 = 120575 gt 0 there is a balance between immigra-tion and immigrant integrationmdashoutsiders are successfullyintegratedmdashand insiders are able to acclimate to the changesin their society Figure 2(b) shows that the 1198691015840 value forGermany is approximately 10000 people per month Beingable to accurately determine the maximum 1198691015840 value is highlyrelevant to the success of globalization According to ourmodel immigration and integration can be the inevitableconsequences of globalization when 119869 lt 1198691015840 If this is notthe case globalization will be threatened by the rise of RWpopulism the democratic system will enter an intolerantmode and cooperation between nations will be downgradedon the list of political priorities

Empirical evidence suggests that we add an oppositeextension (B) as the inflow of outsiders 119869 crosses somethreshold 11986910158401015840 (which does not need to be equal to 1198691015840)society becomes less tolerant Mathematically extension (B)indicates that when 119869 gt 11986910158401015840 the tolerance parameter 1198911015840119868decreases by

Δ1198911015840119868 = minus120574119869 (2)

Using the econometric models in (1) we decrease thetolerance parameter proportional to inflow 119869 where 120574 isa proportionality coefficient expressing the sensitivity ofinsiders to high levels of outsider inflow Figure 3(b) showsthreshold 11986910158401015840 in terms of total population that is 11986910158401015840 = 120572119873where 120572 is another proportionality coefficient Figure 3(b)estimates the 120572 value when all of the EU countries are hitby the migrant crisis The dashed line is annual inflow belowwhich countries have a mixed response to immigration andabove which support for RW populism increases Because120572 = 11986910158401015840119873 and using Figure 3(b) 1211986910158401015840119873 asymp 0004 we obtain

120572 asymp 000033 (3)

Extensions (A) and (B) are opposite limiting cases onein which immigration is slow and the other in whichimmigration is rapid Apart from the empirical evidencethat these extensions are needed brain science for exampleoffers a physiological interpretation political attitudes have acounterpart in brain structure [50ndash52] If outsiders increaseat a rate and in a manner perceived as controllable byinsiders the prefrontal cortex of the human brain responsiblefor decision making and for moderating social behavioracclimates to the new circumstances but if the insidersperceive the outsiders to be invaders the prefrontal cortexis supplanted by the amygdala which induces a fightingreaction and tolerance is suppressed

Although with assumptions (i) and (ii) our modelaccounts for the processes that affect individual insider opin-ion the expansion of RW populism can become extremelyrapid when insiders are influenced by their peers This well-documented phenomenon in human interactions is further

accelerated when social media is added Thus the spread ofRW populism can be highly nonlinear much like the spreadof a highly contagious disease We include this nonlinear col-lective spreading mechanism in our third model assumption

Assumption iii (mutual insider contagion) At any givenmoment 119905 an insider agent 119894 with 119870119894 connections to otherinsiders turns to RW populism at rate 11990110158401015840 if at time 119905 minus 1this agent has at least119872119894 = 1198701198942 RW populist supporters intheir neighborhood Note that for simplicity factor 12 playsa role analogous to the tolerance parameter in assumption(ii) Because of connections between insider agents whenRW populism emerges anywhere in the network the populistmovement is able to spread like a contagion This collectivespreading indicates that insider agents can become RWpopulist supporters even when there are no outsiders inthe immediate neighborhood Thus some EU countries withalmost no immigration are opposed to accepting even a smallgroup of immigrants This may have affected the outcomeof the recent US presidential election in which the winningcandidate was often ridiculed in the mainstream mediamdashanattitude represented in our model by assumption (i)

The model here assumes imitative interactions withouttesting the validity of such an assumption However thisassumption could be tested by themethod of Agliari et al [53]if the required data were available The authors use the toolsfrom statistical mechanics to determine from data the natureof interactions in social customs such as local and mixedmarriages in Italy and neighboring European countries Thefraction of actualmarriages satisfying a given condition scalesdifferently with the fraction of all possible couples that satisfythe same condition depending on the nature of interactionsbetween agents If the interactions are independent (one-body model) the scaling is linear whereas if the interactionsare imitative (two-body model) the scaling is square-rooted

We now turn to the simulation results and their impli-cation Figure 4 shows how in a network of 5000 agents thefraction of RW populist supporters increases when there isa constant inflow of outsiders here 119869 = 2 per month Thisinflow is annually approximately 05 of the total populationwhich is slightly more than the threshold value impliedin Figure 3(b) Simulations with 119869 = 2 per month aredesigned to emulate the rapid limit in extension (B) describedabove After approximately 37 years of rapid globalization thenetwork reaches a tipping point and abruptly shifts to amodedominated by RW populism RW populism dominates whenmore than 50 of the network is made up of RW populistsupporters (ie 119875 gt 05 where 119875 denotes the fraction of RWpopulists) The threshold is 50 because in a democracy themajority rules

Figure 4 also shows how the simulated network underassumptions (i)ndash(iii) responds to shocks (red curve) At thisstage extension (B) does not yet operateThe constant annualinflow of outsiders 119869 = 2 is supplemented by two eventsat times 1199051 and 1199052 when 119869 = 200 The state of the networkcharacterized by the proportion of RWpopulists (119875) exhibitsa much stronger response at 1199052 than at 1199051 although the shockinflow (119869 = 200) is the same This occurs because at 1199052 the

8 Complexity

08

00

02

J = 2

t1

J = 2 except at

t2

Change in tolerance

500

04

06

Time

Prob

abili

ty o

f RW

pop

ulism

t1 amp t2

Figure 4 Nonlinearity a tipping point and the rise of right-wingpopulism Using a network of 119873 = 5000 agents each with anaverage of 15 connections we examine the effect of a constantinflow of outsiders at rate 119869 = 2 at each time step In this setupthe total number of outsiders at any moment in time is 119868(119905) =sum119894 119869119894 = ⟨119869⟩119905 As the fraction of outsiders 119891119868 = 119868(119873 + 119868)approaches the tolerance parameter 1198911015840119868 = 015 the presence of atipping point causes the fraction of RW populist supporters to startincreasing nonlinearly and eventually undergo a sudden jump (iea discontinuous change) at about 37 years (450 months) into thesimulation (black curve) The sudden jump happens much earlierif the inflow of outsiders experiences shocks at times 1199051 and 1199052at which 119869 = 200 outsiders enter the network In particular asthe network approaches the tipping point the effect of exactlythe same shock becomes disproportionately higher (red curve) Inthis case however the tolerance parameter is still kept constantFinally we also examine the case in which shocks at times 1199051 and1199052 affect the tolerance parameter where responsiveness is controlledby parameter 120574 = 00001 Here the second shock at 1199052 is sufficientto instantly tip the network into RW populism (green curve) Otherparameters are 119901 = 0007 120591 = 15 1199011015840 = 05 11990110158401015840 = 05 and 120572 = 0001

system is closer to the tipping point and consequently moreunstable than at 1199051 Because in real-world data the value of119869 can be biased due to estimation errors or misinterpretedinformation our results suggest that approaching the tippingpoint can be concurrent with strong nonlinear effects suchthat even a small shock can trigger a transition to a modedominated by RWpopulismThis can be evenmore explosive(in terms of 119875) if extension (B) is allowed to operate that isif the tolerance parameter 1198911015840119868 changes with 119869

Figure 4 shows a third simulation (green curve) inwhich the dynamics operate under assumptions (i)ndash(iii) withextension (B) The tolerance parameter 119891119868 thus changes with119869 as prescribed by (2) Because of the decreasing tolerancethe second shock at 1199052 can now push the system beyondthe tipping point and thus the dominance of RW populismoccurs earlier than in the two previous simulations

From the simulations in Figure 4 alone it is unclearhow much the local influence of outsiders (assumption (ii))contributes to the rise of RW populism relative to the insider

08

1

06

04

02

00

TotalLocal influence (ii)Contagion (iii)

250 500

Time

Prob

abili

ty o

f RW

pop

ulism

Figure 5 Breakdown of the causes of right-wing populism Figure 4shows that the probability of RW populism 119875 suddenly increasesas society approaches a tipping point but remains silent on theunderlying causes Here we discern between the contributionsof local outsider influence (assumption (ii)) and mutual insidercontagion (assumption (iii)) Far from the tipping point 119875 mainlyresponds to local outsider influence (ii) By contrast as the networkapproaches its tipping point mutual insider contagion (iii) takesover and accelerates the transition to RW populist dominanceParameter values are119873 = 5000 with an average degree of 15 119869 = 2119901 = 0007 120591 = 15 1199011015840 = 07 and 11990110158401015840 = 08

contagion (assumption (iii)) Figure 5 shows these contribu-tions After the initial transients fade the local influence ofoutsiders drives the increase in RW populists in the networkThe contribution of mutual insider contagion is relativelysmall until the system approaches a tipping point Near thetipping point contagion spreads rapidly and overtakes thelocal outsider influence as the main contributor to the riseof RW populism Thereafter the RW populist movement cansustain itself without support from the outside

In the regime of moderate immigration inflows (1198691015840 lt 119869 lt11986910158401015840) we can examine the dynamics of our complex networkusing themean-field theory (MFT) analytic techniqueWhenthe number of agent connections does not deviate greatlyfrom the network average the probability 119875 that a randomlychosen insider agent 119894 is an RWpopulist supporter due to anyof the processes underlying assumptions (i)ndash(iii) is

119875 = 119901lowast + 11990110158401199011 (119896119894 119898119894 119891119868) + 119901101584010158401199011 (119870119894119872119894 119875)

minus 119901lowast11990110158401199011 (119896119894 119898119894 119891119868) minus 119901lowast119901101584010158401199011 (119870119894119872119894 119875)

minus 11990110158401199011 (119896119894 119898119894 119891119868) 119901101584010158401199011 (119870119894119872119894 119875)

(4)

where the last three terms avoid double counting in accor-dance with the probability theory formula 119875(119860 cup 119861 cup 119862) =119875(119860) + 119875(119861) + 119875(119862) minus 119875(119860)119875(119861) minus 119875(119860)119875(119862) minus 119875(119861)119875(119862) forthree mutually independent events 119860 119861 and 119862 that cannotoccur simultaneously In the MFT approximation we candrop index 119894 because no single agent is markedly differentfrom the collective average Previously we set 119872 = 1198702

Complexity 9

417 882

88

200 600 800 1000 1200

Time (months)

13208

1

06

04

02

00

Prob

abili

ty o

f RW

pop

ulism

400

294

263185

555

J = 1 fI = 005

J = 1 fI = 01

J = 1 fIfI

= 015

J = 2 = 005

J = 3 fI = 005

J = 3 fI = 01

J = 3 fI = 015

J = 3 fI = 02

Figure 6 Predicting the timing of RW populism We find that for abroad range of outsider inflows (119869) and tolerance parameter values(1198911015840119868 ) (5) predicts the moment at which 119875 gt 05 in a manner thatagrees favorably with the simulation results Except for 120574 = 0 otherparameters are the same as in Figure 4

for simplicity but the peer pressure measured by the valueof the proportionality factor between 119872 and 119870 can differbetween countries or regions In addition parameters 1199011015840 and11990110158401015840 are constants only in theory The real social dynamicsare such that these parameters may change in responseto rumors political manipulations or outside shocks Ifparameters 1199011015840 and 11990110158401015840 substantially increase the value of 119875also increases thus further improving the prospects for RWpopulism dominance In our framework due to democracyrsquosmajority rule principle when 119875 approaches 05 the nonlinearprocesses embedded in assumptions (ii) and (iii) cause asudden transition to RW populist mode

Because a theoretical model is more useful if it haspredictive power [54 55] we show how a network of agentsunder assumptions (i)ndash(iii) leads to a simple formula for thetiming at which RW populism starts to dominate

119905119905ℎ =1198731198911015840119868

119869 (1 minus 1198911015840119868 )minus 119868 (0)

119869 (5)

Gaining this result involves three steps First if the immi-gration inflow is constant then the number of outsidersin the network after 119905 time steps is 119868(0) + 119869119905 Second thetotal population size thus equals 119873 + 119868(0) + 119869119905 Finally (5)follows if the current fraction of outsiders (119868(0) + 119869119905)(119873 +119868(0) + 119869119905) is equated with the critical parameter 1198911015840119868 Figure 6shows that for a number of immigration inflow-toleranceparameter pairs (119869 1198911015840119868 ) the simulated timing of the shift toRW populist mode (ie 119875 gt 05) fits theoretical predictionsIn conjunction with the empirical data on tolerance towardsimmigrants in the EU countries the formula in (5) could beused to provide an estimate of when a given country mightbe approaching a possible tipping point to RW populismdominance

Numerical simulations allow us to examine not only iso-lated single networks but also the interdependence betweentwo or more networks Note that there is a potential for a cas-cade effect when an RW populist movement in one networkcauses the rise of RW populist movements in other networksThis cascade effect is growing in relevance because expandingglobalization is causing countries to become increasinglysimilar to each another and this similarity increases in suchsupranational organizations as the EU in which bordersbetween nation states are rapidly fading

To examine how interdependence affects the rise of RWpopulism we set up two random economically equivalentER networks (equal 119901 in the model) with different tolerancelevels towards outsiders (different 1198911015840119868 in the model) Tothe usual intraconnections within each network we addinterconnecting agents that linkwith their counterparts in theother network Apart from this addition we retain the samemodel assumptions as previously held

To examine the effects of interconnectedness we first runnumerical simulations of two independent networks withoutinterconnections [see Figure 7(a)] As expected when theinflow level of outsiders is the same (119869 = 2) the networkwith a higher tolerance parameter (11989110158401198681 = 04) reaches thetipping point much later than the network with a lowertolerance parameter (11989110158401198682 = 02) When the networks areinterconnected and the more tolerant network experiencesan increased inflow (1198692 = 4) and a shock (119869 = 500) at time1199051 = 500 its susceptibility to RW populism increases andit also affects the other network and shortens the time oftransition to RW populism [see Figure 7(b)] It would appearthat countries do not want to be the first to cross the line butin an interconnected world being second is easier

4 Discussion and Conclusion

Why some countries (eg the ex-socialist EU countries)strongly oppose receiving immigrants while others (eg theUSA) have a long history of receiving immigrants is animportant topic in the social sciences and more recently amajor issue for the EU Perhaps receiving immigrants hasbeen an ongoing pattern in the USA because there is nosingle dominant ethnicity and thus a clear distinction ismadebetween USA national identity and the country of originof its citizens In addition because USA citizens are peoplefrom all over the world there is no single dominant religiousethnic or cultural group that can organize and threaten theestablished social order In France we find the oppositeThereis a large group of immigrants with different language andreligion from the French majority whose presence can instillfear among themajority Fear exacerbated by an inflow rate ofimmigrants that exceeds the rate of their integration can leadto a volatile situation one that is often resolved in one of twoways Either there is an immigrant uprising as exemplifiedby the Visigoth immigrants and their ex-Roman commanderAlaricwhoplunderedRome in 410 or themajority populationsuppresses the inflow of immigrants Currently this secondoption is often accomplished by supporting populist right-wing parties

10 Complexity

08

1

00

06

0402

J1 = 2

J2 = 2

200015001000500

Prob

abili

ty o

f RW

pop

ulism

Time

(a)

08

1

00

06

2000

0402

J1 = 2

J2 = 4

Time15001000500Pr

obab

ility

of R

W p

opul

ism

(b)

Figure 7 Interconnected networks or why ldquosomebody elsersquos problemrdquo easily turns into ldquomy problemrdquo In (a) we show the case when there areno interlinks between networks The tolerance parameters between the two networks differ 1198911198681 = 02 and 1198911198681 = 04 while the inflows intoboth networks are the same 1198691 = 1198692 = 2 (b) More tolerant network is now exposed to a higher inflow 1198692 = 4 and a shock at 1199051 = 500 Theaverage number of connections for intraconnections (interconnections) in both networks equals 15 (10) The other parameters are the sameas in Figure 4

Although globalization was conceived to allow capitaland labor to move freely across national borders real-worldglobalization is affected by a multitude of noneconomicfactors such as ethnicity culture and religion With so manyfactors at play globalization is an enormously complex pro-cess and often noneconomic factors do not align with purelyeconomic factors This misalignment can lead to frustrationsthat feed populist movements By opposing the collaborationthat comes with globalization populist movements act asa feedback mechanism that pushes the general populationtowards becoming more ideologically rigid and this inturn further accelerates the populism [56] A strengthenedpopulist movement can also trigger tectonic shifts in worldaffairs as exemplified by BREXIT in the UK and the recent2016 US presidential election

Problems in a globalized world rarely confine themselvesto one place Interdependencemakes the developed countriesmore alike and synchronizes their social dynamics Thissynchronization can cause political shifts in one country tospill over into other countries and this is what enables the riseof RW populism to spread across large regions of the worldAfter BREXIT and the 2016 US presidential election politicalelites should not expect to continue business as usual Beingthe first to adopt amajor political shift with a high probabilityof negative economic consequences is difficult but once thatline has been crossed the interdependence of globalizationmakes cascades (domino effects) an active possibility No onewants to be first but many are ready to be second

The tipping point that brings on the rise of RW populismmay be reached more quickly when voters face a binarychoice for example the ldquoyesrdquo or ldquonordquo choice in the UKBREXIT referendum and the two major-party candidates inthe recent US presidential election In both cases the populistoption secured a narrow victory As mentioned above thepolls indicate that the populist party in Austria can expect toreceive 34 of the votes but the populist party candidate inthe presidential race can expect close to 51 A simple wayto understand these percentages is to assume that politicalattitudes of voters are approximately symmetrical in their

distribution across the political spectrum from left to rightConsequently if leftist voters comprise 120595119871 of the totalpopulation then RW populist voters will maintain a similarpresence that is 120595119877 asymp 120595119871 Facing this binary choice thecentrist voters have no one representing their views and thusare likely to vote evenly between the two available optionsAn implication is that when 120595119871 asymp 120595119877 asymp 33 even aslight (statistically significant) imbalance in favor of120595119877over120595119871 can tip the society towards RW populism In Austria120595119877 asymp 36 seems to be sufficient to make the RW populistcandidate a front runner in the presidential race

The final outcome of the battle between the conflictingfactors surrounding globalization will almost surely havetremendous economic implications If a country approachesits tipping point how will the ensuing volatility affect itslong-term credit rating If a domino effect cascades acrossa large region of the EU or the entire EU how will thataffect the Euro and the common banking system Withouta proper resolution of the migrant crisis what will be theimpact on systemic risk In an attempt to shed light onsome of the factors and underlying processes that affectthe success of globalization we offer an empirically backedtheoretical model of the rise of RW populism in response tounsustainable immigration inflows

Our model emphasizes the need for controlled globaliza-tion in which immigration inflows into a society are balancedwith the ability of the society to integrate the immigrantsThis ability is arguably improved when immigrants mixwith the native population which is a principle practicedin Singapore where immigrant hubs are discouraged andtenants in government-built housing (comprising 88 ofall housing) must be of mixed ethnic origin [20] Becausetolerance towards immigrants is conditional when immigra-tion inflows overshadow integration rates the society can betipped into RWpopulism an intolerant mode of functioningThe rise of RW populism can occur because elections areby their nature stochastic and they resemble a mathematicalrandom walker Left to its own devices a random walkerwill eventually hit an absorbing barrier Here this barrier

Complexity 11

of course is a metaphor for the demise of globalizationironically at the very hands of the progressive system (iedemocracy) thatmade globalization possible in the first place

Disclosure

A version of this work was presented in Warsaw at ldquoEcono-physics Colloquium 2017rdquo

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

The authors are grateful to S Galam and T Lipic forhelpful suggestions Boris Podobnik received the supportfrom Slovenian Research Agency (ARRS) via Project J5-8236and from the University of Rijeka Boris Podobnik and H EStanley received support from the Defense Threat ReductionAgency (DTRA) theOffice ofNaval Research (ONR) and theNational Science Foundation (NSF) (Grant CMMI 1125290)Marko Jusup was supported by the Japan Science andTechnology Agency (JST) Program to Disseminate TenureTracking System

References

[1] H Grubel and A Scott ldquoThe international flow of humancapitalrdquo American Economic Review vol 56 pp 268ndash274 1966

[2] G C Hufbauer and K Suominen Globalization at Risk Chal-lenges to Finance and Trade Yale University Press 2010

[3] F Docquier and H Rapoport ldquoGlobalization Brain Drain andDevelopmentrdquo Journal of Economic Literature vol 50 pp 681ndash730 2012

[4] D Rodrik The Globalization Paradox Democracy and theFuture of the World Economy W W Norton amp Company NewYork USA 2011

[5] B R Chiswick ldquoThe Effect of Americanization on the Earningsof Foreign-born Menrdquo Journal of Political Economy vol 86 no5 pp 897ndash921 1978

[6] G J BorjasHeavenrsquosDoor Immigration Policy and theAmericanEconomy Princeton University Press 1999

[7] G J Borjas ldquoThe economics of immigrationrdquo Journal ofEconomic Literature vol 32 pp 1667ndash1717 1994

[8] G J Borjas ldquoAssimilation Changes in Cohort Quality and theEarnings of Immigrantsrdquo Journal of Labor Economics vol 3 no4 pp 463ndash489 1985

[9] D Swank and H G Betz ldquoGlobalization the welfare stateand right-wing populism in Western Europerdquo Socio-EconomicReview vol 1 no 2 pp 215ndash245 2003

[10] J Gibson and D McKenzie ldquoEight Questions about BrainDrainsrdquo Journal of Economic Perspectives vol 25 no 3 pp 107ndash128 2011

[11] H Betz ldquoPolitics of Resentment Right-Wing Radicalism inWest Germanyrdquo Comparative Politics vol 23 no 1 pp 45ndash601990

[12] R Knight ldquoHaider the freedom party and the extreme right inAustriardquo Parliamentary Affairs vol 45 no 3 pp 285ndash299 1992

[13] P Fysh and J Wolfreys ldquoLe pen the National Front and theextreme right in Francerdquo Parliamentary Affairs vol 45 no 3pp 309ndash326 1992

[14] G Voerman and P Lucardie ldquoThe extreme right in the Nether-lands The centrists and their radical rivalsrdquo European Journalof Political Research vol 22 no 1 pp 35ndash54 1992

[15] WD Chapin ldquoExplaining the electoral success of the new rightThe German caserdquoWest European Politics vol 20 no 2 pp 53ndash72 1997

[16] J M Smith ldquoDoes crime pay Issue ownership political oppor-tunity and the populist right in Western Europerdquo ComparativePolitical Studies vol 43 no 11 pp 1471ndash1498 2010

[17] B Bonikowski and PDiMaggio ldquoVarieties of American PopularNationalismrdquo American Sociological Review vol 81 no 5 pp949ndash980 2016

[18] D Rodrik ldquoHow far will international economic integrationgordquo Journal of Economic Perspectives vol 14 no 1 pp 177ndash1862000

[19] B Davis ldquoDespite his heritage prominent economist backs im-migration cutrdquo The Wall Street Journal 1996 httpswwwhksharvardedufsgborjaspublicationspopularWSJ042696pdf

[20] B Podobnik M Jusup Z Wang and H E Stanley ldquoHow fearof future outcomes affects social dynamicsrdquo Journal of StatisticalPhysics vol 167 no 3-4 pp 1007ndash1019 2017

[21] R Reuveny ldquoClimate change-induced migration and violentconflictrdquo Political Geography vol 26 no 6 pp 656ndash673 2007

[22] T C Schelling ldquoDynamic models of segregationrdquo Journal ofMathematical Sociology vol 1 pp 143ndash186 1971

[23] R Axelrod ldquoThe dissemination of culture a model withlocal convergence and global polarizationrdquo Journal of ConflictResolution vol 41 no 2 pp 203ndash226 1997

[24] M McPherson L Smith-Lovin and J M Cook ldquoBirds of aFeather Homophily in Social Networksrdquo Annual Review ofSociology vol 27 pp 415ndash444 2001

[25] T Antal P L Krapivsky and S Redner ldquoDynamics of socialbalance on networksrdquo Physical Review E vol 72 Article ID036121 2005

[26] S Marvel J Jon Kleinbergb D Robert R D Kleinberg andS Strogatz ldquoContinuous-Time Model of Structural BalancerdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 108 no 5 pp 1771ndash1776 2011

[27] C Gracia-Lazaro L M Flora and Y Moreno ldquoSelectiveAdvantage of Tolerant Cultural Traits in the Axelrod-SchellingModelrdquo Physical Review E vol 83 Article ID 056103 2011

[28] F Aguiar and A Parravano ldquoTolerating the IntolerantHomophily Intolerance and Segregation in Social BalancedNetworksrdquo Journal of Conflict Resolution vol 59 no 1 pp 29ndash50 2015

[29] M Lim R Metzler and Y Bar-Yam ldquoGlobal pattern formationand ethniccultural violencerdquo Science vol 317 no 5844 pp1540ndash1544 2007

[30] R M Dancygier Immigration and Conflict in Europe Studies inComparative Politics Cambridge University Press 2010

[31] R M Dancygier and D D Laitin ldquoImmigration into EuropeEconomic Discrimination Violence and Public Policyrdquo AnnualReview of Political Science vol 17 pp 43ndash64 2014

[32] S Galam and M A Javarone ldquoModeling Radicalization Phe-nomena in Heterogeneous Populationsrdquo PLoS ONE vol 11Article ID e0155407 2016

4 Complexity

Presidential election

05

04

03

02

01

40

30

30

2016

2015

2014

10

10

20

20

50

0

R

W v

oter

s

VotersInflow

Imm

igra

tion

inflo

w (J

p

opul

atio

n)

Time (months)

(a)

10

6

2

14

Time (months)

2016201520142012

4812 24 360

1

2

3

4

5

6

7

0

R

W v

oter

s

Imm

igra

tion

inflo

w(J

times104)

VotersInflow

(b)

10

15

5

0

20

1 2 3 4 5 6

R

W v

oter

s

Immigration inflow (J times 104)

(c)

Figure 2 Immigration affects the support for right-wing populism II (a) An unprecedented inflow of immigrants into Austria coincided witha steady increase in the fraction of RW populist voters A solitary black dot represents the results of Austrian presidential election in May2016 in which an RW populist candidate secured almost 50 of votes This election shows that even after the record immigrant inflow atthe end of 2015 had subsided a decreasing trend in the number of immigrants that enter Austria did not automatically translate into lowersupport for the RWpopulist political option that is RWpopulism seems to bemore than just a craze (b) In Germany the increasing inflow ofimmigrants (monthly data [34 35]) rather clearly coincided with the increasing support for an RW populist party (c) A significant regressionemerges when the German case is presented as a scatter plot between the inflow of immigrants and the percentage of far-right voters

Figure 3 indicates that the interplay of factors influencingthe rising popularity of RW populism is more complex than asimple bivariate regression and thus we turn to econometricanalysis and multivariate regression Using the results of thesimple regression in Figure 1 we assume that the fractionof RW voters (response variable RW119905) in a given countryis determined by the fraction of immigrants (IM119871119905 ) livingin the country We further assume that the fraction of RWvoters depends on the overall inflow of immigrants into theEU (IMEU

119905 ) calculated relative to the total EU populationThis variable represents an ldquoimmigration shockrdquo in themodel To also take into account the possibility that violentincidents involving immigrants could affect the popular votewe include the total number of injuries (119868) and casualties

(119863) involving immigrants recorded across the EU [39] in themodel Additionally we take into account the unemploymentrate (119880) that might also affect the popular vote Finally weadd a variable119872119894119905 = (1minusMIPEX100) in whichMIPEX is themigrant integration policy index [40] a proxy for the integra-tion ratemdashthe larger the MIPEX the better the integration

We perform econometric analysis using a pooled time-series cross-section (TSCS) method that combines the cross-sectional data on multiple countries Here the number ofcountries is 119873 = 10 entirely consisting of the so-called olddemocracies Germany France Austria Netherlands Swe-den Norway Denmark Finland Greece and Italy Becausefor each country there are 119879 observations along the temporaldimension the entire dataset has119873 times 119879 = 370 observations

Complexity 5

FRA (116)

GRC (89)

HUN (06)

BGR (12)ESP

PRT

SWE (159)

DK (99)UK (124)

ITA (94)

FIN (54)

05 1

2

minus2

4

0

0

Annual immigration inflow ( population)

Ann

ual

chan

ge in

RW

vot

ers NOR (138) DEU (119)

BEL (104) AUT (157)

NL (117)

Figure 3 Immigrant inflows and the popularity of right-wing populistmovementsmdasha nonlinear threshold Shown is the annualized immi-grant inflow into a given country (horizontal axis) as a percentage ofthat countryrsquos population as well as the corresponding percentagechange in RW populist votes (vertical axis) In parentheses are thefractions of immigrants in the total population of the correspondingcountry For a group of countries in which the annualized increasein the percentage of RW voters exceeded 2 this increase isvirtually independent of the inflow of immigrants Such a result mayreflect the EUrsquos political organization that is the lack of internalborders whereby if one country decides to accept immigrants thedecision may have repercussions for all the other member statesWe also observe a threshold indicated by a dashed line at whichthe immigrant inflow into a given country is sufficiently high toinvariably provoke an increase in the percentage of RW populistvoters In the model construction this threshold suggests 120572 = 0004on an annual basis

We have an extra index 119894 = 1 2 119873 that refers to a cross-sectional unit giving

RW119894119905 = 1205730 + 120573119871IMIM119871119894119905 + 120573

EUIM IMEU119894119905 + 120573

ter119863 119863119894119905 + 120573

ter119868 119868119894119905

+ 120573119880119880119894119905 + 120573119872119872119894119905 + 119890119894119905(1)

where 119890119905 is the random errorTable 1 shows the results of the TSCS regression model

indicating that the fraction of immigrants in the generalpopulation and the immigration inflow into the entire EUare the significant explanatory variables In addition thecoefficient 120573119871IM is not only significant but also higher thanunity Perhaps surprisingly the response variable is notsignificantly affected by the number of injuries and casualtiesin violent incidents involving immigrants We also show thatunemployment insignificantly affects the popular vote

The aforementioned survey data suggest that not everyEU nation is equally tolerant to immigrants We believe thata proxy for this tolerance can be a fraction of RW votes justafter the Second World War (RW0) when the fractions ofimmigrants were considerably smaller than nowadays Notethat by far the largest fraction of RW votes was recordedin Austria 1167 Table 2 show the results of the TSCSregression model when RW0 is included We find that this

Table 1 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression as defined in (1) Test statistics Wald1205942(6) = 14787 and Prob gt 1205942 = 0000

Coeff Std err 119911 119875 gt |119911|120573119871IM 217 0527 412 0000120573EUIM 4311 532 810 0000120573119863ter minus37119890 minus 04 49119890 minus 04 minus078 0437120573119868ter 21119890 minus 04 16119890 minus 04 128 0202120573119872 0208 0212 098 0325120573119880 minus0167 0348 minus048 06301205730 minus0298 0103 minus289 0004

Table 2 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression as defined in (1) Test statistics Wald1205942(6) = 15154 and Prob gt 1205942 = 0000

Coeff Std err 119911 119875 gt |119911|120573119871IM 132 0445 296 0003120573EUIM 4559 535 852 0000120573ter119863 minus42119890 minus 04 49119890 minus 04 minus086 0392120573ter119868 22119890 minus 04 16119890 minus 04 138 0167120573119872 minus0019 02 minus009 0926120573119880 minus0015 0289 minus005 0959RW0 0904 0562 161 01081205730 minus0139 0093 minus150 0133

Table 3 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression Test statistics Wald 1205942(5) = 1424and Prob gt 1205942 = 0000

Coeff Std err 119905 stat 119875 gt 119905120573119871IM 217 0466 465 0000120573EUIM 4672 501 933 00001205730 minus0235 0056 minus415 0000

new regression insignificantly contributes to modern RWpopulism

Table 3 shows the results of the TSCS regression modelwithoutMIPEX unemployment and violent incidents Inter-estingly with the total inflow of immigrants into the EUof 100000 on a monthly basis coefficient values in Table 3suggest that around 30 of immigrants in the total popu-lation of a country are sufficient to cause larger than 50support for RW populist parties which corresponds to theresult obtained by a linear fit in Figure 1

The analyzed data do not indicate whether the rise of RWpopulism is a transient phenomenon or a longer-term changein political orientation One factor is the persistence of votermemory In the October 2015 local election in Vienna theRW populist party won 33 of the popular vote During thepresidential election a fewmonths later themigrant crisis hadreached a peak and the RW populist movement candidatesecured almost 50 of the votes narrowly losing to a leftistrival These results have been contested and a new electionin December 2016 brought around 48 to the RW populistcandidate indicating that this rise in RW populism is not

6 Complexity

short-term This phenomenon has been described in theliterature Betz [41] describes how a substantial increase inthe number of refugees and illegal immigrants in Europeancountries during the 1980s provoked a wave of radical RWpopulism Following these events in the early 1990s thereremained between 11 and 14 of Europeans who felt thepresence of other nationalities races and religions to beunsettling [41]

3 Model

Human interactions are often heterogeneous and prone toabrupt nonlinear responses Because this characterizes therise of the RW populist party and its RW candidate in theAustrian elections such linear approaches as the regressionin (1) yield only partial results A useful intuition is gained bythinking about the election system in a democratic country asa randomwalker Even in a bipolar political system not everyleft government is equally left and similarly not every rightgovernment is equally right Therefore as a result of electionthe randomwalker can move either left or right with the sizeof this move depending on the standard deviation Withoutlimitations such that democracy and the election processpossess an unlimited tolerance for every political option aftera long enough time the random walker is bound to endup either extremely right or extremely left Both of theselimits exhibit ideological rigidity likely to substantially reducethe level of democracy and tolerancemdashin agreement withPopper ldquounlimited tolerance must lead to the disappearanceof tolerancerdquo [42] Thus RW populism can be considered asjust one of the two randomwalk limits However the randomwalker intuition does not explain why a society would moveleft or right nor does it provide a microscopic interpretationof the societal processes at the individual agent level

To mechanistically characterize the rise of RW populismand account for the existence of tipping points in socialdynamics we use a complex network approach [20 43ndash45]Complex network science is able to emulate heterogeneity inhuman interactions and goes beyond capturing the dynam-ics near tipping points Heterogeneity is important whenconsidering immigration and integration issues becauseimmigrants in order to sustain their identity live in ldquohubsrdquothat make them more difficult to integrate than immigrantsmixed into the native population

We construct our model by setting a constant numberof native ldquoinsiderrdquo agents and arranging them in an Erdos-Renyi random network of business and personal contactsImmigrant ldquooutsiderrdquo agents are then added to the networkEach insider notices the percentage of outsiders in theirneighborhood and based on this percentage decides whetherto be supportive of globalization or RW populism Insideragents get information from their neighborhood and theinteraction is local and this data is essential in understandingtipping points [46] but there are also other relevant nonlocalinteractionsThere is furthermore the factor that informationcan be misperceived or misinterpreted In the following weformalize these concepts with three assumptions

Assumption i (media and economic influences) At eachone-month period of time 119905 insider agents are influencedby media at a probability rate 119901 and remain influencedfor a period 120591 We assume that this influence transformsinsiders into RW populist supporters Although media canaffect insiders in both directions we focus on the growthrate and disregard negative values of 119901 The effect of themedia is global and hearing that immigration is occurringcan transform some insiders into RW populist supportersirrespective of their local situation For example most likelyas the effect of media coverage of immigration to boththe EU and the UK during the BREXIT referendum mostUK districts with low immigration voted mainly for Leave[47] The media alongside a lower level of tolerance toimmigration can be an important reason why in ex-socialistcountries the large fraction of voters oppose receiving evena low overall fraction of immigrants The probability rate119901 can also reflect such economic factors as unemploymentThus we use equation [43] 119901lowast = 1 minus exp(minus119901120591) to calculatethe probability that a randomly chosen insider agent is beinginfluenced by the media

Assumption ii (local influence of outsiders) In local electionsin Greece in November 2010 although the far-right GoldenDawn party received only 53 of the vote in some neighbor-hoods of Athens with large immigrant communities the partywon nearly 20 [48] This suggests that contacts betweeninsiders and outsiders do matter Within our network modelwe maintain a constant number of 119873 insiders and then add119868(0) outsidersWe then increase the number of outsiders withan inflow 119869119905 at each moment 119905 (representing one month)We randomly place newly arriving outsiders between insideragents both of whom initially have an average number ofconnections (ie a degree) 119896 The total number of outsiders119868(119905) is obtained by summing the monthly 119869 values accordingto 119868(119905) = 119868(0) + sum119905119904=1 119869119904 At any given moment the fractionof outsiders will equal 119891119868(119905) = 119868(119905)(119873 + 119868(119905)) To accountfor the effect of contacts between insiders and outsiders weassume that any insider agent 119894with 119896119894 total connections turnsto RWpopulism at a rate 1199011015840 when this agent is surrounded byat least 119898119894 = 1198911015840119868119896119894 outsiders [43 46 49] where 0 lt 1198911015840119868 lt 1is a constant model parameter quantifying the toleration ofinsiders This assumption merits a few additional comments

First the probability that randomly chosen insider agent 119894with 119896119894 connections is surrounded by119898119894 outsiders and there-fore prone to RW populism is 1199011(119896119894 119898119894 119891119868) equiv sum119896119894119895=119898119894 119891

119895119868 (1 minus

119891119868)119896119894minus119895 ( 119896119894119896119894minus119895 ) In this formula 119891119868 is the true current state

of the network However the information may be biasedand cause insiders to think there are more outsiders than isactually the case If the bias is Δ119891119868 then 1199011(119896119894 119898119894 119891119868 + Δ119891119868) gt119901(119896119894 119898119894 119891119868) This increased probability 119901(119896119894 119898119894 119891119868 + Δ119891119868)implies that the tolerance parameter 1198911015840119868 must decrease byamount Δ1198911015840119868 which we estimate using condition 119901(119896119894 119898119894 minusΔ1198911015840119868119896119894 119891119868) = 119901(119896119894 119898119894 119891119868 + Δ119891119868) An implicit assumption hereis that all insider agents are equally tolerant to immigrantsbecause the tolerance parameter 1198911015840119868 is defined as a globalnetwork property rather than an individual agent property

Complexity 7

An alternative would be to assume a distribution of tolerancelevels in which case 1198911015840119868 would represent the mean [20]

We now expand assumption (ii) with extension (A) whenthe immigration inflow 119869 is below some threshold 1198691015840 thesociety becomes more tolerant When 119869 lt 1198691015840 at a giventime moment the tolerance parameter 1198911015840119868 increases by anamount Δ1198911015840119868 = 120575 gt 0 there is a balance between immigra-tion and immigrant integrationmdashoutsiders are successfullyintegratedmdashand insiders are able to acclimate to the changesin their society Figure 2(b) shows that the 1198691015840 value forGermany is approximately 10000 people per month Beingable to accurately determine the maximum 1198691015840 value is highlyrelevant to the success of globalization According to ourmodel immigration and integration can be the inevitableconsequences of globalization when 119869 lt 1198691015840 If this is notthe case globalization will be threatened by the rise of RWpopulism the democratic system will enter an intolerantmode and cooperation between nations will be downgradedon the list of political priorities

Empirical evidence suggests that we add an oppositeextension (B) as the inflow of outsiders 119869 crosses somethreshold 11986910158401015840 (which does not need to be equal to 1198691015840)society becomes less tolerant Mathematically extension (B)indicates that when 119869 gt 11986910158401015840 the tolerance parameter 1198911015840119868decreases by

Δ1198911015840119868 = minus120574119869 (2)

Using the econometric models in (1) we decrease thetolerance parameter proportional to inflow 119869 where 120574 isa proportionality coefficient expressing the sensitivity ofinsiders to high levels of outsider inflow Figure 3(b) showsthreshold 11986910158401015840 in terms of total population that is 11986910158401015840 = 120572119873where 120572 is another proportionality coefficient Figure 3(b)estimates the 120572 value when all of the EU countries are hitby the migrant crisis The dashed line is annual inflow belowwhich countries have a mixed response to immigration andabove which support for RW populism increases Because120572 = 11986910158401015840119873 and using Figure 3(b) 1211986910158401015840119873 asymp 0004 we obtain

120572 asymp 000033 (3)

Extensions (A) and (B) are opposite limiting cases onein which immigration is slow and the other in whichimmigration is rapid Apart from the empirical evidencethat these extensions are needed brain science for exampleoffers a physiological interpretation political attitudes have acounterpart in brain structure [50ndash52] If outsiders increaseat a rate and in a manner perceived as controllable byinsiders the prefrontal cortex of the human brain responsiblefor decision making and for moderating social behavioracclimates to the new circumstances but if the insidersperceive the outsiders to be invaders the prefrontal cortexis supplanted by the amygdala which induces a fightingreaction and tolerance is suppressed

Although with assumptions (i) and (ii) our modelaccounts for the processes that affect individual insider opin-ion the expansion of RW populism can become extremelyrapid when insiders are influenced by their peers This well-documented phenomenon in human interactions is further

accelerated when social media is added Thus the spread ofRW populism can be highly nonlinear much like the spreadof a highly contagious disease We include this nonlinear col-lective spreading mechanism in our third model assumption

Assumption iii (mutual insider contagion) At any givenmoment 119905 an insider agent 119894 with 119870119894 connections to otherinsiders turns to RW populism at rate 11990110158401015840 if at time 119905 minus 1this agent has at least119872119894 = 1198701198942 RW populist supporters intheir neighborhood Note that for simplicity factor 12 playsa role analogous to the tolerance parameter in assumption(ii) Because of connections between insider agents whenRW populism emerges anywhere in the network the populistmovement is able to spread like a contagion This collectivespreading indicates that insider agents can become RWpopulist supporters even when there are no outsiders inthe immediate neighborhood Thus some EU countries withalmost no immigration are opposed to accepting even a smallgroup of immigrants This may have affected the outcomeof the recent US presidential election in which the winningcandidate was often ridiculed in the mainstream mediamdashanattitude represented in our model by assumption (i)

The model here assumes imitative interactions withouttesting the validity of such an assumption However thisassumption could be tested by themethod of Agliari et al [53]if the required data were available The authors use the toolsfrom statistical mechanics to determine from data the natureof interactions in social customs such as local and mixedmarriages in Italy and neighboring European countries Thefraction of actualmarriages satisfying a given condition scalesdifferently with the fraction of all possible couples that satisfythe same condition depending on the nature of interactionsbetween agents If the interactions are independent (one-body model) the scaling is linear whereas if the interactionsare imitative (two-body model) the scaling is square-rooted

We now turn to the simulation results and their impli-cation Figure 4 shows how in a network of 5000 agents thefraction of RW populist supporters increases when there isa constant inflow of outsiders here 119869 = 2 per month Thisinflow is annually approximately 05 of the total populationwhich is slightly more than the threshold value impliedin Figure 3(b) Simulations with 119869 = 2 per month aredesigned to emulate the rapid limit in extension (B) describedabove After approximately 37 years of rapid globalization thenetwork reaches a tipping point and abruptly shifts to amodedominated by RW populism RW populism dominates whenmore than 50 of the network is made up of RW populistsupporters (ie 119875 gt 05 where 119875 denotes the fraction of RWpopulists) The threshold is 50 because in a democracy themajority rules

Figure 4 also shows how the simulated network underassumptions (i)ndash(iii) responds to shocks (red curve) At thisstage extension (B) does not yet operateThe constant annualinflow of outsiders 119869 = 2 is supplemented by two eventsat times 1199051 and 1199052 when 119869 = 200 The state of the networkcharacterized by the proportion of RWpopulists (119875) exhibitsa much stronger response at 1199052 than at 1199051 although the shockinflow (119869 = 200) is the same This occurs because at 1199052 the

8 Complexity

08

00

02

J = 2

t1

J = 2 except at

t2

Change in tolerance

500

04

06

Time

Prob

abili

ty o

f RW

pop

ulism

t1 amp t2

Figure 4 Nonlinearity a tipping point and the rise of right-wingpopulism Using a network of 119873 = 5000 agents each with anaverage of 15 connections we examine the effect of a constantinflow of outsiders at rate 119869 = 2 at each time step In this setupthe total number of outsiders at any moment in time is 119868(119905) =sum119894 119869119894 = ⟨119869⟩119905 As the fraction of outsiders 119891119868 = 119868(119873 + 119868)approaches the tolerance parameter 1198911015840119868 = 015 the presence of atipping point causes the fraction of RW populist supporters to startincreasing nonlinearly and eventually undergo a sudden jump (iea discontinuous change) at about 37 years (450 months) into thesimulation (black curve) The sudden jump happens much earlierif the inflow of outsiders experiences shocks at times 1199051 and 1199052at which 119869 = 200 outsiders enter the network In particular asthe network approaches the tipping point the effect of exactlythe same shock becomes disproportionately higher (red curve) Inthis case however the tolerance parameter is still kept constantFinally we also examine the case in which shocks at times 1199051 and1199052 affect the tolerance parameter where responsiveness is controlledby parameter 120574 = 00001 Here the second shock at 1199052 is sufficientto instantly tip the network into RW populism (green curve) Otherparameters are 119901 = 0007 120591 = 15 1199011015840 = 05 11990110158401015840 = 05 and 120572 = 0001

system is closer to the tipping point and consequently moreunstable than at 1199051 Because in real-world data the value of119869 can be biased due to estimation errors or misinterpretedinformation our results suggest that approaching the tippingpoint can be concurrent with strong nonlinear effects suchthat even a small shock can trigger a transition to a modedominated by RWpopulismThis can be evenmore explosive(in terms of 119875) if extension (B) is allowed to operate that isif the tolerance parameter 1198911015840119868 changes with 119869

Figure 4 shows a third simulation (green curve) inwhich the dynamics operate under assumptions (i)ndash(iii) withextension (B) The tolerance parameter 119891119868 thus changes with119869 as prescribed by (2) Because of the decreasing tolerancethe second shock at 1199052 can now push the system beyondthe tipping point and thus the dominance of RW populismoccurs earlier than in the two previous simulations

From the simulations in Figure 4 alone it is unclearhow much the local influence of outsiders (assumption (ii))contributes to the rise of RW populism relative to the insider

08

1

06

04

02

00

TotalLocal influence (ii)Contagion (iii)

250 500

Time

Prob

abili

ty o

f RW

pop

ulism

Figure 5 Breakdown of the causes of right-wing populism Figure 4shows that the probability of RW populism 119875 suddenly increasesas society approaches a tipping point but remains silent on theunderlying causes Here we discern between the contributionsof local outsider influence (assumption (ii)) and mutual insidercontagion (assumption (iii)) Far from the tipping point 119875 mainlyresponds to local outsider influence (ii) By contrast as the networkapproaches its tipping point mutual insider contagion (iii) takesover and accelerates the transition to RW populist dominanceParameter values are119873 = 5000 with an average degree of 15 119869 = 2119901 = 0007 120591 = 15 1199011015840 = 07 and 11990110158401015840 = 08

contagion (assumption (iii)) Figure 5 shows these contribu-tions After the initial transients fade the local influence ofoutsiders drives the increase in RW populists in the networkThe contribution of mutual insider contagion is relativelysmall until the system approaches a tipping point Near thetipping point contagion spreads rapidly and overtakes thelocal outsider influence as the main contributor to the riseof RW populism Thereafter the RW populist movement cansustain itself without support from the outside

In the regime of moderate immigration inflows (1198691015840 lt 119869 lt11986910158401015840) we can examine the dynamics of our complex networkusing themean-field theory (MFT) analytic techniqueWhenthe number of agent connections does not deviate greatlyfrom the network average the probability 119875 that a randomlychosen insider agent 119894 is an RWpopulist supporter due to anyof the processes underlying assumptions (i)ndash(iii) is

119875 = 119901lowast + 11990110158401199011 (119896119894 119898119894 119891119868) + 119901101584010158401199011 (119870119894119872119894 119875)

minus 119901lowast11990110158401199011 (119896119894 119898119894 119891119868) minus 119901lowast119901101584010158401199011 (119870119894119872119894 119875)

minus 11990110158401199011 (119896119894 119898119894 119891119868) 119901101584010158401199011 (119870119894119872119894 119875)

(4)

where the last three terms avoid double counting in accor-dance with the probability theory formula 119875(119860 cup 119861 cup 119862) =119875(119860) + 119875(119861) + 119875(119862) minus 119875(119860)119875(119861) minus 119875(119860)119875(119862) minus 119875(119861)119875(119862) forthree mutually independent events 119860 119861 and 119862 that cannotoccur simultaneously In the MFT approximation we candrop index 119894 because no single agent is markedly differentfrom the collective average Previously we set 119872 = 1198702

Complexity 9

417 882

88

200 600 800 1000 1200

Time (months)

13208

1

06

04

02

00

Prob

abili

ty o

f RW

pop

ulism

400

294

263185

555

J = 1 fI = 005

J = 1 fI = 01

J = 1 fIfI

= 015

J = 2 = 005

J = 3 fI = 005

J = 3 fI = 01

J = 3 fI = 015

J = 3 fI = 02

Figure 6 Predicting the timing of RW populism We find that for abroad range of outsider inflows (119869) and tolerance parameter values(1198911015840119868 ) (5) predicts the moment at which 119875 gt 05 in a manner thatagrees favorably with the simulation results Except for 120574 = 0 otherparameters are the same as in Figure 4

for simplicity but the peer pressure measured by the valueof the proportionality factor between 119872 and 119870 can differbetween countries or regions In addition parameters 1199011015840 and11990110158401015840 are constants only in theory The real social dynamicsare such that these parameters may change in responseto rumors political manipulations or outside shocks Ifparameters 1199011015840 and 11990110158401015840 substantially increase the value of 119875also increases thus further improving the prospects for RWpopulism dominance In our framework due to democracyrsquosmajority rule principle when 119875 approaches 05 the nonlinearprocesses embedded in assumptions (ii) and (iii) cause asudden transition to RW populist mode

Because a theoretical model is more useful if it haspredictive power [54 55] we show how a network of agentsunder assumptions (i)ndash(iii) leads to a simple formula for thetiming at which RW populism starts to dominate

119905119905ℎ =1198731198911015840119868

119869 (1 minus 1198911015840119868 )minus 119868 (0)

119869 (5)

Gaining this result involves three steps First if the immi-gration inflow is constant then the number of outsidersin the network after 119905 time steps is 119868(0) + 119869119905 Second thetotal population size thus equals 119873 + 119868(0) + 119869119905 Finally (5)follows if the current fraction of outsiders (119868(0) + 119869119905)(119873 +119868(0) + 119869119905) is equated with the critical parameter 1198911015840119868 Figure 6shows that for a number of immigration inflow-toleranceparameter pairs (119869 1198911015840119868 ) the simulated timing of the shift toRW populist mode (ie 119875 gt 05) fits theoretical predictionsIn conjunction with the empirical data on tolerance towardsimmigrants in the EU countries the formula in (5) could beused to provide an estimate of when a given country mightbe approaching a possible tipping point to RW populismdominance

Numerical simulations allow us to examine not only iso-lated single networks but also the interdependence betweentwo or more networks Note that there is a potential for a cas-cade effect when an RW populist movement in one networkcauses the rise of RW populist movements in other networksThis cascade effect is growing in relevance because expandingglobalization is causing countries to become increasinglysimilar to each another and this similarity increases in suchsupranational organizations as the EU in which bordersbetween nation states are rapidly fading

To examine how interdependence affects the rise of RWpopulism we set up two random economically equivalentER networks (equal 119901 in the model) with different tolerancelevels towards outsiders (different 1198911015840119868 in the model) Tothe usual intraconnections within each network we addinterconnecting agents that linkwith their counterparts in theother network Apart from this addition we retain the samemodel assumptions as previously held

To examine the effects of interconnectedness we first runnumerical simulations of two independent networks withoutinterconnections [see Figure 7(a)] As expected when theinflow level of outsiders is the same (119869 = 2) the networkwith a higher tolerance parameter (11989110158401198681 = 04) reaches thetipping point much later than the network with a lowertolerance parameter (11989110158401198682 = 02) When the networks areinterconnected and the more tolerant network experiencesan increased inflow (1198692 = 4) and a shock (119869 = 500) at time1199051 = 500 its susceptibility to RW populism increases andit also affects the other network and shortens the time oftransition to RW populism [see Figure 7(b)] It would appearthat countries do not want to be the first to cross the line butin an interconnected world being second is easier

4 Discussion and Conclusion

Why some countries (eg the ex-socialist EU countries)strongly oppose receiving immigrants while others (eg theUSA) have a long history of receiving immigrants is animportant topic in the social sciences and more recently amajor issue for the EU Perhaps receiving immigrants hasbeen an ongoing pattern in the USA because there is nosingle dominant ethnicity and thus a clear distinction ismadebetween USA national identity and the country of originof its citizens In addition because USA citizens are peoplefrom all over the world there is no single dominant religiousethnic or cultural group that can organize and threaten theestablished social order In France we find the oppositeThereis a large group of immigrants with different language andreligion from the French majority whose presence can instillfear among themajority Fear exacerbated by an inflow rate ofimmigrants that exceeds the rate of their integration can leadto a volatile situation one that is often resolved in one of twoways Either there is an immigrant uprising as exemplifiedby the Visigoth immigrants and their ex-Roman commanderAlaricwhoplunderedRome in 410 or themajority populationsuppresses the inflow of immigrants Currently this secondoption is often accomplished by supporting populist right-wing parties

10 Complexity

08

1

00

06

0402

J1 = 2

J2 = 2

200015001000500

Prob

abili

ty o

f RW

pop

ulism

Time

(a)

08

1

00

06

2000

0402

J1 = 2

J2 = 4

Time15001000500Pr

obab

ility

of R

W p

opul

ism

(b)

Figure 7 Interconnected networks or why ldquosomebody elsersquos problemrdquo easily turns into ldquomy problemrdquo In (a) we show the case when there areno interlinks between networks The tolerance parameters between the two networks differ 1198911198681 = 02 and 1198911198681 = 04 while the inflows intoboth networks are the same 1198691 = 1198692 = 2 (b) More tolerant network is now exposed to a higher inflow 1198692 = 4 and a shock at 1199051 = 500 Theaverage number of connections for intraconnections (interconnections) in both networks equals 15 (10) The other parameters are the sameas in Figure 4

Although globalization was conceived to allow capitaland labor to move freely across national borders real-worldglobalization is affected by a multitude of noneconomicfactors such as ethnicity culture and religion With so manyfactors at play globalization is an enormously complex pro-cess and often noneconomic factors do not align with purelyeconomic factors This misalignment can lead to frustrationsthat feed populist movements By opposing the collaborationthat comes with globalization populist movements act asa feedback mechanism that pushes the general populationtowards becoming more ideologically rigid and this inturn further accelerates the populism [56] A strengthenedpopulist movement can also trigger tectonic shifts in worldaffairs as exemplified by BREXIT in the UK and the recent2016 US presidential election

Problems in a globalized world rarely confine themselvesto one place Interdependencemakes the developed countriesmore alike and synchronizes their social dynamics Thissynchronization can cause political shifts in one country tospill over into other countries and this is what enables the riseof RW populism to spread across large regions of the worldAfter BREXIT and the 2016 US presidential election politicalelites should not expect to continue business as usual Beingthe first to adopt amajor political shift with a high probabilityof negative economic consequences is difficult but once thatline has been crossed the interdependence of globalizationmakes cascades (domino effects) an active possibility No onewants to be first but many are ready to be second

The tipping point that brings on the rise of RW populismmay be reached more quickly when voters face a binarychoice for example the ldquoyesrdquo or ldquonordquo choice in the UKBREXIT referendum and the two major-party candidates inthe recent US presidential election In both cases the populistoption secured a narrow victory As mentioned above thepolls indicate that the populist party in Austria can expect toreceive 34 of the votes but the populist party candidate inthe presidential race can expect close to 51 A simple wayto understand these percentages is to assume that politicalattitudes of voters are approximately symmetrical in their

distribution across the political spectrum from left to rightConsequently if leftist voters comprise 120595119871 of the totalpopulation then RW populist voters will maintain a similarpresence that is 120595119877 asymp 120595119871 Facing this binary choice thecentrist voters have no one representing their views and thusare likely to vote evenly between the two available optionsAn implication is that when 120595119871 asymp 120595119877 asymp 33 even aslight (statistically significant) imbalance in favor of120595119877over120595119871 can tip the society towards RW populism In Austria120595119877 asymp 36 seems to be sufficient to make the RW populistcandidate a front runner in the presidential race

The final outcome of the battle between the conflictingfactors surrounding globalization will almost surely havetremendous economic implications If a country approachesits tipping point how will the ensuing volatility affect itslong-term credit rating If a domino effect cascades acrossa large region of the EU or the entire EU how will thataffect the Euro and the common banking system Withouta proper resolution of the migrant crisis what will be theimpact on systemic risk In an attempt to shed light onsome of the factors and underlying processes that affectthe success of globalization we offer an empirically backedtheoretical model of the rise of RW populism in response tounsustainable immigration inflows

Our model emphasizes the need for controlled globaliza-tion in which immigration inflows into a society are balancedwith the ability of the society to integrate the immigrantsThis ability is arguably improved when immigrants mixwith the native population which is a principle practicedin Singapore where immigrant hubs are discouraged andtenants in government-built housing (comprising 88 ofall housing) must be of mixed ethnic origin [20] Becausetolerance towards immigrants is conditional when immigra-tion inflows overshadow integration rates the society can betipped into RWpopulism an intolerant mode of functioningThe rise of RW populism can occur because elections areby their nature stochastic and they resemble a mathematicalrandom walker Left to its own devices a random walkerwill eventually hit an absorbing barrier Here this barrier

Complexity 11

of course is a metaphor for the demise of globalizationironically at the very hands of the progressive system (iedemocracy) thatmade globalization possible in the first place

Disclosure

A version of this work was presented in Warsaw at ldquoEcono-physics Colloquium 2017rdquo

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

The authors are grateful to S Galam and T Lipic forhelpful suggestions Boris Podobnik received the supportfrom Slovenian Research Agency (ARRS) via Project J5-8236and from the University of Rijeka Boris Podobnik and H EStanley received support from the Defense Threat ReductionAgency (DTRA) theOffice ofNaval Research (ONR) and theNational Science Foundation (NSF) (Grant CMMI 1125290)Marko Jusup was supported by the Japan Science andTechnology Agency (JST) Program to Disseminate TenureTracking System

References

[1] H Grubel and A Scott ldquoThe international flow of humancapitalrdquo American Economic Review vol 56 pp 268ndash274 1966

[2] G C Hufbauer and K Suominen Globalization at Risk Chal-lenges to Finance and Trade Yale University Press 2010

[3] F Docquier and H Rapoport ldquoGlobalization Brain Drain andDevelopmentrdquo Journal of Economic Literature vol 50 pp 681ndash730 2012

[4] D Rodrik The Globalization Paradox Democracy and theFuture of the World Economy W W Norton amp Company NewYork USA 2011

[5] B R Chiswick ldquoThe Effect of Americanization on the Earningsof Foreign-born Menrdquo Journal of Political Economy vol 86 no5 pp 897ndash921 1978

[6] G J BorjasHeavenrsquosDoor Immigration Policy and theAmericanEconomy Princeton University Press 1999

[7] G J Borjas ldquoThe economics of immigrationrdquo Journal ofEconomic Literature vol 32 pp 1667ndash1717 1994

[8] G J Borjas ldquoAssimilation Changes in Cohort Quality and theEarnings of Immigrantsrdquo Journal of Labor Economics vol 3 no4 pp 463ndash489 1985

[9] D Swank and H G Betz ldquoGlobalization the welfare stateand right-wing populism in Western Europerdquo Socio-EconomicReview vol 1 no 2 pp 215ndash245 2003

[10] J Gibson and D McKenzie ldquoEight Questions about BrainDrainsrdquo Journal of Economic Perspectives vol 25 no 3 pp 107ndash128 2011

[11] H Betz ldquoPolitics of Resentment Right-Wing Radicalism inWest Germanyrdquo Comparative Politics vol 23 no 1 pp 45ndash601990

[12] R Knight ldquoHaider the freedom party and the extreme right inAustriardquo Parliamentary Affairs vol 45 no 3 pp 285ndash299 1992

[13] P Fysh and J Wolfreys ldquoLe pen the National Front and theextreme right in Francerdquo Parliamentary Affairs vol 45 no 3pp 309ndash326 1992

[14] G Voerman and P Lucardie ldquoThe extreme right in the Nether-lands The centrists and their radical rivalsrdquo European Journalof Political Research vol 22 no 1 pp 35ndash54 1992

[15] WD Chapin ldquoExplaining the electoral success of the new rightThe German caserdquoWest European Politics vol 20 no 2 pp 53ndash72 1997

[16] J M Smith ldquoDoes crime pay Issue ownership political oppor-tunity and the populist right in Western Europerdquo ComparativePolitical Studies vol 43 no 11 pp 1471ndash1498 2010

[17] B Bonikowski and PDiMaggio ldquoVarieties of American PopularNationalismrdquo American Sociological Review vol 81 no 5 pp949ndash980 2016

[18] D Rodrik ldquoHow far will international economic integrationgordquo Journal of Economic Perspectives vol 14 no 1 pp 177ndash1862000

[19] B Davis ldquoDespite his heritage prominent economist backs im-migration cutrdquo The Wall Street Journal 1996 httpswwwhksharvardedufsgborjaspublicationspopularWSJ042696pdf

[20] B Podobnik M Jusup Z Wang and H E Stanley ldquoHow fearof future outcomes affects social dynamicsrdquo Journal of StatisticalPhysics vol 167 no 3-4 pp 1007ndash1019 2017

[21] R Reuveny ldquoClimate change-induced migration and violentconflictrdquo Political Geography vol 26 no 6 pp 656ndash673 2007

[22] T C Schelling ldquoDynamic models of segregationrdquo Journal ofMathematical Sociology vol 1 pp 143ndash186 1971

[23] R Axelrod ldquoThe dissemination of culture a model withlocal convergence and global polarizationrdquo Journal of ConflictResolution vol 41 no 2 pp 203ndash226 1997

[24] M McPherson L Smith-Lovin and J M Cook ldquoBirds of aFeather Homophily in Social Networksrdquo Annual Review ofSociology vol 27 pp 415ndash444 2001

[25] T Antal P L Krapivsky and S Redner ldquoDynamics of socialbalance on networksrdquo Physical Review E vol 72 Article ID036121 2005

[26] S Marvel J Jon Kleinbergb D Robert R D Kleinberg andS Strogatz ldquoContinuous-Time Model of Structural BalancerdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 108 no 5 pp 1771ndash1776 2011

[27] C Gracia-Lazaro L M Flora and Y Moreno ldquoSelectiveAdvantage of Tolerant Cultural Traits in the Axelrod-SchellingModelrdquo Physical Review E vol 83 Article ID 056103 2011

[28] F Aguiar and A Parravano ldquoTolerating the IntolerantHomophily Intolerance and Segregation in Social BalancedNetworksrdquo Journal of Conflict Resolution vol 59 no 1 pp 29ndash50 2015

[29] M Lim R Metzler and Y Bar-Yam ldquoGlobal pattern formationand ethniccultural violencerdquo Science vol 317 no 5844 pp1540ndash1544 2007

[30] R M Dancygier Immigration and Conflict in Europe Studies inComparative Politics Cambridge University Press 2010

[31] R M Dancygier and D D Laitin ldquoImmigration into EuropeEconomic Discrimination Violence and Public Policyrdquo AnnualReview of Political Science vol 17 pp 43ndash64 2014

[32] S Galam and M A Javarone ldquoModeling Radicalization Phe-nomena in Heterogeneous Populationsrdquo PLoS ONE vol 11Article ID e0155407 2016

Complexity 5

FRA (116)

GRC (89)

HUN (06)

BGR (12)ESP

PRT

SWE (159)

DK (99)UK (124)

ITA (94)

FIN (54)

05 1

2

minus2

4

0

0

Annual immigration inflow ( population)

Ann

ual

chan

ge in

RW

vot

ers NOR (138) DEU (119)

BEL (104) AUT (157)

NL (117)

Figure 3 Immigrant inflows and the popularity of right-wing populistmovementsmdasha nonlinear threshold Shown is the annualized immi-grant inflow into a given country (horizontal axis) as a percentage ofthat countryrsquos population as well as the corresponding percentagechange in RW populist votes (vertical axis) In parentheses are thefractions of immigrants in the total population of the correspondingcountry For a group of countries in which the annualized increasein the percentage of RW voters exceeded 2 this increase isvirtually independent of the inflow of immigrants Such a result mayreflect the EUrsquos political organization that is the lack of internalborders whereby if one country decides to accept immigrants thedecision may have repercussions for all the other member statesWe also observe a threshold indicated by a dashed line at whichthe immigrant inflow into a given country is sufficiently high toinvariably provoke an increase in the percentage of RW populistvoters In the model construction this threshold suggests 120572 = 0004on an annual basis

We have an extra index 119894 = 1 2 119873 that refers to a cross-sectional unit giving

RW119894119905 = 1205730 + 120573119871IMIM119871119894119905 + 120573

EUIM IMEU119894119905 + 120573

ter119863 119863119894119905 + 120573

ter119868 119868119894119905

+ 120573119880119880119894119905 + 120573119872119872119894119905 + 119890119894119905(1)

where 119890119905 is the random errorTable 1 shows the results of the TSCS regression model

indicating that the fraction of immigrants in the generalpopulation and the immigration inflow into the entire EUare the significant explanatory variables In addition thecoefficient 120573119871IM is not only significant but also higher thanunity Perhaps surprisingly the response variable is notsignificantly affected by the number of injuries and casualtiesin violent incidents involving immigrants We also show thatunemployment insignificantly affects the popular vote

The aforementioned survey data suggest that not everyEU nation is equally tolerant to immigrants We believe thata proxy for this tolerance can be a fraction of RW votes justafter the Second World War (RW0) when the fractions ofimmigrants were considerably smaller than nowadays Notethat by far the largest fraction of RW votes was recordedin Austria 1167 Table 2 show the results of the TSCSregression model when RW0 is included We find that this

Table 1 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression as defined in (1) Test statistics Wald1205942(6) = 14787 and Prob gt 1205942 = 0000

Coeff Std err 119911 119875 gt |119911|120573119871IM 217 0527 412 0000120573EUIM 4311 532 810 0000120573119863ter minus37119890 minus 04 49119890 minus 04 minus078 0437120573119868ter 21119890 minus 04 16119890 minus 04 128 0202120573119872 0208 0212 098 0325120573119880 minus0167 0348 minus048 06301205730 minus0298 0103 minus289 0004

Table 2 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression as defined in (1) Test statistics Wald1205942(6) = 15154 and Prob gt 1205942 = 0000

Coeff Std err 119911 119875 gt |119911|120573119871IM 132 0445 296 0003120573EUIM 4559 535 852 0000120573ter119863 minus42119890 minus 04 49119890 minus 04 minus086 0392120573ter119868 22119890 minus 04 16119890 minus 04 138 0167120573119872 minus0019 02 minus009 0926120573119880 minus0015 0289 minus005 0959RW0 0904 0562 161 01081205730 minus0139 0093 minus150 0133

Table 3 Pooled time series cross-section (TSCS) analysis withrandom-effects GLS regression Test statistics Wald 1205942(5) = 1424and Prob gt 1205942 = 0000

Coeff Std err 119905 stat 119875 gt 119905120573119871IM 217 0466 465 0000120573EUIM 4672 501 933 00001205730 minus0235 0056 minus415 0000

new regression insignificantly contributes to modern RWpopulism

Table 3 shows the results of the TSCS regression modelwithoutMIPEX unemployment and violent incidents Inter-estingly with the total inflow of immigrants into the EUof 100000 on a monthly basis coefficient values in Table 3suggest that around 30 of immigrants in the total popu-lation of a country are sufficient to cause larger than 50support for RW populist parties which corresponds to theresult obtained by a linear fit in Figure 1

The analyzed data do not indicate whether the rise of RWpopulism is a transient phenomenon or a longer-term changein political orientation One factor is the persistence of votermemory In the October 2015 local election in Vienna theRW populist party won 33 of the popular vote During thepresidential election a fewmonths later themigrant crisis hadreached a peak and the RW populist movement candidatesecured almost 50 of the votes narrowly losing to a leftistrival These results have been contested and a new electionin December 2016 brought around 48 to the RW populistcandidate indicating that this rise in RW populism is not

6 Complexity

short-term This phenomenon has been described in theliterature Betz [41] describes how a substantial increase inthe number of refugees and illegal immigrants in Europeancountries during the 1980s provoked a wave of radical RWpopulism Following these events in the early 1990s thereremained between 11 and 14 of Europeans who felt thepresence of other nationalities races and religions to beunsettling [41]

3 Model

Human interactions are often heterogeneous and prone toabrupt nonlinear responses Because this characterizes therise of the RW populist party and its RW candidate in theAustrian elections such linear approaches as the regressionin (1) yield only partial results A useful intuition is gained bythinking about the election system in a democratic country asa randomwalker Even in a bipolar political system not everyleft government is equally left and similarly not every rightgovernment is equally right Therefore as a result of electionthe randomwalker can move either left or right with the sizeof this move depending on the standard deviation Withoutlimitations such that democracy and the election processpossess an unlimited tolerance for every political option aftera long enough time the random walker is bound to endup either extremely right or extremely left Both of theselimits exhibit ideological rigidity likely to substantially reducethe level of democracy and tolerancemdashin agreement withPopper ldquounlimited tolerance must lead to the disappearanceof tolerancerdquo [42] Thus RW populism can be considered asjust one of the two randomwalk limits However the randomwalker intuition does not explain why a society would moveleft or right nor does it provide a microscopic interpretationof the societal processes at the individual agent level

To mechanistically characterize the rise of RW populismand account for the existence of tipping points in socialdynamics we use a complex network approach [20 43ndash45]Complex network science is able to emulate heterogeneity inhuman interactions and goes beyond capturing the dynam-ics near tipping points Heterogeneity is important whenconsidering immigration and integration issues becauseimmigrants in order to sustain their identity live in ldquohubsrdquothat make them more difficult to integrate than immigrantsmixed into the native population

We construct our model by setting a constant numberof native ldquoinsiderrdquo agents and arranging them in an Erdos-Renyi random network of business and personal contactsImmigrant ldquooutsiderrdquo agents are then added to the networkEach insider notices the percentage of outsiders in theirneighborhood and based on this percentage decides whetherto be supportive of globalization or RW populism Insideragents get information from their neighborhood and theinteraction is local and this data is essential in understandingtipping points [46] but there are also other relevant nonlocalinteractionsThere is furthermore the factor that informationcan be misperceived or misinterpreted In the following weformalize these concepts with three assumptions

Assumption i (media and economic influences) At eachone-month period of time 119905 insider agents are influencedby media at a probability rate 119901 and remain influencedfor a period 120591 We assume that this influence transformsinsiders into RW populist supporters Although media canaffect insiders in both directions we focus on the growthrate and disregard negative values of 119901 The effect of themedia is global and hearing that immigration is occurringcan transform some insiders into RW populist supportersirrespective of their local situation For example most likelyas the effect of media coverage of immigration to boththe EU and the UK during the BREXIT referendum mostUK districts with low immigration voted mainly for Leave[47] The media alongside a lower level of tolerance toimmigration can be an important reason why in ex-socialistcountries the large fraction of voters oppose receiving evena low overall fraction of immigrants The probability rate119901 can also reflect such economic factors as unemploymentThus we use equation [43] 119901lowast = 1 minus exp(minus119901120591) to calculatethe probability that a randomly chosen insider agent is beinginfluenced by the media

Assumption ii (local influence of outsiders) In local electionsin Greece in November 2010 although the far-right GoldenDawn party received only 53 of the vote in some neighbor-hoods of Athens with large immigrant communities the partywon nearly 20 [48] This suggests that contacts betweeninsiders and outsiders do matter Within our network modelwe maintain a constant number of 119873 insiders and then add119868(0) outsidersWe then increase the number of outsiders withan inflow 119869119905 at each moment 119905 (representing one month)We randomly place newly arriving outsiders between insideragents both of whom initially have an average number ofconnections (ie a degree) 119896 The total number of outsiders119868(119905) is obtained by summing the monthly 119869 values accordingto 119868(119905) = 119868(0) + sum119905119904=1 119869119904 At any given moment the fractionof outsiders will equal 119891119868(119905) = 119868(119905)(119873 + 119868(119905)) To accountfor the effect of contacts between insiders and outsiders weassume that any insider agent 119894with 119896119894 total connections turnsto RWpopulism at a rate 1199011015840 when this agent is surrounded byat least 119898119894 = 1198911015840119868119896119894 outsiders [43 46 49] where 0 lt 1198911015840119868 lt 1is a constant model parameter quantifying the toleration ofinsiders This assumption merits a few additional comments

First the probability that randomly chosen insider agent 119894with 119896119894 connections is surrounded by119898119894 outsiders and there-fore prone to RW populism is 1199011(119896119894 119898119894 119891119868) equiv sum119896119894119895=119898119894 119891

119895119868 (1 minus

119891119868)119896119894minus119895 ( 119896119894119896119894minus119895 ) In this formula 119891119868 is the true current state

of the network However the information may be biasedand cause insiders to think there are more outsiders than isactually the case If the bias is Δ119891119868 then 1199011(119896119894 119898119894 119891119868 + Δ119891119868) gt119901(119896119894 119898119894 119891119868) This increased probability 119901(119896119894 119898119894 119891119868 + Δ119891119868)implies that the tolerance parameter 1198911015840119868 must decrease byamount Δ1198911015840119868 which we estimate using condition 119901(119896119894 119898119894 minusΔ1198911015840119868119896119894 119891119868) = 119901(119896119894 119898119894 119891119868 + Δ119891119868) An implicit assumption hereis that all insider agents are equally tolerant to immigrantsbecause the tolerance parameter 1198911015840119868 is defined as a globalnetwork property rather than an individual agent property

Complexity 7

An alternative would be to assume a distribution of tolerancelevels in which case 1198911015840119868 would represent the mean [20]

We now expand assumption (ii) with extension (A) whenthe immigration inflow 119869 is below some threshold 1198691015840 thesociety becomes more tolerant When 119869 lt 1198691015840 at a giventime moment the tolerance parameter 1198911015840119868 increases by anamount Δ1198911015840119868 = 120575 gt 0 there is a balance between immigra-tion and immigrant integrationmdashoutsiders are successfullyintegratedmdashand insiders are able to acclimate to the changesin their society Figure 2(b) shows that the 1198691015840 value forGermany is approximately 10000 people per month Beingable to accurately determine the maximum 1198691015840 value is highlyrelevant to the success of globalization According to ourmodel immigration and integration can be the inevitableconsequences of globalization when 119869 lt 1198691015840 If this is notthe case globalization will be threatened by the rise of RWpopulism the democratic system will enter an intolerantmode and cooperation between nations will be downgradedon the list of political priorities

Empirical evidence suggests that we add an oppositeextension (B) as the inflow of outsiders 119869 crosses somethreshold 11986910158401015840 (which does not need to be equal to 1198691015840)society becomes less tolerant Mathematically extension (B)indicates that when 119869 gt 11986910158401015840 the tolerance parameter 1198911015840119868decreases by

Δ1198911015840119868 = minus120574119869 (2)

Using the econometric models in (1) we decrease thetolerance parameter proportional to inflow 119869 where 120574 isa proportionality coefficient expressing the sensitivity ofinsiders to high levels of outsider inflow Figure 3(b) showsthreshold 11986910158401015840 in terms of total population that is 11986910158401015840 = 120572119873where 120572 is another proportionality coefficient Figure 3(b)estimates the 120572 value when all of the EU countries are hitby the migrant crisis The dashed line is annual inflow belowwhich countries have a mixed response to immigration andabove which support for RW populism increases Because120572 = 11986910158401015840119873 and using Figure 3(b) 1211986910158401015840119873 asymp 0004 we obtain

120572 asymp 000033 (3)

Extensions (A) and (B) are opposite limiting cases onein which immigration is slow and the other in whichimmigration is rapid Apart from the empirical evidencethat these extensions are needed brain science for exampleoffers a physiological interpretation political attitudes have acounterpart in brain structure [50ndash52] If outsiders increaseat a rate and in a manner perceived as controllable byinsiders the prefrontal cortex of the human brain responsiblefor decision making and for moderating social behavioracclimates to the new circumstances but if the insidersperceive the outsiders to be invaders the prefrontal cortexis supplanted by the amygdala which induces a fightingreaction and tolerance is suppressed

Although with assumptions (i) and (ii) our modelaccounts for the processes that affect individual insider opin-ion the expansion of RW populism can become extremelyrapid when insiders are influenced by their peers This well-documented phenomenon in human interactions is further

accelerated when social media is added Thus the spread ofRW populism can be highly nonlinear much like the spreadof a highly contagious disease We include this nonlinear col-lective spreading mechanism in our third model assumption

Assumption iii (mutual insider contagion) At any givenmoment 119905 an insider agent 119894 with 119870119894 connections to otherinsiders turns to RW populism at rate 11990110158401015840 if at time 119905 minus 1this agent has at least119872119894 = 1198701198942 RW populist supporters intheir neighborhood Note that for simplicity factor 12 playsa role analogous to the tolerance parameter in assumption(ii) Because of connections between insider agents whenRW populism emerges anywhere in the network the populistmovement is able to spread like a contagion This collectivespreading indicates that insider agents can become RWpopulist supporters even when there are no outsiders inthe immediate neighborhood Thus some EU countries withalmost no immigration are opposed to accepting even a smallgroup of immigrants This may have affected the outcomeof the recent US presidential election in which the winningcandidate was often ridiculed in the mainstream mediamdashanattitude represented in our model by assumption (i)

The model here assumes imitative interactions withouttesting the validity of such an assumption However thisassumption could be tested by themethod of Agliari et al [53]if the required data were available The authors use the toolsfrom statistical mechanics to determine from data the natureof interactions in social customs such as local and mixedmarriages in Italy and neighboring European countries Thefraction of actualmarriages satisfying a given condition scalesdifferently with the fraction of all possible couples that satisfythe same condition depending on the nature of interactionsbetween agents If the interactions are independent (one-body model) the scaling is linear whereas if the interactionsare imitative (two-body model) the scaling is square-rooted

We now turn to the simulation results and their impli-cation Figure 4 shows how in a network of 5000 agents thefraction of RW populist supporters increases when there isa constant inflow of outsiders here 119869 = 2 per month Thisinflow is annually approximately 05 of the total populationwhich is slightly more than the threshold value impliedin Figure 3(b) Simulations with 119869 = 2 per month aredesigned to emulate the rapid limit in extension (B) describedabove After approximately 37 years of rapid globalization thenetwork reaches a tipping point and abruptly shifts to amodedominated by RW populism RW populism dominates whenmore than 50 of the network is made up of RW populistsupporters (ie 119875 gt 05 where 119875 denotes the fraction of RWpopulists) The threshold is 50 because in a democracy themajority rules

Figure 4 also shows how the simulated network underassumptions (i)ndash(iii) responds to shocks (red curve) At thisstage extension (B) does not yet operateThe constant annualinflow of outsiders 119869 = 2 is supplemented by two eventsat times 1199051 and 1199052 when 119869 = 200 The state of the networkcharacterized by the proportion of RWpopulists (119875) exhibitsa much stronger response at 1199052 than at 1199051 although the shockinflow (119869 = 200) is the same This occurs because at 1199052 the

8 Complexity

08

00

02

J = 2

t1

J = 2 except at

t2

Change in tolerance

500

04

06

Time

Prob

abili

ty o

f RW

pop

ulism

t1 amp t2

Figure 4 Nonlinearity a tipping point and the rise of right-wingpopulism Using a network of 119873 = 5000 agents each with anaverage of 15 connections we examine the effect of a constantinflow of outsiders at rate 119869 = 2 at each time step In this setupthe total number of outsiders at any moment in time is 119868(119905) =sum119894 119869119894 = ⟨119869⟩119905 As the fraction of outsiders 119891119868 = 119868(119873 + 119868)approaches the tolerance parameter 1198911015840119868 = 015 the presence of atipping point causes the fraction of RW populist supporters to startincreasing nonlinearly and eventually undergo a sudden jump (iea discontinuous change) at about 37 years (450 months) into thesimulation (black curve) The sudden jump happens much earlierif the inflow of outsiders experiences shocks at times 1199051 and 1199052at which 119869 = 200 outsiders enter the network In particular asthe network approaches the tipping point the effect of exactlythe same shock becomes disproportionately higher (red curve) Inthis case however the tolerance parameter is still kept constantFinally we also examine the case in which shocks at times 1199051 and1199052 affect the tolerance parameter where responsiveness is controlledby parameter 120574 = 00001 Here the second shock at 1199052 is sufficientto instantly tip the network into RW populism (green curve) Otherparameters are 119901 = 0007 120591 = 15 1199011015840 = 05 11990110158401015840 = 05 and 120572 = 0001

system is closer to the tipping point and consequently moreunstable than at 1199051 Because in real-world data the value of119869 can be biased due to estimation errors or misinterpretedinformation our results suggest that approaching the tippingpoint can be concurrent with strong nonlinear effects suchthat even a small shock can trigger a transition to a modedominated by RWpopulismThis can be evenmore explosive(in terms of 119875) if extension (B) is allowed to operate that isif the tolerance parameter 1198911015840119868 changes with 119869

Figure 4 shows a third simulation (green curve) inwhich the dynamics operate under assumptions (i)ndash(iii) withextension (B) The tolerance parameter 119891119868 thus changes with119869 as prescribed by (2) Because of the decreasing tolerancethe second shock at 1199052 can now push the system beyondthe tipping point and thus the dominance of RW populismoccurs earlier than in the two previous simulations

From the simulations in Figure 4 alone it is unclearhow much the local influence of outsiders (assumption (ii))contributes to the rise of RW populism relative to the insider

08

1

06

04

02

00

TotalLocal influence (ii)Contagion (iii)

250 500

Time

Prob

abili

ty o

f RW

pop

ulism

Figure 5 Breakdown of the causes of right-wing populism Figure 4shows that the probability of RW populism 119875 suddenly increasesas society approaches a tipping point but remains silent on theunderlying causes Here we discern between the contributionsof local outsider influence (assumption (ii)) and mutual insidercontagion (assumption (iii)) Far from the tipping point 119875 mainlyresponds to local outsider influence (ii) By contrast as the networkapproaches its tipping point mutual insider contagion (iii) takesover and accelerates the transition to RW populist dominanceParameter values are119873 = 5000 with an average degree of 15 119869 = 2119901 = 0007 120591 = 15 1199011015840 = 07 and 11990110158401015840 = 08

contagion (assumption (iii)) Figure 5 shows these contribu-tions After the initial transients fade the local influence ofoutsiders drives the increase in RW populists in the networkThe contribution of mutual insider contagion is relativelysmall until the system approaches a tipping point Near thetipping point contagion spreads rapidly and overtakes thelocal outsider influence as the main contributor to the riseof RW populism Thereafter the RW populist movement cansustain itself without support from the outside

In the regime of moderate immigration inflows (1198691015840 lt 119869 lt11986910158401015840) we can examine the dynamics of our complex networkusing themean-field theory (MFT) analytic techniqueWhenthe number of agent connections does not deviate greatlyfrom the network average the probability 119875 that a randomlychosen insider agent 119894 is an RWpopulist supporter due to anyof the processes underlying assumptions (i)ndash(iii) is

119875 = 119901lowast + 11990110158401199011 (119896119894 119898119894 119891119868) + 119901101584010158401199011 (119870119894119872119894 119875)

minus 119901lowast11990110158401199011 (119896119894 119898119894 119891119868) minus 119901lowast119901101584010158401199011 (119870119894119872119894 119875)

minus 11990110158401199011 (119896119894 119898119894 119891119868) 119901101584010158401199011 (119870119894119872119894 119875)

(4)

where the last three terms avoid double counting in accor-dance with the probability theory formula 119875(119860 cup 119861 cup 119862) =119875(119860) + 119875(119861) + 119875(119862) minus 119875(119860)119875(119861) minus 119875(119860)119875(119862) minus 119875(119861)119875(119862) forthree mutually independent events 119860 119861 and 119862 that cannotoccur simultaneously In the MFT approximation we candrop index 119894 because no single agent is markedly differentfrom the collective average Previously we set 119872 = 1198702

Complexity 9

417 882

88

200 600 800 1000 1200

Time (months)

13208

1

06

04

02

00

Prob

abili

ty o

f RW

pop

ulism

400

294

263185

555

J = 1 fI = 005

J = 1 fI = 01

J = 1 fIfI

= 015

J = 2 = 005

J = 3 fI = 005

J = 3 fI = 01

J = 3 fI = 015

J = 3 fI = 02

Figure 6 Predicting the timing of RW populism We find that for abroad range of outsider inflows (119869) and tolerance parameter values(1198911015840119868 ) (5) predicts the moment at which 119875 gt 05 in a manner thatagrees favorably with the simulation results Except for 120574 = 0 otherparameters are the same as in Figure 4

for simplicity but the peer pressure measured by the valueof the proportionality factor between 119872 and 119870 can differbetween countries or regions In addition parameters 1199011015840 and11990110158401015840 are constants only in theory The real social dynamicsare such that these parameters may change in responseto rumors political manipulations or outside shocks Ifparameters 1199011015840 and 11990110158401015840 substantially increase the value of 119875also increases thus further improving the prospects for RWpopulism dominance In our framework due to democracyrsquosmajority rule principle when 119875 approaches 05 the nonlinearprocesses embedded in assumptions (ii) and (iii) cause asudden transition to RW populist mode

Because a theoretical model is more useful if it haspredictive power [54 55] we show how a network of agentsunder assumptions (i)ndash(iii) leads to a simple formula for thetiming at which RW populism starts to dominate

119905119905ℎ =1198731198911015840119868

119869 (1 minus 1198911015840119868 )minus 119868 (0)

119869 (5)

Gaining this result involves three steps First if the immi-gration inflow is constant then the number of outsidersin the network after 119905 time steps is 119868(0) + 119869119905 Second thetotal population size thus equals 119873 + 119868(0) + 119869119905 Finally (5)follows if the current fraction of outsiders (119868(0) + 119869119905)(119873 +119868(0) + 119869119905) is equated with the critical parameter 1198911015840119868 Figure 6shows that for a number of immigration inflow-toleranceparameter pairs (119869 1198911015840119868 ) the simulated timing of the shift toRW populist mode (ie 119875 gt 05) fits theoretical predictionsIn conjunction with the empirical data on tolerance towardsimmigrants in the EU countries the formula in (5) could beused to provide an estimate of when a given country mightbe approaching a possible tipping point to RW populismdominance

Numerical simulations allow us to examine not only iso-lated single networks but also the interdependence betweentwo or more networks Note that there is a potential for a cas-cade effect when an RW populist movement in one networkcauses the rise of RW populist movements in other networksThis cascade effect is growing in relevance because expandingglobalization is causing countries to become increasinglysimilar to each another and this similarity increases in suchsupranational organizations as the EU in which bordersbetween nation states are rapidly fading

To examine how interdependence affects the rise of RWpopulism we set up two random economically equivalentER networks (equal 119901 in the model) with different tolerancelevels towards outsiders (different 1198911015840119868 in the model) Tothe usual intraconnections within each network we addinterconnecting agents that linkwith their counterparts in theother network Apart from this addition we retain the samemodel assumptions as previously held

To examine the effects of interconnectedness we first runnumerical simulations of two independent networks withoutinterconnections [see Figure 7(a)] As expected when theinflow level of outsiders is the same (119869 = 2) the networkwith a higher tolerance parameter (11989110158401198681 = 04) reaches thetipping point much later than the network with a lowertolerance parameter (11989110158401198682 = 02) When the networks areinterconnected and the more tolerant network experiencesan increased inflow (1198692 = 4) and a shock (119869 = 500) at time1199051 = 500 its susceptibility to RW populism increases andit also affects the other network and shortens the time oftransition to RW populism [see Figure 7(b)] It would appearthat countries do not want to be the first to cross the line butin an interconnected world being second is easier

4 Discussion and Conclusion

Why some countries (eg the ex-socialist EU countries)strongly oppose receiving immigrants while others (eg theUSA) have a long history of receiving immigrants is animportant topic in the social sciences and more recently amajor issue for the EU Perhaps receiving immigrants hasbeen an ongoing pattern in the USA because there is nosingle dominant ethnicity and thus a clear distinction ismadebetween USA national identity and the country of originof its citizens In addition because USA citizens are peoplefrom all over the world there is no single dominant religiousethnic or cultural group that can organize and threaten theestablished social order In France we find the oppositeThereis a large group of immigrants with different language andreligion from the French majority whose presence can instillfear among themajority Fear exacerbated by an inflow rate ofimmigrants that exceeds the rate of their integration can leadto a volatile situation one that is often resolved in one of twoways Either there is an immigrant uprising as exemplifiedby the Visigoth immigrants and their ex-Roman commanderAlaricwhoplunderedRome in 410 or themajority populationsuppresses the inflow of immigrants Currently this secondoption is often accomplished by supporting populist right-wing parties

10 Complexity

08

1

00

06

0402

J1 = 2

J2 = 2

200015001000500

Prob

abili

ty o

f RW

pop

ulism

Time

(a)

08

1

00

06

2000

0402

J1 = 2

J2 = 4

Time15001000500Pr

obab

ility

of R

W p

opul

ism

(b)

Figure 7 Interconnected networks or why ldquosomebody elsersquos problemrdquo easily turns into ldquomy problemrdquo In (a) we show the case when there areno interlinks between networks The tolerance parameters between the two networks differ 1198911198681 = 02 and 1198911198681 = 04 while the inflows intoboth networks are the same 1198691 = 1198692 = 2 (b) More tolerant network is now exposed to a higher inflow 1198692 = 4 and a shock at 1199051 = 500 Theaverage number of connections for intraconnections (interconnections) in both networks equals 15 (10) The other parameters are the sameas in Figure 4

Although globalization was conceived to allow capitaland labor to move freely across national borders real-worldglobalization is affected by a multitude of noneconomicfactors such as ethnicity culture and religion With so manyfactors at play globalization is an enormously complex pro-cess and often noneconomic factors do not align with purelyeconomic factors This misalignment can lead to frustrationsthat feed populist movements By opposing the collaborationthat comes with globalization populist movements act asa feedback mechanism that pushes the general populationtowards becoming more ideologically rigid and this inturn further accelerates the populism [56] A strengthenedpopulist movement can also trigger tectonic shifts in worldaffairs as exemplified by BREXIT in the UK and the recent2016 US presidential election

Problems in a globalized world rarely confine themselvesto one place Interdependencemakes the developed countriesmore alike and synchronizes their social dynamics Thissynchronization can cause political shifts in one country tospill over into other countries and this is what enables the riseof RW populism to spread across large regions of the worldAfter BREXIT and the 2016 US presidential election politicalelites should not expect to continue business as usual Beingthe first to adopt amajor political shift with a high probabilityof negative economic consequences is difficult but once thatline has been crossed the interdependence of globalizationmakes cascades (domino effects) an active possibility No onewants to be first but many are ready to be second

The tipping point that brings on the rise of RW populismmay be reached more quickly when voters face a binarychoice for example the ldquoyesrdquo or ldquonordquo choice in the UKBREXIT referendum and the two major-party candidates inthe recent US presidential election In both cases the populistoption secured a narrow victory As mentioned above thepolls indicate that the populist party in Austria can expect toreceive 34 of the votes but the populist party candidate inthe presidential race can expect close to 51 A simple wayto understand these percentages is to assume that politicalattitudes of voters are approximately symmetrical in their

distribution across the political spectrum from left to rightConsequently if leftist voters comprise 120595119871 of the totalpopulation then RW populist voters will maintain a similarpresence that is 120595119877 asymp 120595119871 Facing this binary choice thecentrist voters have no one representing their views and thusare likely to vote evenly between the two available optionsAn implication is that when 120595119871 asymp 120595119877 asymp 33 even aslight (statistically significant) imbalance in favor of120595119877over120595119871 can tip the society towards RW populism In Austria120595119877 asymp 36 seems to be sufficient to make the RW populistcandidate a front runner in the presidential race

The final outcome of the battle between the conflictingfactors surrounding globalization will almost surely havetremendous economic implications If a country approachesits tipping point how will the ensuing volatility affect itslong-term credit rating If a domino effect cascades acrossa large region of the EU or the entire EU how will thataffect the Euro and the common banking system Withouta proper resolution of the migrant crisis what will be theimpact on systemic risk In an attempt to shed light onsome of the factors and underlying processes that affectthe success of globalization we offer an empirically backedtheoretical model of the rise of RW populism in response tounsustainable immigration inflows

Our model emphasizes the need for controlled globaliza-tion in which immigration inflows into a society are balancedwith the ability of the society to integrate the immigrantsThis ability is arguably improved when immigrants mixwith the native population which is a principle practicedin Singapore where immigrant hubs are discouraged andtenants in government-built housing (comprising 88 ofall housing) must be of mixed ethnic origin [20] Becausetolerance towards immigrants is conditional when immigra-tion inflows overshadow integration rates the society can betipped into RWpopulism an intolerant mode of functioningThe rise of RW populism can occur because elections areby their nature stochastic and they resemble a mathematicalrandom walker Left to its own devices a random walkerwill eventually hit an absorbing barrier Here this barrier

Complexity 11

of course is a metaphor for the demise of globalizationironically at the very hands of the progressive system (iedemocracy) thatmade globalization possible in the first place

Disclosure

A version of this work was presented in Warsaw at ldquoEcono-physics Colloquium 2017rdquo

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

The authors are grateful to S Galam and T Lipic forhelpful suggestions Boris Podobnik received the supportfrom Slovenian Research Agency (ARRS) via Project J5-8236and from the University of Rijeka Boris Podobnik and H EStanley received support from the Defense Threat ReductionAgency (DTRA) theOffice ofNaval Research (ONR) and theNational Science Foundation (NSF) (Grant CMMI 1125290)Marko Jusup was supported by the Japan Science andTechnology Agency (JST) Program to Disseminate TenureTracking System

References

[1] H Grubel and A Scott ldquoThe international flow of humancapitalrdquo American Economic Review vol 56 pp 268ndash274 1966

[2] G C Hufbauer and K Suominen Globalization at Risk Chal-lenges to Finance and Trade Yale University Press 2010

[3] F Docquier and H Rapoport ldquoGlobalization Brain Drain andDevelopmentrdquo Journal of Economic Literature vol 50 pp 681ndash730 2012

[4] D Rodrik The Globalization Paradox Democracy and theFuture of the World Economy W W Norton amp Company NewYork USA 2011

[5] B R Chiswick ldquoThe Effect of Americanization on the Earningsof Foreign-born Menrdquo Journal of Political Economy vol 86 no5 pp 897ndash921 1978

[6] G J BorjasHeavenrsquosDoor Immigration Policy and theAmericanEconomy Princeton University Press 1999

[7] G J Borjas ldquoThe economics of immigrationrdquo Journal ofEconomic Literature vol 32 pp 1667ndash1717 1994

[8] G J Borjas ldquoAssimilation Changes in Cohort Quality and theEarnings of Immigrantsrdquo Journal of Labor Economics vol 3 no4 pp 463ndash489 1985

[9] D Swank and H G Betz ldquoGlobalization the welfare stateand right-wing populism in Western Europerdquo Socio-EconomicReview vol 1 no 2 pp 215ndash245 2003

[10] J Gibson and D McKenzie ldquoEight Questions about BrainDrainsrdquo Journal of Economic Perspectives vol 25 no 3 pp 107ndash128 2011

[11] H Betz ldquoPolitics of Resentment Right-Wing Radicalism inWest Germanyrdquo Comparative Politics vol 23 no 1 pp 45ndash601990

[12] R Knight ldquoHaider the freedom party and the extreme right inAustriardquo Parliamentary Affairs vol 45 no 3 pp 285ndash299 1992

[13] P Fysh and J Wolfreys ldquoLe pen the National Front and theextreme right in Francerdquo Parliamentary Affairs vol 45 no 3pp 309ndash326 1992

[14] G Voerman and P Lucardie ldquoThe extreme right in the Nether-lands The centrists and their radical rivalsrdquo European Journalof Political Research vol 22 no 1 pp 35ndash54 1992

[15] WD Chapin ldquoExplaining the electoral success of the new rightThe German caserdquoWest European Politics vol 20 no 2 pp 53ndash72 1997

[16] J M Smith ldquoDoes crime pay Issue ownership political oppor-tunity and the populist right in Western Europerdquo ComparativePolitical Studies vol 43 no 11 pp 1471ndash1498 2010

[17] B Bonikowski and PDiMaggio ldquoVarieties of American PopularNationalismrdquo American Sociological Review vol 81 no 5 pp949ndash980 2016

[18] D Rodrik ldquoHow far will international economic integrationgordquo Journal of Economic Perspectives vol 14 no 1 pp 177ndash1862000

[19] B Davis ldquoDespite his heritage prominent economist backs im-migration cutrdquo The Wall Street Journal 1996 httpswwwhksharvardedufsgborjaspublicationspopularWSJ042696pdf

[20] B Podobnik M Jusup Z Wang and H E Stanley ldquoHow fearof future outcomes affects social dynamicsrdquo Journal of StatisticalPhysics vol 167 no 3-4 pp 1007ndash1019 2017

[21] R Reuveny ldquoClimate change-induced migration and violentconflictrdquo Political Geography vol 26 no 6 pp 656ndash673 2007

[22] T C Schelling ldquoDynamic models of segregationrdquo Journal ofMathematical Sociology vol 1 pp 143ndash186 1971

[23] R Axelrod ldquoThe dissemination of culture a model withlocal convergence and global polarizationrdquo Journal of ConflictResolution vol 41 no 2 pp 203ndash226 1997

[24] M McPherson L Smith-Lovin and J M Cook ldquoBirds of aFeather Homophily in Social Networksrdquo Annual Review ofSociology vol 27 pp 415ndash444 2001

[25] T Antal P L Krapivsky and S Redner ldquoDynamics of socialbalance on networksrdquo Physical Review E vol 72 Article ID036121 2005

[26] S Marvel J Jon Kleinbergb D Robert R D Kleinberg andS Strogatz ldquoContinuous-Time Model of Structural BalancerdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 108 no 5 pp 1771ndash1776 2011

[27] C Gracia-Lazaro L M Flora and Y Moreno ldquoSelectiveAdvantage of Tolerant Cultural Traits in the Axelrod-SchellingModelrdquo Physical Review E vol 83 Article ID 056103 2011

[28] F Aguiar and A Parravano ldquoTolerating the IntolerantHomophily Intolerance and Segregation in Social BalancedNetworksrdquo Journal of Conflict Resolution vol 59 no 1 pp 29ndash50 2015

[29] M Lim R Metzler and Y Bar-Yam ldquoGlobal pattern formationand ethniccultural violencerdquo Science vol 317 no 5844 pp1540ndash1544 2007

[30] R M Dancygier Immigration and Conflict in Europe Studies inComparative Politics Cambridge University Press 2010

[31] R M Dancygier and D D Laitin ldquoImmigration into EuropeEconomic Discrimination Violence and Public Policyrdquo AnnualReview of Political Science vol 17 pp 43ndash64 2014

[32] S Galam and M A Javarone ldquoModeling Radicalization Phe-nomena in Heterogeneous Populationsrdquo PLoS ONE vol 11Article ID e0155407 2016

6 Complexity

short-term This phenomenon has been described in theliterature Betz [41] describes how a substantial increase inthe number of refugees and illegal immigrants in Europeancountries during the 1980s provoked a wave of radical RWpopulism Following these events in the early 1990s thereremained between 11 and 14 of Europeans who felt thepresence of other nationalities races and religions to beunsettling [41]

3 Model

Human interactions are often heterogeneous and prone toabrupt nonlinear responses Because this characterizes therise of the RW populist party and its RW candidate in theAustrian elections such linear approaches as the regressionin (1) yield only partial results A useful intuition is gained bythinking about the election system in a democratic country asa randomwalker Even in a bipolar political system not everyleft government is equally left and similarly not every rightgovernment is equally right Therefore as a result of electionthe randomwalker can move either left or right with the sizeof this move depending on the standard deviation Withoutlimitations such that democracy and the election processpossess an unlimited tolerance for every political option aftera long enough time the random walker is bound to endup either extremely right or extremely left Both of theselimits exhibit ideological rigidity likely to substantially reducethe level of democracy and tolerancemdashin agreement withPopper ldquounlimited tolerance must lead to the disappearanceof tolerancerdquo [42] Thus RW populism can be considered asjust one of the two randomwalk limits However the randomwalker intuition does not explain why a society would moveleft or right nor does it provide a microscopic interpretationof the societal processes at the individual agent level

To mechanistically characterize the rise of RW populismand account for the existence of tipping points in socialdynamics we use a complex network approach [20 43ndash45]Complex network science is able to emulate heterogeneity inhuman interactions and goes beyond capturing the dynam-ics near tipping points Heterogeneity is important whenconsidering immigration and integration issues becauseimmigrants in order to sustain their identity live in ldquohubsrdquothat make them more difficult to integrate than immigrantsmixed into the native population

We construct our model by setting a constant numberof native ldquoinsiderrdquo agents and arranging them in an Erdos-Renyi random network of business and personal contactsImmigrant ldquooutsiderrdquo agents are then added to the networkEach insider notices the percentage of outsiders in theirneighborhood and based on this percentage decides whetherto be supportive of globalization or RW populism Insideragents get information from their neighborhood and theinteraction is local and this data is essential in understandingtipping points [46] but there are also other relevant nonlocalinteractionsThere is furthermore the factor that informationcan be misperceived or misinterpreted In the following weformalize these concepts with three assumptions

Assumption i (media and economic influences) At eachone-month period of time 119905 insider agents are influencedby media at a probability rate 119901 and remain influencedfor a period 120591 We assume that this influence transformsinsiders into RW populist supporters Although media canaffect insiders in both directions we focus on the growthrate and disregard negative values of 119901 The effect of themedia is global and hearing that immigration is occurringcan transform some insiders into RW populist supportersirrespective of their local situation For example most likelyas the effect of media coverage of immigration to boththe EU and the UK during the BREXIT referendum mostUK districts with low immigration voted mainly for Leave[47] The media alongside a lower level of tolerance toimmigration can be an important reason why in ex-socialistcountries the large fraction of voters oppose receiving evena low overall fraction of immigrants The probability rate119901 can also reflect such economic factors as unemploymentThus we use equation [43] 119901lowast = 1 minus exp(minus119901120591) to calculatethe probability that a randomly chosen insider agent is beinginfluenced by the media

Assumption ii (local influence of outsiders) In local electionsin Greece in November 2010 although the far-right GoldenDawn party received only 53 of the vote in some neighbor-hoods of Athens with large immigrant communities the partywon nearly 20 [48] This suggests that contacts betweeninsiders and outsiders do matter Within our network modelwe maintain a constant number of 119873 insiders and then add119868(0) outsidersWe then increase the number of outsiders withan inflow 119869119905 at each moment 119905 (representing one month)We randomly place newly arriving outsiders between insideragents both of whom initially have an average number ofconnections (ie a degree) 119896 The total number of outsiders119868(119905) is obtained by summing the monthly 119869 values accordingto 119868(119905) = 119868(0) + sum119905119904=1 119869119904 At any given moment the fractionof outsiders will equal 119891119868(119905) = 119868(119905)(119873 + 119868(119905)) To accountfor the effect of contacts between insiders and outsiders weassume that any insider agent 119894with 119896119894 total connections turnsto RWpopulism at a rate 1199011015840 when this agent is surrounded byat least 119898119894 = 1198911015840119868119896119894 outsiders [43 46 49] where 0 lt 1198911015840119868 lt 1is a constant model parameter quantifying the toleration ofinsiders This assumption merits a few additional comments

First the probability that randomly chosen insider agent 119894with 119896119894 connections is surrounded by119898119894 outsiders and there-fore prone to RW populism is 1199011(119896119894 119898119894 119891119868) equiv sum119896119894119895=119898119894 119891

119895119868 (1 minus

119891119868)119896119894minus119895 ( 119896119894119896119894minus119895 ) In this formula 119891119868 is the true current state

of the network However the information may be biasedand cause insiders to think there are more outsiders than isactually the case If the bias is Δ119891119868 then 1199011(119896119894 119898119894 119891119868 + Δ119891119868) gt119901(119896119894 119898119894 119891119868) This increased probability 119901(119896119894 119898119894 119891119868 + Δ119891119868)implies that the tolerance parameter 1198911015840119868 must decrease byamount Δ1198911015840119868 which we estimate using condition 119901(119896119894 119898119894 minusΔ1198911015840119868119896119894 119891119868) = 119901(119896119894 119898119894 119891119868 + Δ119891119868) An implicit assumption hereis that all insider agents are equally tolerant to immigrantsbecause the tolerance parameter 1198911015840119868 is defined as a globalnetwork property rather than an individual agent property

Complexity 7

An alternative would be to assume a distribution of tolerancelevels in which case 1198911015840119868 would represent the mean [20]

We now expand assumption (ii) with extension (A) whenthe immigration inflow 119869 is below some threshold 1198691015840 thesociety becomes more tolerant When 119869 lt 1198691015840 at a giventime moment the tolerance parameter 1198911015840119868 increases by anamount Δ1198911015840119868 = 120575 gt 0 there is a balance between immigra-tion and immigrant integrationmdashoutsiders are successfullyintegratedmdashand insiders are able to acclimate to the changesin their society Figure 2(b) shows that the 1198691015840 value forGermany is approximately 10000 people per month Beingable to accurately determine the maximum 1198691015840 value is highlyrelevant to the success of globalization According to ourmodel immigration and integration can be the inevitableconsequences of globalization when 119869 lt 1198691015840 If this is notthe case globalization will be threatened by the rise of RWpopulism the democratic system will enter an intolerantmode and cooperation between nations will be downgradedon the list of political priorities

Empirical evidence suggests that we add an oppositeextension (B) as the inflow of outsiders 119869 crosses somethreshold 11986910158401015840 (which does not need to be equal to 1198691015840)society becomes less tolerant Mathematically extension (B)indicates that when 119869 gt 11986910158401015840 the tolerance parameter 1198911015840119868decreases by

Δ1198911015840119868 = minus120574119869 (2)

Using the econometric models in (1) we decrease thetolerance parameter proportional to inflow 119869 where 120574 isa proportionality coefficient expressing the sensitivity ofinsiders to high levels of outsider inflow Figure 3(b) showsthreshold 11986910158401015840 in terms of total population that is 11986910158401015840 = 120572119873where 120572 is another proportionality coefficient Figure 3(b)estimates the 120572 value when all of the EU countries are hitby the migrant crisis The dashed line is annual inflow belowwhich countries have a mixed response to immigration andabove which support for RW populism increases Because120572 = 11986910158401015840119873 and using Figure 3(b) 1211986910158401015840119873 asymp 0004 we obtain

120572 asymp 000033 (3)

Extensions (A) and (B) are opposite limiting cases onein which immigration is slow and the other in whichimmigration is rapid Apart from the empirical evidencethat these extensions are needed brain science for exampleoffers a physiological interpretation political attitudes have acounterpart in brain structure [50ndash52] If outsiders increaseat a rate and in a manner perceived as controllable byinsiders the prefrontal cortex of the human brain responsiblefor decision making and for moderating social behavioracclimates to the new circumstances but if the insidersperceive the outsiders to be invaders the prefrontal cortexis supplanted by the amygdala which induces a fightingreaction and tolerance is suppressed

Although with assumptions (i) and (ii) our modelaccounts for the processes that affect individual insider opin-ion the expansion of RW populism can become extremelyrapid when insiders are influenced by their peers This well-documented phenomenon in human interactions is further

accelerated when social media is added Thus the spread ofRW populism can be highly nonlinear much like the spreadof a highly contagious disease We include this nonlinear col-lective spreading mechanism in our third model assumption

Assumption iii (mutual insider contagion) At any givenmoment 119905 an insider agent 119894 with 119870119894 connections to otherinsiders turns to RW populism at rate 11990110158401015840 if at time 119905 minus 1this agent has at least119872119894 = 1198701198942 RW populist supporters intheir neighborhood Note that for simplicity factor 12 playsa role analogous to the tolerance parameter in assumption(ii) Because of connections between insider agents whenRW populism emerges anywhere in the network the populistmovement is able to spread like a contagion This collectivespreading indicates that insider agents can become RWpopulist supporters even when there are no outsiders inthe immediate neighborhood Thus some EU countries withalmost no immigration are opposed to accepting even a smallgroup of immigrants This may have affected the outcomeof the recent US presidential election in which the winningcandidate was often ridiculed in the mainstream mediamdashanattitude represented in our model by assumption (i)

The model here assumes imitative interactions withouttesting the validity of such an assumption However thisassumption could be tested by themethod of Agliari et al [53]if the required data were available The authors use the toolsfrom statistical mechanics to determine from data the natureof interactions in social customs such as local and mixedmarriages in Italy and neighboring European countries Thefraction of actualmarriages satisfying a given condition scalesdifferently with the fraction of all possible couples that satisfythe same condition depending on the nature of interactionsbetween agents If the interactions are independent (one-body model) the scaling is linear whereas if the interactionsare imitative (two-body model) the scaling is square-rooted

We now turn to the simulation results and their impli-cation Figure 4 shows how in a network of 5000 agents thefraction of RW populist supporters increases when there isa constant inflow of outsiders here 119869 = 2 per month Thisinflow is annually approximately 05 of the total populationwhich is slightly more than the threshold value impliedin Figure 3(b) Simulations with 119869 = 2 per month aredesigned to emulate the rapid limit in extension (B) describedabove After approximately 37 years of rapid globalization thenetwork reaches a tipping point and abruptly shifts to amodedominated by RW populism RW populism dominates whenmore than 50 of the network is made up of RW populistsupporters (ie 119875 gt 05 where 119875 denotes the fraction of RWpopulists) The threshold is 50 because in a democracy themajority rules

Figure 4 also shows how the simulated network underassumptions (i)ndash(iii) responds to shocks (red curve) At thisstage extension (B) does not yet operateThe constant annualinflow of outsiders 119869 = 2 is supplemented by two eventsat times 1199051 and 1199052 when 119869 = 200 The state of the networkcharacterized by the proportion of RWpopulists (119875) exhibitsa much stronger response at 1199052 than at 1199051 although the shockinflow (119869 = 200) is the same This occurs because at 1199052 the

8 Complexity

08

00

02

J = 2

t1

J = 2 except at

t2

Change in tolerance

500

04

06

Time

Prob

abili

ty o

f RW

pop

ulism

t1 amp t2

Figure 4 Nonlinearity a tipping point and the rise of right-wingpopulism Using a network of 119873 = 5000 agents each with anaverage of 15 connections we examine the effect of a constantinflow of outsiders at rate 119869 = 2 at each time step In this setupthe total number of outsiders at any moment in time is 119868(119905) =sum119894 119869119894 = ⟨119869⟩119905 As the fraction of outsiders 119891119868 = 119868(119873 + 119868)approaches the tolerance parameter 1198911015840119868 = 015 the presence of atipping point causes the fraction of RW populist supporters to startincreasing nonlinearly and eventually undergo a sudden jump (iea discontinuous change) at about 37 years (450 months) into thesimulation (black curve) The sudden jump happens much earlierif the inflow of outsiders experiences shocks at times 1199051 and 1199052at which 119869 = 200 outsiders enter the network In particular asthe network approaches the tipping point the effect of exactlythe same shock becomes disproportionately higher (red curve) Inthis case however the tolerance parameter is still kept constantFinally we also examine the case in which shocks at times 1199051 and1199052 affect the tolerance parameter where responsiveness is controlledby parameter 120574 = 00001 Here the second shock at 1199052 is sufficientto instantly tip the network into RW populism (green curve) Otherparameters are 119901 = 0007 120591 = 15 1199011015840 = 05 11990110158401015840 = 05 and 120572 = 0001

system is closer to the tipping point and consequently moreunstable than at 1199051 Because in real-world data the value of119869 can be biased due to estimation errors or misinterpretedinformation our results suggest that approaching the tippingpoint can be concurrent with strong nonlinear effects suchthat even a small shock can trigger a transition to a modedominated by RWpopulismThis can be evenmore explosive(in terms of 119875) if extension (B) is allowed to operate that isif the tolerance parameter 1198911015840119868 changes with 119869

Figure 4 shows a third simulation (green curve) inwhich the dynamics operate under assumptions (i)ndash(iii) withextension (B) The tolerance parameter 119891119868 thus changes with119869 as prescribed by (2) Because of the decreasing tolerancethe second shock at 1199052 can now push the system beyondthe tipping point and thus the dominance of RW populismoccurs earlier than in the two previous simulations

From the simulations in Figure 4 alone it is unclearhow much the local influence of outsiders (assumption (ii))contributes to the rise of RW populism relative to the insider

08

1

06

04

02

00

TotalLocal influence (ii)Contagion (iii)

250 500

Time

Prob

abili

ty o

f RW

pop

ulism

Figure 5 Breakdown of the causes of right-wing populism Figure 4shows that the probability of RW populism 119875 suddenly increasesas society approaches a tipping point but remains silent on theunderlying causes Here we discern between the contributionsof local outsider influence (assumption (ii)) and mutual insidercontagion (assumption (iii)) Far from the tipping point 119875 mainlyresponds to local outsider influence (ii) By contrast as the networkapproaches its tipping point mutual insider contagion (iii) takesover and accelerates the transition to RW populist dominanceParameter values are119873 = 5000 with an average degree of 15 119869 = 2119901 = 0007 120591 = 15 1199011015840 = 07 and 11990110158401015840 = 08

contagion (assumption (iii)) Figure 5 shows these contribu-tions After the initial transients fade the local influence ofoutsiders drives the increase in RW populists in the networkThe contribution of mutual insider contagion is relativelysmall until the system approaches a tipping point Near thetipping point contagion spreads rapidly and overtakes thelocal outsider influence as the main contributor to the riseof RW populism Thereafter the RW populist movement cansustain itself without support from the outside

In the regime of moderate immigration inflows (1198691015840 lt 119869 lt11986910158401015840) we can examine the dynamics of our complex networkusing themean-field theory (MFT) analytic techniqueWhenthe number of agent connections does not deviate greatlyfrom the network average the probability 119875 that a randomlychosen insider agent 119894 is an RWpopulist supporter due to anyof the processes underlying assumptions (i)ndash(iii) is

119875 = 119901lowast + 11990110158401199011 (119896119894 119898119894 119891119868) + 119901101584010158401199011 (119870119894119872119894 119875)

minus 119901lowast11990110158401199011 (119896119894 119898119894 119891119868) minus 119901lowast119901101584010158401199011 (119870119894119872119894 119875)

minus 11990110158401199011 (119896119894 119898119894 119891119868) 119901101584010158401199011 (119870119894119872119894 119875)

(4)

where the last three terms avoid double counting in accor-dance with the probability theory formula 119875(119860 cup 119861 cup 119862) =119875(119860) + 119875(119861) + 119875(119862) minus 119875(119860)119875(119861) minus 119875(119860)119875(119862) minus 119875(119861)119875(119862) forthree mutually independent events 119860 119861 and 119862 that cannotoccur simultaneously In the MFT approximation we candrop index 119894 because no single agent is markedly differentfrom the collective average Previously we set 119872 = 1198702

Complexity 9

417 882

88

200 600 800 1000 1200

Time (months)

13208

1

06

04

02

00

Prob

abili

ty o

f RW

pop

ulism

400

294

263185

555

J = 1 fI = 005

J = 1 fI = 01

J = 1 fIfI

= 015

J = 2 = 005

J = 3 fI = 005

J = 3 fI = 01

J = 3 fI = 015

J = 3 fI = 02

Figure 6 Predicting the timing of RW populism We find that for abroad range of outsider inflows (119869) and tolerance parameter values(1198911015840119868 ) (5) predicts the moment at which 119875 gt 05 in a manner thatagrees favorably with the simulation results Except for 120574 = 0 otherparameters are the same as in Figure 4

for simplicity but the peer pressure measured by the valueof the proportionality factor between 119872 and 119870 can differbetween countries or regions In addition parameters 1199011015840 and11990110158401015840 are constants only in theory The real social dynamicsare such that these parameters may change in responseto rumors political manipulations or outside shocks Ifparameters 1199011015840 and 11990110158401015840 substantially increase the value of 119875also increases thus further improving the prospects for RWpopulism dominance In our framework due to democracyrsquosmajority rule principle when 119875 approaches 05 the nonlinearprocesses embedded in assumptions (ii) and (iii) cause asudden transition to RW populist mode

Because a theoretical model is more useful if it haspredictive power [54 55] we show how a network of agentsunder assumptions (i)ndash(iii) leads to a simple formula for thetiming at which RW populism starts to dominate

119905119905ℎ =1198731198911015840119868

119869 (1 minus 1198911015840119868 )minus 119868 (0)

119869 (5)

Gaining this result involves three steps First if the immi-gration inflow is constant then the number of outsidersin the network after 119905 time steps is 119868(0) + 119869119905 Second thetotal population size thus equals 119873 + 119868(0) + 119869119905 Finally (5)follows if the current fraction of outsiders (119868(0) + 119869119905)(119873 +119868(0) + 119869119905) is equated with the critical parameter 1198911015840119868 Figure 6shows that for a number of immigration inflow-toleranceparameter pairs (119869 1198911015840119868 ) the simulated timing of the shift toRW populist mode (ie 119875 gt 05) fits theoretical predictionsIn conjunction with the empirical data on tolerance towardsimmigrants in the EU countries the formula in (5) could beused to provide an estimate of when a given country mightbe approaching a possible tipping point to RW populismdominance

Numerical simulations allow us to examine not only iso-lated single networks but also the interdependence betweentwo or more networks Note that there is a potential for a cas-cade effect when an RW populist movement in one networkcauses the rise of RW populist movements in other networksThis cascade effect is growing in relevance because expandingglobalization is causing countries to become increasinglysimilar to each another and this similarity increases in suchsupranational organizations as the EU in which bordersbetween nation states are rapidly fading

To examine how interdependence affects the rise of RWpopulism we set up two random economically equivalentER networks (equal 119901 in the model) with different tolerancelevels towards outsiders (different 1198911015840119868 in the model) Tothe usual intraconnections within each network we addinterconnecting agents that linkwith their counterparts in theother network Apart from this addition we retain the samemodel assumptions as previously held

To examine the effects of interconnectedness we first runnumerical simulations of two independent networks withoutinterconnections [see Figure 7(a)] As expected when theinflow level of outsiders is the same (119869 = 2) the networkwith a higher tolerance parameter (11989110158401198681 = 04) reaches thetipping point much later than the network with a lowertolerance parameter (11989110158401198682 = 02) When the networks areinterconnected and the more tolerant network experiencesan increased inflow (1198692 = 4) and a shock (119869 = 500) at time1199051 = 500 its susceptibility to RW populism increases andit also affects the other network and shortens the time oftransition to RW populism [see Figure 7(b)] It would appearthat countries do not want to be the first to cross the line butin an interconnected world being second is easier

4 Discussion and Conclusion

Why some countries (eg the ex-socialist EU countries)strongly oppose receiving immigrants while others (eg theUSA) have a long history of receiving immigrants is animportant topic in the social sciences and more recently amajor issue for the EU Perhaps receiving immigrants hasbeen an ongoing pattern in the USA because there is nosingle dominant ethnicity and thus a clear distinction ismadebetween USA national identity and the country of originof its citizens In addition because USA citizens are peoplefrom all over the world there is no single dominant religiousethnic or cultural group that can organize and threaten theestablished social order In France we find the oppositeThereis a large group of immigrants with different language andreligion from the French majority whose presence can instillfear among themajority Fear exacerbated by an inflow rate ofimmigrants that exceeds the rate of their integration can leadto a volatile situation one that is often resolved in one of twoways Either there is an immigrant uprising as exemplifiedby the Visigoth immigrants and their ex-Roman commanderAlaricwhoplunderedRome in 410 or themajority populationsuppresses the inflow of immigrants Currently this secondoption is often accomplished by supporting populist right-wing parties

10 Complexity

08

1

00

06

0402

J1 = 2

J2 = 2

200015001000500

Prob

abili

ty o

f RW

pop

ulism

Time

(a)

08

1

00

06

2000

0402

J1 = 2

J2 = 4

Time15001000500Pr

obab

ility

of R

W p

opul

ism

(b)

Figure 7 Interconnected networks or why ldquosomebody elsersquos problemrdquo easily turns into ldquomy problemrdquo In (a) we show the case when there areno interlinks between networks The tolerance parameters between the two networks differ 1198911198681 = 02 and 1198911198681 = 04 while the inflows intoboth networks are the same 1198691 = 1198692 = 2 (b) More tolerant network is now exposed to a higher inflow 1198692 = 4 and a shock at 1199051 = 500 Theaverage number of connections for intraconnections (interconnections) in both networks equals 15 (10) The other parameters are the sameas in Figure 4

Although globalization was conceived to allow capitaland labor to move freely across national borders real-worldglobalization is affected by a multitude of noneconomicfactors such as ethnicity culture and religion With so manyfactors at play globalization is an enormously complex pro-cess and often noneconomic factors do not align with purelyeconomic factors This misalignment can lead to frustrationsthat feed populist movements By opposing the collaborationthat comes with globalization populist movements act asa feedback mechanism that pushes the general populationtowards becoming more ideologically rigid and this inturn further accelerates the populism [56] A strengthenedpopulist movement can also trigger tectonic shifts in worldaffairs as exemplified by BREXIT in the UK and the recent2016 US presidential election

Problems in a globalized world rarely confine themselvesto one place Interdependencemakes the developed countriesmore alike and synchronizes their social dynamics Thissynchronization can cause political shifts in one country tospill over into other countries and this is what enables the riseof RW populism to spread across large regions of the worldAfter BREXIT and the 2016 US presidential election politicalelites should not expect to continue business as usual Beingthe first to adopt amajor political shift with a high probabilityof negative economic consequences is difficult but once thatline has been crossed the interdependence of globalizationmakes cascades (domino effects) an active possibility No onewants to be first but many are ready to be second

The tipping point that brings on the rise of RW populismmay be reached more quickly when voters face a binarychoice for example the ldquoyesrdquo or ldquonordquo choice in the UKBREXIT referendum and the two major-party candidates inthe recent US presidential election In both cases the populistoption secured a narrow victory As mentioned above thepolls indicate that the populist party in Austria can expect toreceive 34 of the votes but the populist party candidate inthe presidential race can expect close to 51 A simple wayto understand these percentages is to assume that politicalattitudes of voters are approximately symmetrical in their

distribution across the political spectrum from left to rightConsequently if leftist voters comprise 120595119871 of the totalpopulation then RW populist voters will maintain a similarpresence that is 120595119877 asymp 120595119871 Facing this binary choice thecentrist voters have no one representing their views and thusare likely to vote evenly between the two available optionsAn implication is that when 120595119871 asymp 120595119877 asymp 33 even aslight (statistically significant) imbalance in favor of120595119877over120595119871 can tip the society towards RW populism In Austria120595119877 asymp 36 seems to be sufficient to make the RW populistcandidate a front runner in the presidential race

The final outcome of the battle between the conflictingfactors surrounding globalization will almost surely havetremendous economic implications If a country approachesits tipping point how will the ensuing volatility affect itslong-term credit rating If a domino effect cascades acrossa large region of the EU or the entire EU how will thataffect the Euro and the common banking system Withouta proper resolution of the migrant crisis what will be theimpact on systemic risk In an attempt to shed light onsome of the factors and underlying processes that affectthe success of globalization we offer an empirically backedtheoretical model of the rise of RW populism in response tounsustainable immigration inflows

Our model emphasizes the need for controlled globaliza-tion in which immigration inflows into a society are balancedwith the ability of the society to integrate the immigrantsThis ability is arguably improved when immigrants mixwith the native population which is a principle practicedin Singapore where immigrant hubs are discouraged andtenants in government-built housing (comprising 88 ofall housing) must be of mixed ethnic origin [20] Becausetolerance towards immigrants is conditional when immigra-tion inflows overshadow integration rates the society can betipped into RWpopulism an intolerant mode of functioningThe rise of RW populism can occur because elections areby their nature stochastic and they resemble a mathematicalrandom walker Left to its own devices a random walkerwill eventually hit an absorbing barrier Here this barrier

Complexity 11

of course is a metaphor for the demise of globalizationironically at the very hands of the progressive system (iedemocracy) thatmade globalization possible in the first place

Disclosure

A version of this work was presented in Warsaw at ldquoEcono-physics Colloquium 2017rdquo

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

The authors are grateful to S Galam and T Lipic forhelpful suggestions Boris Podobnik received the supportfrom Slovenian Research Agency (ARRS) via Project J5-8236and from the University of Rijeka Boris Podobnik and H EStanley received support from the Defense Threat ReductionAgency (DTRA) theOffice ofNaval Research (ONR) and theNational Science Foundation (NSF) (Grant CMMI 1125290)Marko Jusup was supported by the Japan Science andTechnology Agency (JST) Program to Disseminate TenureTracking System

References

[1] H Grubel and A Scott ldquoThe international flow of humancapitalrdquo American Economic Review vol 56 pp 268ndash274 1966

[2] G C Hufbauer and K Suominen Globalization at Risk Chal-lenges to Finance and Trade Yale University Press 2010

[3] F Docquier and H Rapoport ldquoGlobalization Brain Drain andDevelopmentrdquo Journal of Economic Literature vol 50 pp 681ndash730 2012

[4] D Rodrik The Globalization Paradox Democracy and theFuture of the World Economy W W Norton amp Company NewYork USA 2011

[5] B R Chiswick ldquoThe Effect of Americanization on the Earningsof Foreign-born Menrdquo Journal of Political Economy vol 86 no5 pp 897ndash921 1978

[6] G J BorjasHeavenrsquosDoor Immigration Policy and theAmericanEconomy Princeton University Press 1999

[7] G J Borjas ldquoThe economics of immigrationrdquo Journal ofEconomic Literature vol 32 pp 1667ndash1717 1994

[8] G J Borjas ldquoAssimilation Changes in Cohort Quality and theEarnings of Immigrantsrdquo Journal of Labor Economics vol 3 no4 pp 463ndash489 1985

[9] D Swank and H G Betz ldquoGlobalization the welfare stateand right-wing populism in Western Europerdquo Socio-EconomicReview vol 1 no 2 pp 215ndash245 2003

[10] J Gibson and D McKenzie ldquoEight Questions about BrainDrainsrdquo Journal of Economic Perspectives vol 25 no 3 pp 107ndash128 2011

[11] H Betz ldquoPolitics of Resentment Right-Wing Radicalism inWest Germanyrdquo Comparative Politics vol 23 no 1 pp 45ndash601990

[12] R Knight ldquoHaider the freedom party and the extreme right inAustriardquo Parliamentary Affairs vol 45 no 3 pp 285ndash299 1992

[13] P Fysh and J Wolfreys ldquoLe pen the National Front and theextreme right in Francerdquo Parliamentary Affairs vol 45 no 3pp 309ndash326 1992

[14] G Voerman and P Lucardie ldquoThe extreme right in the Nether-lands The centrists and their radical rivalsrdquo European Journalof Political Research vol 22 no 1 pp 35ndash54 1992

[15] WD Chapin ldquoExplaining the electoral success of the new rightThe German caserdquoWest European Politics vol 20 no 2 pp 53ndash72 1997

[16] J M Smith ldquoDoes crime pay Issue ownership political oppor-tunity and the populist right in Western Europerdquo ComparativePolitical Studies vol 43 no 11 pp 1471ndash1498 2010

[17] B Bonikowski and PDiMaggio ldquoVarieties of American PopularNationalismrdquo American Sociological Review vol 81 no 5 pp949ndash980 2016

[18] D Rodrik ldquoHow far will international economic integrationgordquo Journal of Economic Perspectives vol 14 no 1 pp 177ndash1862000

[19] B Davis ldquoDespite his heritage prominent economist backs im-migration cutrdquo The Wall Street Journal 1996 httpswwwhksharvardedufsgborjaspublicationspopularWSJ042696pdf

[20] B Podobnik M Jusup Z Wang and H E Stanley ldquoHow fearof future outcomes affects social dynamicsrdquo Journal of StatisticalPhysics vol 167 no 3-4 pp 1007ndash1019 2017

[21] R Reuveny ldquoClimate change-induced migration and violentconflictrdquo Political Geography vol 26 no 6 pp 656ndash673 2007

[22] T C Schelling ldquoDynamic models of segregationrdquo Journal ofMathematical Sociology vol 1 pp 143ndash186 1971

[23] R Axelrod ldquoThe dissemination of culture a model withlocal convergence and global polarizationrdquo Journal of ConflictResolution vol 41 no 2 pp 203ndash226 1997

[24] M McPherson L Smith-Lovin and J M Cook ldquoBirds of aFeather Homophily in Social Networksrdquo Annual Review ofSociology vol 27 pp 415ndash444 2001

[25] T Antal P L Krapivsky and S Redner ldquoDynamics of socialbalance on networksrdquo Physical Review E vol 72 Article ID036121 2005

[26] S Marvel J Jon Kleinbergb D Robert R D Kleinberg andS Strogatz ldquoContinuous-Time Model of Structural BalancerdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 108 no 5 pp 1771ndash1776 2011

[27] C Gracia-Lazaro L M Flora and Y Moreno ldquoSelectiveAdvantage of Tolerant Cultural Traits in the Axelrod-SchellingModelrdquo Physical Review E vol 83 Article ID 056103 2011

[28] F Aguiar and A Parravano ldquoTolerating the IntolerantHomophily Intolerance and Segregation in Social BalancedNetworksrdquo Journal of Conflict Resolution vol 59 no 1 pp 29ndash50 2015

[29] M Lim R Metzler and Y Bar-Yam ldquoGlobal pattern formationand ethniccultural violencerdquo Science vol 317 no 5844 pp1540ndash1544 2007

[30] R M Dancygier Immigration and Conflict in Europe Studies inComparative Politics Cambridge University Press 2010

[31] R M Dancygier and D D Laitin ldquoImmigration into EuropeEconomic Discrimination Violence and Public Policyrdquo AnnualReview of Political Science vol 17 pp 43ndash64 2014

[32] S Galam and M A Javarone ldquoModeling Radicalization Phe-nomena in Heterogeneous Populationsrdquo PLoS ONE vol 11Article ID e0155407 2016

Complexity 7

An alternative would be to assume a distribution of tolerancelevels in which case 1198911015840119868 would represent the mean [20]

We now expand assumption (ii) with extension (A) whenthe immigration inflow 119869 is below some threshold 1198691015840 thesociety becomes more tolerant When 119869 lt 1198691015840 at a giventime moment the tolerance parameter 1198911015840119868 increases by anamount Δ1198911015840119868 = 120575 gt 0 there is a balance between immigra-tion and immigrant integrationmdashoutsiders are successfullyintegratedmdashand insiders are able to acclimate to the changesin their society Figure 2(b) shows that the 1198691015840 value forGermany is approximately 10000 people per month Beingable to accurately determine the maximum 1198691015840 value is highlyrelevant to the success of globalization According to ourmodel immigration and integration can be the inevitableconsequences of globalization when 119869 lt 1198691015840 If this is notthe case globalization will be threatened by the rise of RWpopulism the democratic system will enter an intolerantmode and cooperation between nations will be downgradedon the list of political priorities

Empirical evidence suggests that we add an oppositeextension (B) as the inflow of outsiders 119869 crosses somethreshold 11986910158401015840 (which does not need to be equal to 1198691015840)society becomes less tolerant Mathematically extension (B)indicates that when 119869 gt 11986910158401015840 the tolerance parameter 1198911015840119868decreases by

Δ1198911015840119868 = minus120574119869 (2)

Using the econometric models in (1) we decrease thetolerance parameter proportional to inflow 119869 where 120574 isa proportionality coefficient expressing the sensitivity ofinsiders to high levels of outsider inflow Figure 3(b) showsthreshold 11986910158401015840 in terms of total population that is 11986910158401015840 = 120572119873where 120572 is another proportionality coefficient Figure 3(b)estimates the 120572 value when all of the EU countries are hitby the migrant crisis The dashed line is annual inflow belowwhich countries have a mixed response to immigration andabove which support for RW populism increases Because120572 = 11986910158401015840119873 and using Figure 3(b) 1211986910158401015840119873 asymp 0004 we obtain

120572 asymp 000033 (3)

Extensions (A) and (B) are opposite limiting cases onein which immigration is slow and the other in whichimmigration is rapid Apart from the empirical evidencethat these extensions are needed brain science for exampleoffers a physiological interpretation political attitudes have acounterpart in brain structure [50ndash52] If outsiders increaseat a rate and in a manner perceived as controllable byinsiders the prefrontal cortex of the human brain responsiblefor decision making and for moderating social behavioracclimates to the new circumstances but if the insidersperceive the outsiders to be invaders the prefrontal cortexis supplanted by the amygdala which induces a fightingreaction and tolerance is suppressed

Although with assumptions (i) and (ii) our modelaccounts for the processes that affect individual insider opin-ion the expansion of RW populism can become extremelyrapid when insiders are influenced by their peers This well-documented phenomenon in human interactions is further

accelerated when social media is added Thus the spread ofRW populism can be highly nonlinear much like the spreadof a highly contagious disease We include this nonlinear col-lective spreading mechanism in our third model assumption

Assumption iii (mutual insider contagion) At any givenmoment 119905 an insider agent 119894 with 119870119894 connections to otherinsiders turns to RW populism at rate 11990110158401015840 if at time 119905 minus 1this agent has at least119872119894 = 1198701198942 RW populist supporters intheir neighborhood Note that for simplicity factor 12 playsa role analogous to the tolerance parameter in assumption(ii) Because of connections between insider agents whenRW populism emerges anywhere in the network the populistmovement is able to spread like a contagion This collectivespreading indicates that insider agents can become RWpopulist supporters even when there are no outsiders inthe immediate neighborhood Thus some EU countries withalmost no immigration are opposed to accepting even a smallgroup of immigrants This may have affected the outcomeof the recent US presidential election in which the winningcandidate was often ridiculed in the mainstream mediamdashanattitude represented in our model by assumption (i)

The model here assumes imitative interactions withouttesting the validity of such an assumption However thisassumption could be tested by themethod of Agliari et al [53]if the required data were available The authors use the toolsfrom statistical mechanics to determine from data the natureof interactions in social customs such as local and mixedmarriages in Italy and neighboring European countries Thefraction of actualmarriages satisfying a given condition scalesdifferently with the fraction of all possible couples that satisfythe same condition depending on the nature of interactionsbetween agents If the interactions are independent (one-body model) the scaling is linear whereas if the interactionsare imitative (two-body model) the scaling is square-rooted

We now turn to the simulation results and their impli-cation Figure 4 shows how in a network of 5000 agents thefraction of RW populist supporters increases when there isa constant inflow of outsiders here 119869 = 2 per month Thisinflow is annually approximately 05 of the total populationwhich is slightly more than the threshold value impliedin Figure 3(b) Simulations with 119869 = 2 per month aredesigned to emulate the rapid limit in extension (B) describedabove After approximately 37 years of rapid globalization thenetwork reaches a tipping point and abruptly shifts to amodedominated by RW populism RW populism dominates whenmore than 50 of the network is made up of RW populistsupporters (ie 119875 gt 05 where 119875 denotes the fraction of RWpopulists) The threshold is 50 because in a democracy themajority rules

Figure 4 also shows how the simulated network underassumptions (i)ndash(iii) responds to shocks (red curve) At thisstage extension (B) does not yet operateThe constant annualinflow of outsiders 119869 = 2 is supplemented by two eventsat times 1199051 and 1199052 when 119869 = 200 The state of the networkcharacterized by the proportion of RWpopulists (119875) exhibitsa much stronger response at 1199052 than at 1199051 although the shockinflow (119869 = 200) is the same This occurs because at 1199052 the

8 Complexity

08

00

02

J = 2

t1

J = 2 except at

t2

Change in tolerance

500

04

06

Time

Prob

abili

ty o

f RW

pop

ulism

t1 amp t2

Figure 4 Nonlinearity a tipping point and the rise of right-wingpopulism Using a network of 119873 = 5000 agents each with anaverage of 15 connections we examine the effect of a constantinflow of outsiders at rate 119869 = 2 at each time step In this setupthe total number of outsiders at any moment in time is 119868(119905) =sum119894 119869119894 = ⟨119869⟩119905 As the fraction of outsiders 119891119868 = 119868(119873 + 119868)approaches the tolerance parameter 1198911015840119868 = 015 the presence of atipping point causes the fraction of RW populist supporters to startincreasing nonlinearly and eventually undergo a sudden jump (iea discontinuous change) at about 37 years (450 months) into thesimulation (black curve) The sudden jump happens much earlierif the inflow of outsiders experiences shocks at times 1199051 and 1199052at which 119869 = 200 outsiders enter the network In particular asthe network approaches the tipping point the effect of exactlythe same shock becomes disproportionately higher (red curve) Inthis case however the tolerance parameter is still kept constantFinally we also examine the case in which shocks at times 1199051 and1199052 affect the tolerance parameter where responsiveness is controlledby parameter 120574 = 00001 Here the second shock at 1199052 is sufficientto instantly tip the network into RW populism (green curve) Otherparameters are 119901 = 0007 120591 = 15 1199011015840 = 05 11990110158401015840 = 05 and 120572 = 0001

system is closer to the tipping point and consequently moreunstable than at 1199051 Because in real-world data the value of119869 can be biased due to estimation errors or misinterpretedinformation our results suggest that approaching the tippingpoint can be concurrent with strong nonlinear effects suchthat even a small shock can trigger a transition to a modedominated by RWpopulismThis can be evenmore explosive(in terms of 119875) if extension (B) is allowed to operate that isif the tolerance parameter 1198911015840119868 changes with 119869

Figure 4 shows a third simulation (green curve) inwhich the dynamics operate under assumptions (i)ndash(iii) withextension (B) The tolerance parameter 119891119868 thus changes with119869 as prescribed by (2) Because of the decreasing tolerancethe second shock at 1199052 can now push the system beyondthe tipping point and thus the dominance of RW populismoccurs earlier than in the two previous simulations

From the simulations in Figure 4 alone it is unclearhow much the local influence of outsiders (assumption (ii))contributes to the rise of RW populism relative to the insider

08

1

06

04

02

00

TotalLocal influence (ii)Contagion (iii)

250 500

Time

Prob

abili

ty o

f RW

pop

ulism

Figure 5 Breakdown of the causes of right-wing populism Figure 4shows that the probability of RW populism 119875 suddenly increasesas society approaches a tipping point but remains silent on theunderlying causes Here we discern between the contributionsof local outsider influence (assumption (ii)) and mutual insidercontagion (assumption (iii)) Far from the tipping point 119875 mainlyresponds to local outsider influence (ii) By contrast as the networkapproaches its tipping point mutual insider contagion (iii) takesover and accelerates the transition to RW populist dominanceParameter values are119873 = 5000 with an average degree of 15 119869 = 2119901 = 0007 120591 = 15 1199011015840 = 07 and 11990110158401015840 = 08

contagion (assumption (iii)) Figure 5 shows these contribu-tions After the initial transients fade the local influence ofoutsiders drives the increase in RW populists in the networkThe contribution of mutual insider contagion is relativelysmall until the system approaches a tipping point Near thetipping point contagion spreads rapidly and overtakes thelocal outsider influence as the main contributor to the riseof RW populism Thereafter the RW populist movement cansustain itself without support from the outside

In the regime of moderate immigration inflows (1198691015840 lt 119869 lt11986910158401015840) we can examine the dynamics of our complex networkusing themean-field theory (MFT) analytic techniqueWhenthe number of agent connections does not deviate greatlyfrom the network average the probability 119875 that a randomlychosen insider agent 119894 is an RWpopulist supporter due to anyof the processes underlying assumptions (i)ndash(iii) is

119875 = 119901lowast + 11990110158401199011 (119896119894 119898119894 119891119868) + 119901101584010158401199011 (119870119894119872119894 119875)

minus 119901lowast11990110158401199011 (119896119894 119898119894 119891119868) minus 119901lowast119901101584010158401199011 (119870119894119872119894 119875)

minus 11990110158401199011 (119896119894 119898119894 119891119868) 119901101584010158401199011 (119870119894119872119894 119875)

(4)

where the last three terms avoid double counting in accor-dance with the probability theory formula 119875(119860 cup 119861 cup 119862) =119875(119860) + 119875(119861) + 119875(119862) minus 119875(119860)119875(119861) minus 119875(119860)119875(119862) minus 119875(119861)119875(119862) forthree mutually independent events 119860 119861 and 119862 that cannotoccur simultaneously In the MFT approximation we candrop index 119894 because no single agent is markedly differentfrom the collective average Previously we set 119872 = 1198702

Complexity 9

417 882

88

200 600 800 1000 1200

Time (months)

13208

1

06

04

02

00

Prob

abili

ty o

f RW

pop

ulism

400

294

263185

555

J = 1 fI = 005

J = 1 fI = 01

J = 1 fIfI

= 015

J = 2 = 005

J = 3 fI = 005

J = 3 fI = 01

J = 3 fI = 015

J = 3 fI = 02

Figure 6 Predicting the timing of RW populism We find that for abroad range of outsider inflows (119869) and tolerance parameter values(1198911015840119868 ) (5) predicts the moment at which 119875 gt 05 in a manner thatagrees favorably with the simulation results Except for 120574 = 0 otherparameters are the same as in Figure 4

for simplicity but the peer pressure measured by the valueof the proportionality factor between 119872 and 119870 can differbetween countries or regions In addition parameters 1199011015840 and11990110158401015840 are constants only in theory The real social dynamicsare such that these parameters may change in responseto rumors political manipulations or outside shocks Ifparameters 1199011015840 and 11990110158401015840 substantially increase the value of 119875also increases thus further improving the prospects for RWpopulism dominance In our framework due to democracyrsquosmajority rule principle when 119875 approaches 05 the nonlinearprocesses embedded in assumptions (ii) and (iii) cause asudden transition to RW populist mode

Because a theoretical model is more useful if it haspredictive power [54 55] we show how a network of agentsunder assumptions (i)ndash(iii) leads to a simple formula for thetiming at which RW populism starts to dominate

119905119905ℎ =1198731198911015840119868

119869 (1 minus 1198911015840119868 )minus 119868 (0)

119869 (5)

Gaining this result involves three steps First if the immi-gration inflow is constant then the number of outsidersin the network after 119905 time steps is 119868(0) + 119869119905 Second thetotal population size thus equals 119873 + 119868(0) + 119869119905 Finally (5)follows if the current fraction of outsiders (119868(0) + 119869119905)(119873 +119868(0) + 119869119905) is equated with the critical parameter 1198911015840119868 Figure 6shows that for a number of immigration inflow-toleranceparameter pairs (119869 1198911015840119868 ) the simulated timing of the shift toRW populist mode (ie 119875 gt 05) fits theoretical predictionsIn conjunction with the empirical data on tolerance towardsimmigrants in the EU countries the formula in (5) could beused to provide an estimate of when a given country mightbe approaching a possible tipping point to RW populismdominance

Numerical simulations allow us to examine not only iso-lated single networks but also the interdependence betweentwo or more networks Note that there is a potential for a cas-cade effect when an RW populist movement in one networkcauses the rise of RW populist movements in other networksThis cascade effect is growing in relevance because expandingglobalization is causing countries to become increasinglysimilar to each another and this similarity increases in suchsupranational organizations as the EU in which bordersbetween nation states are rapidly fading

To examine how interdependence affects the rise of RWpopulism we set up two random economically equivalentER networks (equal 119901 in the model) with different tolerancelevels towards outsiders (different 1198911015840119868 in the model) Tothe usual intraconnections within each network we addinterconnecting agents that linkwith their counterparts in theother network Apart from this addition we retain the samemodel assumptions as previously held

To examine the effects of interconnectedness we first runnumerical simulations of two independent networks withoutinterconnections [see Figure 7(a)] As expected when theinflow level of outsiders is the same (119869 = 2) the networkwith a higher tolerance parameter (11989110158401198681 = 04) reaches thetipping point much later than the network with a lowertolerance parameter (11989110158401198682 = 02) When the networks areinterconnected and the more tolerant network experiencesan increased inflow (1198692 = 4) and a shock (119869 = 500) at time1199051 = 500 its susceptibility to RW populism increases andit also affects the other network and shortens the time oftransition to RW populism [see Figure 7(b)] It would appearthat countries do not want to be the first to cross the line butin an interconnected world being second is easier

4 Discussion and Conclusion

Why some countries (eg the ex-socialist EU countries)strongly oppose receiving immigrants while others (eg theUSA) have a long history of receiving immigrants is animportant topic in the social sciences and more recently amajor issue for the EU Perhaps receiving immigrants hasbeen an ongoing pattern in the USA because there is nosingle dominant ethnicity and thus a clear distinction ismadebetween USA national identity and the country of originof its citizens In addition because USA citizens are peoplefrom all over the world there is no single dominant religiousethnic or cultural group that can organize and threaten theestablished social order In France we find the oppositeThereis a large group of immigrants with different language andreligion from the French majority whose presence can instillfear among themajority Fear exacerbated by an inflow rate ofimmigrants that exceeds the rate of their integration can leadto a volatile situation one that is often resolved in one of twoways Either there is an immigrant uprising as exemplifiedby the Visigoth immigrants and their ex-Roman commanderAlaricwhoplunderedRome in 410 or themajority populationsuppresses the inflow of immigrants Currently this secondoption is often accomplished by supporting populist right-wing parties

10 Complexity

08

1

00

06

0402

J1 = 2

J2 = 2

200015001000500

Prob

abili

ty o

f RW

pop

ulism

Time

(a)

08

1

00

06

2000

0402

J1 = 2

J2 = 4

Time15001000500Pr

obab

ility

of R

W p

opul

ism

(b)

Figure 7 Interconnected networks or why ldquosomebody elsersquos problemrdquo easily turns into ldquomy problemrdquo In (a) we show the case when there areno interlinks between networks The tolerance parameters between the two networks differ 1198911198681 = 02 and 1198911198681 = 04 while the inflows intoboth networks are the same 1198691 = 1198692 = 2 (b) More tolerant network is now exposed to a higher inflow 1198692 = 4 and a shock at 1199051 = 500 Theaverage number of connections for intraconnections (interconnections) in both networks equals 15 (10) The other parameters are the sameas in Figure 4

Although globalization was conceived to allow capitaland labor to move freely across national borders real-worldglobalization is affected by a multitude of noneconomicfactors such as ethnicity culture and religion With so manyfactors at play globalization is an enormously complex pro-cess and often noneconomic factors do not align with purelyeconomic factors This misalignment can lead to frustrationsthat feed populist movements By opposing the collaborationthat comes with globalization populist movements act asa feedback mechanism that pushes the general populationtowards becoming more ideologically rigid and this inturn further accelerates the populism [56] A strengthenedpopulist movement can also trigger tectonic shifts in worldaffairs as exemplified by BREXIT in the UK and the recent2016 US presidential election

Problems in a globalized world rarely confine themselvesto one place Interdependencemakes the developed countriesmore alike and synchronizes their social dynamics Thissynchronization can cause political shifts in one country tospill over into other countries and this is what enables the riseof RW populism to spread across large regions of the worldAfter BREXIT and the 2016 US presidential election politicalelites should not expect to continue business as usual Beingthe first to adopt amajor political shift with a high probabilityof negative economic consequences is difficult but once thatline has been crossed the interdependence of globalizationmakes cascades (domino effects) an active possibility No onewants to be first but many are ready to be second

The tipping point that brings on the rise of RW populismmay be reached more quickly when voters face a binarychoice for example the ldquoyesrdquo or ldquonordquo choice in the UKBREXIT referendum and the two major-party candidates inthe recent US presidential election In both cases the populistoption secured a narrow victory As mentioned above thepolls indicate that the populist party in Austria can expect toreceive 34 of the votes but the populist party candidate inthe presidential race can expect close to 51 A simple wayto understand these percentages is to assume that politicalattitudes of voters are approximately symmetrical in their

distribution across the political spectrum from left to rightConsequently if leftist voters comprise 120595119871 of the totalpopulation then RW populist voters will maintain a similarpresence that is 120595119877 asymp 120595119871 Facing this binary choice thecentrist voters have no one representing their views and thusare likely to vote evenly between the two available optionsAn implication is that when 120595119871 asymp 120595119877 asymp 33 even aslight (statistically significant) imbalance in favor of120595119877over120595119871 can tip the society towards RW populism In Austria120595119877 asymp 36 seems to be sufficient to make the RW populistcandidate a front runner in the presidential race

The final outcome of the battle between the conflictingfactors surrounding globalization will almost surely havetremendous economic implications If a country approachesits tipping point how will the ensuing volatility affect itslong-term credit rating If a domino effect cascades acrossa large region of the EU or the entire EU how will thataffect the Euro and the common banking system Withouta proper resolution of the migrant crisis what will be theimpact on systemic risk In an attempt to shed light onsome of the factors and underlying processes that affectthe success of globalization we offer an empirically backedtheoretical model of the rise of RW populism in response tounsustainable immigration inflows

Our model emphasizes the need for controlled globaliza-tion in which immigration inflows into a society are balancedwith the ability of the society to integrate the immigrantsThis ability is arguably improved when immigrants mixwith the native population which is a principle practicedin Singapore where immigrant hubs are discouraged andtenants in government-built housing (comprising 88 ofall housing) must be of mixed ethnic origin [20] Becausetolerance towards immigrants is conditional when immigra-tion inflows overshadow integration rates the society can betipped into RWpopulism an intolerant mode of functioningThe rise of RW populism can occur because elections areby their nature stochastic and they resemble a mathematicalrandom walker Left to its own devices a random walkerwill eventually hit an absorbing barrier Here this barrier

Complexity 11

of course is a metaphor for the demise of globalizationironically at the very hands of the progressive system (iedemocracy) thatmade globalization possible in the first place

Disclosure

A version of this work was presented in Warsaw at ldquoEcono-physics Colloquium 2017rdquo

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

The authors are grateful to S Galam and T Lipic forhelpful suggestions Boris Podobnik received the supportfrom Slovenian Research Agency (ARRS) via Project J5-8236and from the University of Rijeka Boris Podobnik and H EStanley received support from the Defense Threat ReductionAgency (DTRA) theOffice ofNaval Research (ONR) and theNational Science Foundation (NSF) (Grant CMMI 1125290)Marko Jusup was supported by the Japan Science andTechnology Agency (JST) Program to Disseminate TenureTracking System

References

[1] H Grubel and A Scott ldquoThe international flow of humancapitalrdquo American Economic Review vol 56 pp 268ndash274 1966

[2] G C Hufbauer and K Suominen Globalization at Risk Chal-lenges to Finance and Trade Yale University Press 2010

[3] F Docquier and H Rapoport ldquoGlobalization Brain Drain andDevelopmentrdquo Journal of Economic Literature vol 50 pp 681ndash730 2012

[4] D Rodrik The Globalization Paradox Democracy and theFuture of the World Economy W W Norton amp Company NewYork USA 2011

[5] B R Chiswick ldquoThe Effect of Americanization on the Earningsof Foreign-born Menrdquo Journal of Political Economy vol 86 no5 pp 897ndash921 1978

[6] G J BorjasHeavenrsquosDoor Immigration Policy and theAmericanEconomy Princeton University Press 1999

[7] G J Borjas ldquoThe economics of immigrationrdquo Journal ofEconomic Literature vol 32 pp 1667ndash1717 1994

[8] G J Borjas ldquoAssimilation Changes in Cohort Quality and theEarnings of Immigrantsrdquo Journal of Labor Economics vol 3 no4 pp 463ndash489 1985

[9] D Swank and H G Betz ldquoGlobalization the welfare stateand right-wing populism in Western Europerdquo Socio-EconomicReview vol 1 no 2 pp 215ndash245 2003

[10] J Gibson and D McKenzie ldquoEight Questions about BrainDrainsrdquo Journal of Economic Perspectives vol 25 no 3 pp 107ndash128 2011

[11] H Betz ldquoPolitics of Resentment Right-Wing Radicalism inWest Germanyrdquo Comparative Politics vol 23 no 1 pp 45ndash601990

[12] R Knight ldquoHaider the freedom party and the extreme right inAustriardquo Parliamentary Affairs vol 45 no 3 pp 285ndash299 1992

[13] P Fysh and J Wolfreys ldquoLe pen the National Front and theextreme right in Francerdquo Parliamentary Affairs vol 45 no 3pp 309ndash326 1992

[14] G Voerman and P Lucardie ldquoThe extreme right in the Nether-lands The centrists and their radical rivalsrdquo European Journalof Political Research vol 22 no 1 pp 35ndash54 1992

[15] WD Chapin ldquoExplaining the electoral success of the new rightThe German caserdquoWest European Politics vol 20 no 2 pp 53ndash72 1997

[16] J M Smith ldquoDoes crime pay Issue ownership political oppor-tunity and the populist right in Western Europerdquo ComparativePolitical Studies vol 43 no 11 pp 1471ndash1498 2010

[17] B Bonikowski and PDiMaggio ldquoVarieties of American PopularNationalismrdquo American Sociological Review vol 81 no 5 pp949ndash980 2016

[18] D Rodrik ldquoHow far will international economic integrationgordquo Journal of Economic Perspectives vol 14 no 1 pp 177ndash1862000

[19] B Davis ldquoDespite his heritage prominent economist backs im-migration cutrdquo The Wall Street Journal 1996 httpswwwhksharvardedufsgborjaspublicationspopularWSJ042696pdf

[20] B Podobnik M Jusup Z Wang and H E Stanley ldquoHow fearof future outcomes affects social dynamicsrdquo Journal of StatisticalPhysics vol 167 no 3-4 pp 1007ndash1019 2017

[21] R Reuveny ldquoClimate change-induced migration and violentconflictrdquo Political Geography vol 26 no 6 pp 656ndash673 2007

[22] T C Schelling ldquoDynamic models of segregationrdquo Journal ofMathematical Sociology vol 1 pp 143ndash186 1971

[23] R Axelrod ldquoThe dissemination of culture a model withlocal convergence and global polarizationrdquo Journal of ConflictResolution vol 41 no 2 pp 203ndash226 1997

[24] M McPherson L Smith-Lovin and J M Cook ldquoBirds of aFeather Homophily in Social Networksrdquo Annual Review ofSociology vol 27 pp 415ndash444 2001

[25] T Antal P L Krapivsky and S Redner ldquoDynamics of socialbalance on networksrdquo Physical Review E vol 72 Article ID036121 2005

[26] S Marvel J Jon Kleinbergb D Robert R D Kleinberg andS Strogatz ldquoContinuous-Time Model of Structural BalancerdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 108 no 5 pp 1771ndash1776 2011

[27] C Gracia-Lazaro L M Flora and Y Moreno ldquoSelectiveAdvantage of Tolerant Cultural Traits in the Axelrod-SchellingModelrdquo Physical Review E vol 83 Article ID 056103 2011

[28] F Aguiar and A Parravano ldquoTolerating the IntolerantHomophily Intolerance and Segregation in Social BalancedNetworksrdquo Journal of Conflict Resolution vol 59 no 1 pp 29ndash50 2015

[29] M Lim R Metzler and Y Bar-Yam ldquoGlobal pattern formationand ethniccultural violencerdquo Science vol 317 no 5844 pp1540ndash1544 2007

[30] R M Dancygier Immigration and Conflict in Europe Studies inComparative Politics Cambridge University Press 2010

[31] R M Dancygier and D D Laitin ldquoImmigration into EuropeEconomic Discrimination Violence and Public Policyrdquo AnnualReview of Political Science vol 17 pp 43ndash64 2014

[32] S Galam and M A Javarone ldquoModeling Radicalization Phe-nomena in Heterogeneous Populationsrdquo PLoS ONE vol 11Article ID e0155407 2016

8 Complexity

08

00

02

J = 2

t1

J = 2 except at

t2

Change in tolerance

500

04

06

Time

Prob

abili

ty o

f RW

pop

ulism

t1 amp t2

Figure 4 Nonlinearity a tipping point and the rise of right-wingpopulism Using a network of 119873 = 5000 agents each with anaverage of 15 connections we examine the effect of a constantinflow of outsiders at rate 119869 = 2 at each time step In this setupthe total number of outsiders at any moment in time is 119868(119905) =sum119894 119869119894 = ⟨119869⟩119905 As the fraction of outsiders 119891119868 = 119868(119873 + 119868)approaches the tolerance parameter 1198911015840119868 = 015 the presence of atipping point causes the fraction of RW populist supporters to startincreasing nonlinearly and eventually undergo a sudden jump (iea discontinuous change) at about 37 years (450 months) into thesimulation (black curve) The sudden jump happens much earlierif the inflow of outsiders experiences shocks at times 1199051 and 1199052at which 119869 = 200 outsiders enter the network In particular asthe network approaches the tipping point the effect of exactlythe same shock becomes disproportionately higher (red curve) Inthis case however the tolerance parameter is still kept constantFinally we also examine the case in which shocks at times 1199051 and1199052 affect the tolerance parameter where responsiveness is controlledby parameter 120574 = 00001 Here the second shock at 1199052 is sufficientto instantly tip the network into RW populism (green curve) Otherparameters are 119901 = 0007 120591 = 15 1199011015840 = 05 11990110158401015840 = 05 and 120572 = 0001

system is closer to the tipping point and consequently moreunstable than at 1199051 Because in real-world data the value of119869 can be biased due to estimation errors or misinterpretedinformation our results suggest that approaching the tippingpoint can be concurrent with strong nonlinear effects suchthat even a small shock can trigger a transition to a modedominated by RWpopulismThis can be evenmore explosive(in terms of 119875) if extension (B) is allowed to operate that isif the tolerance parameter 1198911015840119868 changes with 119869

Figure 4 shows a third simulation (green curve) inwhich the dynamics operate under assumptions (i)ndash(iii) withextension (B) The tolerance parameter 119891119868 thus changes with119869 as prescribed by (2) Because of the decreasing tolerancethe second shock at 1199052 can now push the system beyondthe tipping point and thus the dominance of RW populismoccurs earlier than in the two previous simulations

From the simulations in Figure 4 alone it is unclearhow much the local influence of outsiders (assumption (ii))contributes to the rise of RW populism relative to the insider

08

1

06

04

02

00

TotalLocal influence (ii)Contagion (iii)

250 500

Time

Prob

abili

ty o

f RW

pop

ulism

Figure 5 Breakdown of the causes of right-wing populism Figure 4shows that the probability of RW populism 119875 suddenly increasesas society approaches a tipping point but remains silent on theunderlying causes Here we discern between the contributionsof local outsider influence (assumption (ii)) and mutual insidercontagion (assumption (iii)) Far from the tipping point 119875 mainlyresponds to local outsider influence (ii) By contrast as the networkapproaches its tipping point mutual insider contagion (iii) takesover and accelerates the transition to RW populist dominanceParameter values are119873 = 5000 with an average degree of 15 119869 = 2119901 = 0007 120591 = 15 1199011015840 = 07 and 11990110158401015840 = 08

contagion (assumption (iii)) Figure 5 shows these contribu-tions After the initial transients fade the local influence ofoutsiders drives the increase in RW populists in the networkThe contribution of mutual insider contagion is relativelysmall until the system approaches a tipping point Near thetipping point contagion spreads rapidly and overtakes thelocal outsider influence as the main contributor to the riseof RW populism Thereafter the RW populist movement cansustain itself without support from the outside

In the regime of moderate immigration inflows (1198691015840 lt 119869 lt11986910158401015840) we can examine the dynamics of our complex networkusing themean-field theory (MFT) analytic techniqueWhenthe number of agent connections does not deviate greatlyfrom the network average the probability 119875 that a randomlychosen insider agent 119894 is an RWpopulist supporter due to anyof the processes underlying assumptions (i)ndash(iii) is

119875 = 119901lowast + 11990110158401199011 (119896119894 119898119894 119891119868) + 119901101584010158401199011 (119870119894119872119894 119875)

minus 119901lowast11990110158401199011 (119896119894 119898119894 119891119868) minus 119901lowast119901101584010158401199011 (119870119894119872119894 119875)

minus 11990110158401199011 (119896119894 119898119894 119891119868) 119901101584010158401199011 (119870119894119872119894 119875)

(4)

where the last three terms avoid double counting in accor-dance with the probability theory formula 119875(119860 cup 119861 cup 119862) =119875(119860) + 119875(119861) + 119875(119862) minus 119875(119860)119875(119861) minus 119875(119860)119875(119862) minus 119875(119861)119875(119862) forthree mutually independent events 119860 119861 and 119862 that cannotoccur simultaneously In the MFT approximation we candrop index 119894 because no single agent is markedly differentfrom the collective average Previously we set 119872 = 1198702

Complexity 9

417 882

88

200 600 800 1000 1200

Time (months)

13208

1

06

04

02

00

Prob

abili

ty o

f RW

pop

ulism

400

294

263185

555

J = 1 fI = 005

J = 1 fI = 01

J = 1 fIfI

= 015

J = 2 = 005

J = 3 fI = 005

J = 3 fI = 01

J = 3 fI = 015

J = 3 fI = 02

Figure 6 Predicting the timing of RW populism We find that for abroad range of outsider inflows (119869) and tolerance parameter values(1198911015840119868 ) (5) predicts the moment at which 119875 gt 05 in a manner thatagrees favorably with the simulation results Except for 120574 = 0 otherparameters are the same as in Figure 4

for simplicity but the peer pressure measured by the valueof the proportionality factor between 119872 and 119870 can differbetween countries or regions In addition parameters 1199011015840 and11990110158401015840 are constants only in theory The real social dynamicsare such that these parameters may change in responseto rumors political manipulations or outside shocks Ifparameters 1199011015840 and 11990110158401015840 substantially increase the value of 119875also increases thus further improving the prospects for RWpopulism dominance In our framework due to democracyrsquosmajority rule principle when 119875 approaches 05 the nonlinearprocesses embedded in assumptions (ii) and (iii) cause asudden transition to RW populist mode

Because a theoretical model is more useful if it haspredictive power [54 55] we show how a network of agentsunder assumptions (i)ndash(iii) leads to a simple formula for thetiming at which RW populism starts to dominate

119905119905ℎ =1198731198911015840119868

119869 (1 minus 1198911015840119868 )minus 119868 (0)

119869 (5)

Gaining this result involves three steps First if the immi-gration inflow is constant then the number of outsidersin the network after 119905 time steps is 119868(0) + 119869119905 Second thetotal population size thus equals 119873 + 119868(0) + 119869119905 Finally (5)follows if the current fraction of outsiders (119868(0) + 119869119905)(119873 +119868(0) + 119869119905) is equated with the critical parameter 1198911015840119868 Figure 6shows that for a number of immigration inflow-toleranceparameter pairs (119869 1198911015840119868 ) the simulated timing of the shift toRW populist mode (ie 119875 gt 05) fits theoretical predictionsIn conjunction with the empirical data on tolerance towardsimmigrants in the EU countries the formula in (5) could beused to provide an estimate of when a given country mightbe approaching a possible tipping point to RW populismdominance

Numerical simulations allow us to examine not only iso-lated single networks but also the interdependence betweentwo or more networks Note that there is a potential for a cas-cade effect when an RW populist movement in one networkcauses the rise of RW populist movements in other networksThis cascade effect is growing in relevance because expandingglobalization is causing countries to become increasinglysimilar to each another and this similarity increases in suchsupranational organizations as the EU in which bordersbetween nation states are rapidly fading

To examine how interdependence affects the rise of RWpopulism we set up two random economically equivalentER networks (equal 119901 in the model) with different tolerancelevels towards outsiders (different 1198911015840119868 in the model) Tothe usual intraconnections within each network we addinterconnecting agents that linkwith their counterparts in theother network Apart from this addition we retain the samemodel assumptions as previously held

To examine the effects of interconnectedness we first runnumerical simulations of two independent networks withoutinterconnections [see Figure 7(a)] As expected when theinflow level of outsiders is the same (119869 = 2) the networkwith a higher tolerance parameter (11989110158401198681 = 04) reaches thetipping point much later than the network with a lowertolerance parameter (11989110158401198682 = 02) When the networks areinterconnected and the more tolerant network experiencesan increased inflow (1198692 = 4) and a shock (119869 = 500) at time1199051 = 500 its susceptibility to RW populism increases andit also affects the other network and shortens the time oftransition to RW populism [see Figure 7(b)] It would appearthat countries do not want to be the first to cross the line butin an interconnected world being second is easier

4 Discussion and Conclusion

Why some countries (eg the ex-socialist EU countries)strongly oppose receiving immigrants while others (eg theUSA) have a long history of receiving immigrants is animportant topic in the social sciences and more recently amajor issue for the EU Perhaps receiving immigrants hasbeen an ongoing pattern in the USA because there is nosingle dominant ethnicity and thus a clear distinction ismadebetween USA national identity and the country of originof its citizens In addition because USA citizens are peoplefrom all over the world there is no single dominant religiousethnic or cultural group that can organize and threaten theestablished social order In France we find the oppositeThereis a large group of immigrants with different language andreligion from the French majority whose presence can instillfear among themajority Fear exacerbated by an inflow rate ofimmigrants that exceeds the rate of their integration can leadto a volatile situation one that is often resolved in one of twoways Either there is an immigrant uprising as exemplifiedby the Visigoth immigrants and their ex-Roman commanderAlaricwhoplunderedRome in 410 or themajority populationsuppresses the inflow of immigrants Currently this secondoption is often accomplished by supporting populist right-wing parties

10 Complexity

08

1

00

06

0402

J1 = 2

J2 = 2

200015001000500

Prob

abili

ty o

f RW

pop

ulism

Time

(a)

08

1

00

06

2000

0402

J1 = 2

J2 = 4

Time15001000500Pr

obab

ility

of R

W p

opul

ism

(b)

Figure 7 Interconnected networks or why ldquosomebody elsersquos problemrdquo easily turns into ldquomy problemrdquo In (a) we show the case when there areno interlinks between networks The tolerance parameters between the two networks differ 1198911198681 = 02 and 1198911198681 = 04 while the inflows intoboth networks are the same 1198691 = 1198692 = 2 (b) More tolerant network is now exposed to a higher inflow 1198692 = 4 and a shock at 1199051 = 500 Theaverage number of connections for intraconnections (interconnections) in both networks equals 15 (10) The other parameters are the sameas in Figure 4

Although globalization was conceived to allow capitaland labor to move freely across national borders real-worldglobalization is affected by a multitude of noneconomicfactors such as ethnicity culture and religion With so manyfactors at play globalization is an enormously complex pro-cess and often noneconomic factors do not align with purelyeconomic factors This misalignment can lead to frustrationsthat feed populist movements By opposing the collaborationthat comes with globalization populist movements act asa feedback mechanism that pushes the general populationtowards becoming more ideologically rigid and this inturn further accelerates the populism [56] A strengthenedpopulist movement can also trigger tectonic shifts in worldaffairs as exemplified by BREXIT in the UK and the recent2016 US presidential election

Problems in a globalized world rarely confine themselvesto one place Interdependencemakes the developed countriesmore alike and synchronizes their social dynamics Thissynchronization can cause political shifts in one country tospill over into other countries and this is what enables the riseof RW populism to spread across large regions of the worldAfter BREXIT and the 2016 US presidential election politicalelites should not expect to continue business as usual Beingthe first to adopt amajor political shift with a high probabilityof negative economic consequences is difficult but once thatline has been crossed the interdependence of globalizationmakes cascades (domino effects) an active possibility No onewants to be first but many are ready to be second

The tipping point that brings on the rise of RW populismmay be reached more quickly when voters face a binarychoice for example the ldquoyesrdquo or ldquonordquo choice in the UKBREXIT referendum and the two major-party candidates inthe recent US presidential election In both cases the populistoption secured a narrow victory As mentioned above thepolls indicate that the populist party in Austria can expect toreceive 34 of the votes but the populist party candidate inthe presidential race can expect close to 51 A simple wayto understand these percentages is to assume that politicalattitudes of voters are approximately symmetrical in their

distribution across the political spectrum from left to rightConsequently if leftist voters comprise 120595119871 of the totalpopulation then RW populist voters will maintain a similarpresence that is 120595119877 asymp 120595119871 Facing this binary choice thecentrist voters have no one representing their views and thusare likely to vote evenly between the two available optionsAn implication is that when 120595119871 asymp 120595119877 asymp 33 even aslight (statistically significant) imbalance in favor of120595119877over120595119871 can tip the society towards RW populism In Austria120595119877 asymp 36 seems to be sufficient to make the RW populistcandidate a front runner in the presidential race

The final outcome of the battle between the conflictingfactors surrounding globalization will almost surely havetremendous economic implications If a country approachesits tipping point how will the ensuing volatility affect itslong-term credit rating If a domino effect cascades acrossa large region of the EU or the entire EU how will thataffect the Euro and the common banking system Withouta proper resolution of the migrant crisis what will be theimpact on systemic risk In an attempt to shed light onsome of the factors and underlying processes that affectthe success of globalization we offer an empirically backedtheoretical model of the rise of RW populism in response tounsustainable immigration inflows

Our model emphasizes the need for controlled globaliza-tion in which immigration inflows into a society are balancedwith the ability of the society to integrate the immigrantsThis ability is arguably improved when immigrants mixwith the native population which is a principle practicedin Singapore where immigrant hubs are discouraged andtenants in government-built housing (comprising 88 ofall housing) must be of mixed ethnic origin [20] Becausetolerance towards immigrants is conditional when immigra-tion inflows overshadow integration rates the society can betipped into RWpopulism an intolerant mode of functioningThe rise of RW populism can occur because elections areby their nature stochastic and they resemble a mathematicalrandom walker Left to its own devices a random walkerwill eventually hit an absorbing barrier Here this barrier

Complexity 11

of course is a metaphor for the demise of globalizationironically at the very hands of the progressive system (iedemocracy) thatmade globalization possible in the first place

Disclosure

A version of this work was presented in Warsaw at ldquoEcono-physics Colloquium 2017rdquo

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

The authors are grateful to S Galam and T Lipic forhelpful suggestions Boris Podobnik received the supportfrom Slovenian Research Agency (ARRS) via Project J5-8236and from the University of Rijeka Boris Podobnik and H EStanley received support from the Defense Threat ReductionAgency (DTRA) theOffice ofNaval Research (ONR) and theNational Science Foundation (NSF) (Grant CMMI 1125290)Marko Jusup was supported by the Japan Science andTechnology Agency (JST) Program to Disseminate TenureTracking System

References

[1] H Grubel and A Scott ldquoThe international flow of humancapitalrdquo American Economic Review vol 56 pp 268ndash274 1966

[2] G C Hufbauer and K Suominen Globalization at Risk Chal-lenges to Finance and Trade Yale University Press 2010

[3] F Docquier and H Rapoport ldquoGlobalization Brain Drain andDevelopmentrdquo Journal of Economic Literature vol 50 pp 681ndash730 2012

[4] D Rodrik The Globalization Paradox Democracy and theFuture of the World Economy W W Norton amp Company NewYork USA 2011

[5] B R Chiswick ldquoThe Effect of Americanization on the Earningsof Foreign-born Menrdquo Journal of Political Economy vol 86 no5 pp 897ndash921 1978

[6] G J BorjasHeavenrsquosDoor Immigration Policy and theAmericanEconomy Princeton University Press 1999

[7] G J Borjas ldquoThe economics of immigrationrdquo Journal ofEconomic Literature vol 32 pp 1667ndash1717 1994

[8] G J Borjas ldquoAssimilation Changes in Cohort Quality and theEarnings of Immigrantsrdquo Journal of Labor Economics vol 3 no4 pp 463ndash489 1985

[9] D Swank and H G Betz ldquoGlobalization the welfare stateand right-wing populism in Western Europerdquo Socio-EconomicReview vol 1 no 2 pp 215ndash245 2003

[10] J Gibson and D McKenzie ldquoEight Questions about BrainDrainsrdquo Journal of Economic Perspectives vol 25 no 3 pp 107ndash128 2011

[11] H Betz ldquoPolitics of Resentment Right-Wing Radicalism inWest Germanyrdquo Comparative Politics vol 23 no 1 pp 45ndash601990

[12] R Knight ldquoHaider the freedom party and the extreme right inAustriardquo Parliamentary Affairs vol 45 no 3 pp 285ndash299 1992

[13] P Fysh and J Wolfreys ldquoLe pen the National Front and theextreme right in Francerdquo Parliamentary Affairs vol 45 no 3pp 309ndash326 1992

[14] G Voerman and P Lucardie ldquoThe extreme right in the Nether-lands The centrists and their radical rivalsrdquo European Journalof Political Research vol 22 no 1 pp 35ndash54 1992

[15] WD Chapin ldquoExplaining the electoral success of the new rightThe German caserdquoWest European Politics vol 20 no 2 pp 53ndash72 1997

[16] J M Smith ldquoDoes crime pay Issue ownership political oppor-tunity and the populist right in Western Europerdquo ComparativePolitical Studies vol 43 no 11 pp 1471ndash1498 2010

[17] B Bonikowski and PDiMaggio ldquoVarieties of American PopularNationalismrdquo American Sociological Review vol 81 no 5 pp949ndash980 2016

[18] D Rodrik ldquoHow far will international economic integrationgordquo Journal of Economic Perspectives vol 14 no 1 pp 177ndash1862000

[19] B Davis ldquoDespite his heritage prominent economist backs im-migration cutrdquo The Wall Street Journal 1996 httpswwwhksharvardedufsgborjaspublicationspopularWSJ042696pdf

[20] B Podobnik M Jusup Z Wang and H E Stanley ldquoHow fearof future outcomes affects social dynamicsrdquo Journal of StatisticalPhysics vol 167 no 3-4 pp 1007ndash1019 2017

[21] R Reuveny ldquoClimate change-induced migration and violentconflictrdquo Political Geography vol 26 no 6 pp 656ndash673 2007

[22] T C Schelling ldquoDynamic models of segregationrdquo Journal ofMathematical Sociology vol 1 pp 143ndash186 1971

[23] R Axelrod ldquoThe dissemination of culture a model withlocal convergence and global polarizationrdquo Journal of ConflictResolution vol 41 no 2 pp 203ndash226 1997

[24] M McPherson L Smith-Lovin and J M Cook ldquoBirds of aFeather Homophily in Social Networksrdquo Annual Review ofSociology vol 27 pp 415ndash444 2001

[25] T Antal P L Krapivsky and S Redner ldquoDynamics of socialbalance on networksrdquo Physical Review E vol 72 Article ID036121 2005

[26] S Marvel J Jon Kleinbergb D Robert R D Kleinberg andS Strogatz ldquoContinuous-Time Model of Structural BalancerdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 108 no 5 pp 1771ndash1776 2011

[27] C Gracia-Lazaro L M Flora and Y Moreno ldquoSelectiveAdvantage of Tolerant Cultural Traits in the Axelrod-SchellingModelrdquo Physical Review E vol 83 Article ID 056103 2011

[28] F Aguiar and A Parravano ldquoTolerating the IntolerantHomophily Intolerance and Segregation in Social BalancedNetworksrdquo Journal of Conflict Resolution vol 59 no 1 pp 29ndash50 2015

[29] M Lim R Metzler and Y Bar-Yam ldquoGlobal pattern formationand ethniccultural violencerdquo Science vol 317 no 5844 pp1540ndash1544 2007

[30] R M Dancygier Immigration and Conflict in Europe Studies inComparative Politics Cambridge University Press 2010

[31] R M Dancygier and D D Laitin ldquoImmigration into EuropeEconomic Discrimination Violence and Public Policyrdquo AnnualReview of Political Science vol 17 pp 43ndash64 2014

[32] S Galam and M A Javarone ldquoModeling Radicalization Phe-nomena in Heterogeneous Populationsrdquo PLoS ONE vol 11Article ID e0155407 2016

Complexity 9

417 882

88

200 600 800 1000 1200

Time (months)

13208

1

06

04

02

00

Prob

abili

ty o

f RW

pop

ulism

400

294

263185

555

J = 1 fI = 005

J = 1 fI = 01

J = 1 fIfI

= 015

J = 2 = 005

J = 3 fI = 005

J = 3 fI = 01

J = 3 fI = 015

J = 3 fI = 02

Figure 6 Predicting the timing of RW populism We find that for abroad range of outsider inflows (119869) and tolerance parameter values(1198911015840119868 ) (5) predicts the moment at which 119875 gt 05 in a manner thatagrees favorably with the simulation results Except for 120574 = 0 otherparameters are the same as in Figure 4

for simplicity but the peer pressure measured by the valueof the proportionality factor between 119872 and 119870 can differbetween countries or regions In addition parameters 1199011015840 and11990110158401015840 are constants only in theory The real social dynamicsare such that these parameters may change in responseto rumors political manipulations or outside shocks Ifparameters 1199011015840 and 11990110158401015840 substantially increase the value of 119875also increases thus further improving the prospects for RWpopulism dominance In our framework due to democracyrsquosmajority rule principle when 119875 approaches 05 the nonlinearprocesses embedded in assumptions (ii) and (iii) cause asudden transition to RW populist mode

Because a theoretical model is more useful if it haspredictive power [54 55] we show how a network of agentsunder assumptions (i)ndash(iii) leads to a simple formula for thetiming at which RW populism starts to dominate

119905119905ℎ =1198731198911015840119868

119869 (1 minus 1198911015840119868 )minus 119868 (0)

119869 (5)

Gaining this result involves three steps First if the immi-gration inflow is constant then the number of outsidersin the network after 119905 time steps is 119868(0) + 119869119905 Second thetotal population size thus equals 119873 + 119868(0) + 119869119905 Finally (5)follows if the current fraction of outsiders (119868(0) + 119869119905)(119873 +119868(0) + 119869119905) is equated with the critical parameter 1198911015840119868 Figure 6shows that for a number of immigration inflow-toleranceparameter pairs (119869 1198911015840119868 ) the simulated timing of the shift toRW populist mode (ie 119875 gt 05) fits theoretical predictionsIn conjunction with the empirical data on tolerance towardsimmigrants in the EU countries the formula in (5) could beused to provide an estimate of when a given country mightbe approaching a possible tipping point to RW populismdominance

Numerical simulations allow us to examine not only iso-lated single networks but also the interdependence betweentwo or more networks Note that there is a potential for a cas-cade effect when an RW populist movement in one networkcauses the rise of RW populist movements in other networksThis cascade effect is growing in relevance because expandingglobalization is causing countries to become increasinglysimilar to each another and this similarity increases in suchsupranational organizations as the EU in which bordersbetween nation states are rapidly fading

To examine how interdependence affects the rise of RWpopulism we set up two random economically equivalentER networks (equal 119901 in the model) with different tolerancelevels towards outsiders (different 1198911015840119868 in the model) Tothe usual intraconnections within each network we addinterconnecting agents that linkwith their counterparts in theother network Apart from this addition we retain the samemodel assumptions as previously held

To examine the effects of interconnectedness we first runnumerical simulations of two independent networks withoutinterconnections [see Figure 7(a)] As expected when theinflow level of outsiders is the same (119869 = 2) the networkwith a higher tolerance parameter (11989110158401198681 = 04) reaches thetipping point much later than the network with a lowertolerance parameter (11989110158401198682 = 02) When the networks areinterconnected and the more tolerant network experiencesan increased inflow (1198692 = 4) and a shock (119869 = 500) at time1199051 = 500 its susceptibility to RW populism increases andit also affects the other network and shortens the time oftransition to RW populism [see Figure 7(b)] It would appearthat countries do not want to be the first to cross the line butin an interconnected world being second is easier

4 Discussion and Conclusion

Why some countries (eg the ex-socialist EU countries)strongly oppose receiving immigrants while others (eg theUSA) have a long history of receiving immigrants is animportant topic in the social sciences and more recently amajor issue for the EU Perhaps receiving immigrants hasbeen an ongoing pattern in the USA because there is nosingle dominant ethnicity and thus a clear distinction ismadebetween USA national identity and the country of originof its citizens In addition because USA citizens are peoplefrom all over the world there is no single dominant religiousethnic or cultural group that can organize and threaten theestablished social order In France we find the oppositeThereis a large group of immigrants with different language andreligion from the French majority whose presence can instillfear among themajority Fear exacerbated by an inflow rate ofimmigrants that exceeds the rate of their integration can leadto a volatile situation one that is often resolved in one of twoways Either there is an immigrant uprising as exemplifiedby the Visigoth immigrants and their ex-Roman commanderAlaricwhoplunderedRome in 410 or themajority populationsuppresses the inflow of immigrants Currently this secondoption is often accomplished by supporting populist right-wing parties

10 Complexity

08

1

00

06

0402

J1 = 2

J2 = 2

200015001000500

Prob

abili

ty o

f RW

pop

ulism

Time

(a)

08

1

00

06

2000

0402

J1 = 2

J2 = 4

Time15001000500Pr

obab

ility

of R

W p

opul

ism

(b)

Figure 7 Interconnected networks or why ldquosomebody elsersquos problemrdquo easily turns into ldquomy problemrdquo In (a) we show the case when there areno interlinks between networks The tolerance parameters between the two networks differ 1198911198681 = 02 and 1198911198681 = 04 while the inflows intoboth networks are the same 1198691 = 1198692 = 2 (b) More tolerant network is now exposed to a higher inflow 1198692 = 4 and a shock at 1199051 = 500 Theaverage number of connections for intraconnections (interconnections) in both networks equals 15 (10) The other parameters are the sameas in Figure 4

Although globalization was conceived to allow capitaland labor to move freely across national borders real-worldglobalization is affected by a multitude of noneconomicfactors such as ethnicity culture and religion With so manyfactors at play globalization is an enormously complex pro-cess and often noneconomic factors do not align with purelyeconomic factors This misalignment can lead to frustrationsthat feed populist movements By opposing the collaborationthat comes with globalization populist movements act asa feedback mechanism that pushes the general populationtowards becoming more ideologically rigid and this inturn further accelerates the populism [56] A strengthenedpopulist movement can also trigger tectonic shifts in worldaffairs as exemplified by BREXIT in the UK and the recent2016 US presidential election

Problems in a globalized world rarely confine themselvesto one place Interdependencemakes the developed countriesmore alike and synchronizes their social dynamics Thissynchronization can cause political shifts in one country tospill over into other countries and this is what enables the riseof RW populism to spread across large regions of the worldAfter BREXIT and the 2016 US presidential election politicalelites should not expect to continue business as usual Beingthe first to adopt amajor political shift with a high probabilityof negative economic consequences is difficult but once thatline has been crossed the interdependence of globalizationmakes cascades (domino effects) an active possibility No onewants to be first but many are ready to be second

The tipping point that brings on the rise of RW populismmay be reached more quickly when voters face a binarychoice for example the ldquoyesrdquo or ldquonordquo choice in the UKBREXIT referendum and the two major-party candidates inthe recent US presidential election In both cases the populistoption secured a narrow victory As mentioned above thepolls indicate that the populist party in Austria can expect toreceive 34 of the votes but the populist party candidate inthe presidential race can expect close to 51 A simple wayto understand these percentages is to assume that politicalattitudes of voters are approximately symmetrical in their

distribution across the political spectrum from left to rightConsequently if leftist voters comprise 120595119871 of the totalpopulation then RW populist voters will maintain a similarpresence that is 120595119877 asymp 120595119871 Facing this binary choice thecentrist voters have no one representing their views and thusare likely to vote evenly between the two available optionsAn implication is that when 120595119871 asymp 120595119877 asymp 33 even aslight (statistically significant) imbalance in favor of120595119877over120595119871 can tip the society towards RW populism In Austria120595119877 asymp 36 seems to be sufficient to make the RW populistcandidate a front runner in the presidential race

The final outcome of the battle between the conflictingfactors surrounding globalization will almost surely havetremendous economic implications If a country approachesits tipping point how will the ensuing volatility affect itslong-term credit rating If a domino effect cascades acrossa large region of the EU or the entire EU how will thataffect the Euro and the common banking system Withouta proper resolution of the migrant crisis what will be theimpact on systemic risk In an attempt to shed light onsome of the factors and underlying processes that affectthe success of globalization we offer an empirically backedtheoretical model of the rise of RW populism in response tounsustainable immigration inflows

Our model emphasizes the need for controlled globaliza-tion in which immigration inflows into a society are balancedwith the ability of the society to integrate the immigrantsThis ability is arguably improved when immigrants mixwith the native population which is a principle practicedin Singapore where immigrant hubs are discouraged andtenants in government-built housing (comprising 88 ofall housing) must be of mixed ethnic origin [20] Becausetolerance towards immigrants is conditional when immigra-tion inflows overshadow integration rates the society can betipped into RWpopulism an intolerant mode of functioningThe rise of RW populism can occur because elections areby their nature stochastic and they resemble a mathematicalrandom walker Left to its own devices a random walkerwill eventually hit an absorbing barrier Here this barrier

Complexity 11

of course is a metaphor for the demise of globalizationironically at the very hands of the progressive system (iedemocracy) thatmade globalization possible in the first place

Disclosure

A version of this work was presented in Warsaw at ldquoEcono-physics Colloquium 2017rdquo

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

The authors are grateful to S Galam and T Lipic forhelpful suggestions Boris Podobnik received the supportfrom Slovenian Research Agency (ARRS) via Project J5-8236and from the University of Rijeka Boris Podobnik and H EStanley received support from the Defense Threat ReductionAgency (DTRA) theOffice ofNaval Research (ONR) and theNational Science Foundation (NSF) (Grant CMMI 1125290)Marko Jusup was supported by the Japan Science andTechnology Agency (JST) Program to Disseminate TenureTracking System

References

[1] H Grubel and A Scott ldquoThe international flow of humancapitalrdquo American Economic Review vol 56 pp 268ndash274 1966

[2] G C Hufbauer and K Suominen Globalization at Risk Chal-lenges to Finance and Trade Yale University Press 2010

[3] F Docquier and H Rapoport ldquoGlobalization Brain Drain andDevelopmentrdquo Journal of Economic Literature vol 50 pp 681ndash730 2012

[4] D Rodrik The Globalization Paradox Democracy and theFuture of the World Economy W W Norton amp Company NewYork USA 2011

[5] B R Chiswick ldquoThe Effect of Americanization on the Earningsof Foreign-born Menrdquo Journal of Political Economy vol 86 no5 pp 897ndash921 1978

[6] G J BorjasHeavenrsquosDoor Immigration Policy and theAmericanEconomy Princeton University Press 1999

[7] G J Borjas ldquoThe economics of immigrationrdquo Journal ofEconomic Literature vol 32 pp 1667ndash1717 1994

[8] G J Borjas ldquoAssimilation Changes in Cohort Quality and theEarnings of Immigrantsrdquo Journal of Labor Economics vol 3 no4 pp 463ndash489 1985

[9] D Swank and H G Betz ldquoGlobalization the welfare stateand right-wing populism in Western Europerdquo Socio-EconomicReview vol 1 no 2 pp 215ndash245 2003

[10] J Gibson and D McKenzie ldquoEight Questions about BrainDrainsrdquo Journal of Economic Perspectives vol 25 no 3 pp 107ndash128 2011

[11] H Betz ldquoPolitics of Resentment Right-Wing Radicalism inWest Germanyrdquo Comparative Politics vol 23 no 1 pp 45ndash601990

[12] R Knight ldquoHaider the freedom party and the extreme right inAustriardquo Parliamentary Affairs vol 45 no 3 pp 285ndash299 1992

[13] P Fysh and J Wolfreys ldquoLe pen the National Front and theextreme right in Francerdquo Parliamentary Affairs vol 45 no 3pp 309ndash326 1992

[14] G Voerman and P Lucardie ldquoThe extreme right in the Nether-lands The centrists and their radical rivalsrdquo European Journalof Political Research vol 22 no 1 pp 35ndash54 1992

[15] WD Chapin ldquoExplaining the electoral success of the new rightThe German caserdquoWest European Politics vol 20 no 2 pp 53ndash72 1997

[16] J M Smith ldquoDoes crime pay Issue ownership political oppor-tunity and the populist right in Western Europerdquo ComparativePolitical Studies vol 43 no 11 pp 1471ndash1498 2010

[17] B Bonikowski and PDiMaggio ldquoVarieties of American PopularNationalismrdquo American Sociological Review vol 81 no 5 pp949ndash980 2016

[18] D Rodrik ldquoHow far will international economic integrationgordquo Journal of Economic Perspectives vol 14 no 1 pp 177ndash1862000

[19] B Davis ldquoDespite his heritage prominent economist backs im-migration cutrdquo The Wall Street Journal 1996 httpswwwhksharvardedufsgborjaspublicationspopularWSJ042696pdf

[20] B Podobnik M Jusup Z Wang and H E Stanley ldquoHow fearof future outcomes affects social dynamicsrdquo Journal of StatisticalPhysics vol 167 no 3-4 pp 1007ndash1019 2017

[21] R Reuveny ldquoClimate change-induced migration and violentconflictrdquo Political Geography vol 26 no 6 pp 656ndash673 2007

[22] T C Schelling ldquoDynamic models of segregationrdquo Journal ofMathematical Sociology vol 1 pp 143ndash186 1971

[23] R Axelrod ldquoThe dissemination of culture a model withlocal convergence and global polarizationrdquo Journal of ConflictResolution vol 41 no 2 pp 203ndash226 1997

[24] M McPherson L Smith-Lovin and J M Cook ldquoBirds of aFeather Homophily in Social Networksrdquo Annual Review ofSociology vol 27 pp 415ndash444 2001

[25] T Antal P L Krapivsky and S Redner ldquoDynamics of socialbalance on networksrdquo Physical Review E vol 72 Article ID036121 2005

[26] S Marvel J Jon Kleinbergb D Robert R D Kleinberg andS Strogatz ldquoContinuous-Time Model of Structural BalancerdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 108 no 5 pp 1771ndash1776 2011

[27] C Gracia-Lazaro L M Flora and Y Moreno ldquoSelectiveAdvantage of Tolerant Cultural Traits in the Axelrod-SchellingModelrdquo Physical Review E vol 83 Article ID 056103 2011

[28] F Aguiar and A Parravano ldquoTolerating the IntolerantHomophily Intolerance and Segregation in Social BalancedNetworksrdquo Journal of Conflict Resolution vol 59 no 1 pp 29ndash50 2015

[29] M Lim R Metzler and Y Bar-Yam ldquoGlobal pattern formationand ethniccultural violencerdquo Science vol 317 no 5844 pp1540ndash1544 2007

[30] R M Dancygier Immigration and Conflict in Europe Studies inComparative Politics Cambridge University Press 2010

[31] R M Dancygier and D D Laitin ldquoImmigration into EuropeEconomic Discrimination Violence and Public Policyrdquo AnnualReview of Political Science vol 17 pp 43ndash64 2014

[32] S Galam and M A Javarone ldquoModeling Radicalization Phe-nomena in Heterogeneous Populationsrdquo PLoS ONE vol 11Article ID e0155407 2016

10 Complexity

08

1

00

06

0402

J1 = 2

J2 = 2

200015001000500

Prob

abili

ty o

f RW

pop

ulism

Time

(a)

08

1

00

06

2000

0402

J1 = 2

J2 = 4

Time15001000500Pr

obab

ility

of R

W p

opul

ism

(b)

Figure 7 Interconnected networks or why ldquosomebody elsersquos problemrdquo easily turns into ldquomy problemrdquo In (a) we show the case when there areno interlinks between networks The tolerance parameters between the two networks differ 1198911198681 = 02 and 1198911198681 = 04 while the inflows intoboth networks are the same 1198691 = 1198692 = 2 (b) More tolerant network is now exposed to a higher inflow 1198692 = 4 and a shock at 1199051 = 500 Theaverage number of connections for intraconnections (interconnections) in both networks equals 15 (10) The other parameters are the sameas in Figure 4

Although globalization was conceived to allow capitaland labor to move freely across national borders real-worldglobalization is affected by a multitude of noneconomicfactors such as ethnicity culture and religion With so manyfactors at play globalization is an enormously complex pro-cess and often noneconomic factors do not align with purelyeconomic factors This misalignment can lead to frustrationsthat feed populist movements By opposing the collaborationthat comes with globalization populist movements act asa feedback mechanism that pushes the general populationtowards becoming more ideologically rigid and this inturn further accelerates the populism [56] A strengthenedpopulist movement can also trigger tectonic shifts in worldaffairs as exemplified by BREXIT in the UK and the recent2016 US presidential election

Problems in a globalized world rarely confine themselvesto one place Interdependencemakes the developed countriesmore alike and synchronizes their social dynamics Thissynchronization can cause political shifts in one country tospill over into other countries and this is what enables the riseof RW populism to spread across large regions of the worldAfter BREXIT and the 2016 US presidential election politicalelites should not expect to continue business as usual Beingthe first to adopt amajor political shift with a high probabilityof negative economic consequences is difficult but once thatline has been crossed the interdependence of globalizationmakes cascades (domino effects) an active possibility No onewants to be first but many are ready to be second

The tipping point that brings on the rise of RW populismmay be reached more quickly when voters face a binarychoice for example the ldquoyesrdquo or ldquonordquo choice in the UKBREXIT referendum and the two major-party candidates inthe recent US presidential election In both cases the populistoption secured a narrow victory As mentioned above thepolls indicate that the populist party in Austria can expect toreceive 34 of the votes but the populist party candidate inthe presidential race can expect close to 51 A simple wayto understand these percentages is to assume that politicalattitudes of voters are approximately symmetrical in their

distribution across the political spectrum from left to rightConsequently if leftist voters comprise 120595119871 of the totalpopulation then RW populist voters will maintain a similarpresence that is 120595119877 asymp 120595119871 Facing this binary choice thecentrist voters have no one representing their views and thusare likely to vote evenly between the two available optionsAn implication is that when 120595119871 asymp 120595119877 asymp 33 even aslight (statistically significant) imbalance in favor of120595119877over120595119871 can tip the society towards RW populism In Austria120595119877 asymp 36 seems to be sufficient to make the RW populistcandidate a front runner in the presidential race

The final outcome of the battle between the conflictingfactors surrounding globalization will almost surely havetremendous economic implications If a country approachesits tipping point how will the ensuing volatility affect itslong-term credit rating If a domino effect cascades acrossa large region of the EU or the entire EU how will thataffect the Euro and the common banking system Withouta proper resolution of the migrant crisis what will be theimpact on systemic risk In an attempt to shed light onsome of the factors and underlying processes that affectthe success of globalization we offer an empirically backedtheoretical model of the rise of RW populism in response tounsustainable immigration inflows

Our model emphasizes the need for controlled globaliza-tion in which immigration inflows into a society are balancedwith the ability of the society to integrate the immigrantsThis ability is arguably improved when immigrants mixwith the native population which is a principle practicedin Singapore where immigrant hubs are discouraged andtenants in government-built housing (comprising 88 ofall housing) must be of mixed ethnic origin [20] Becausetolerance towards immigrants is conditional when immigra-tion inflows overshadow integration rates the society can betipped into RWpopulism an intolerant mode of functioningThe rise of RW populism can occur because elections areby their nature stochastic and they resemble a mathematicalrandom walker Left to its own devices a random walkerwill eventually hit an absorbing barrier Here this barrier

Complexity 11

of course is a metaphor for the demise of globalizationironically at the very hands of the progressive system (iedemocracy) thatmade globalization possible in the first place

Disclosure

A version of this work was presented in Warsaw at ldquoEcono-physics Colloquium 2017rdquo

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

The authors are grateful to S Galam and T Lipic forhelpful suggestions Boris Podobnik received the supportfrom Slovenian Research Agency (ARRS) via Project J5-8236and from the University of Rijeka Boris Podobnik and H EStanley received support from the Defense Threat ReductionAgency (DTRA) theOffice ofNaval Research (ONR) and theNational Science Foundation (NSF) (Grant CMMI 1125290)Marko Jusup was supported by the Japan Science andTechnology Agency (JST) Program to Disseminate TenureTracking System

References

[1] H Grubel and A Scott ldquoThe international flow of humancapitalrdquo American Economic Review vol 56 pp 268ndash274 1966

[2] G C Hufbauer and K Suominen Globalization at Risk Chal-lenges to Finance and Trade Yale University Press 2010

[3] F Docquier and H Rapoport ldquoGlobalization Brain Drain andDevelopmentrdquo Journal of Economic Literature vol 50 pp 681ndash730 2012

[4] D Rodrik The Globalization Paradox Democracy and theFuture of the World Economy W W Norton amp Company NewYork USA 2011

[5] B R Chiswick ldquoThe Effect of Americanization on the Earningsof Foreign-born Menrdquo Journal of Political Economy vol 86 no5 pp 897ndash921 1978

[6] G J BorjasHeavenrsquosDoor Immigration Policy and theAmericanEconomy Princeton University Press 1999

[7] G J Borjas ldquoThe economics of immigrationrdquo Journal ofEconomic Literature vol 32 pp 1667ndash1717 1994

[8] G J Borjas ldquoAssimilation Changes in Cohort Quality and theEarnings of Immigrantsrdquo Journal of Labor Economics vol 3 no4 pp 463ndash489 1985

[9] D Swank and H G Betz ldquoGlobalization the welfare stateand right-wing populism in Western Europerdquo Socio-EconomicReview vol 1 no 2 pp 215ndash245 2003

[10] J Gibson and D McKenzie ldquoEight Questions about BrainDrainsrdquo Journal of Economic Perspectives vol 25 no 3 pp 107ndash128 2011

[11] H Betz ldquoPolitics of Resentment Right-Wing Radicalism inWest Germanyrdquo Comparative Politics vol 23 no 1 pp 45ndash601990

[12] R Knight ldquoHaider the freedom party and the extreme right inAustriardquo Parliamentary Affairs vol 45 no 3 pp 285ndash299 1992

[13] P Fysh and J Wolfreys ldquoLe pen the National Front and theextreme right in Francerdquo Parliamentary Affairs vol 45 no 3pp 309ndash326 1992

[14] G Voerman and P Lucardie ldquoThe extreme right in the Nether-lands The centrists and their radical rivalsrdquo European Journalof Political Research vol 22 no 1 pp 35ndash54 1992

[15] WD Chapin ldquoExplaining the electoral success of the new rightThe German caserdquoWest European Politics vol 20 no 2 pp 53ndash72 1997

[16] J M Smith ldquoDoes crime pay Issue ownership political oppor-tunity and the populist right in Western Europerdquo ComparativePolitical Studies vol 43 no 11 pp 1471ndash1498 2010

[17] B Bonikowski and PDiMaggio ldquoVarieties of American PopularNationalismrdquo American Sociological Review vol 81 no 5 pp949ndash980 2016

[18] D Rodrik ldquoHow far will international economic integrationgordquo Journal of Economic Perspectives vol 14 no 1 pp 177ndash1862000

[19] B Davis ldquoDespite his heritage prominent economist backs im-migration cutrdquo The Wall Street Journal 1996 httpswwwhksharvardedufsgborjaspublicationspopularWSJ042696pdf

[20] B Podobnik M Jusup Z Wang and H E Stanley ldquoHow fearof future outcomes affects social dynamicsrdquo Journal of StatisticalPhysics vol 167 no 3-4 pp 1007ndash1019 2017

[21] R Reuveny ldquoClimate change-induced migration and violentconflictrdquo Political Geography vol 26 no 6 pp 656ndash673 2007

[22] T C Schelling ldquoDynamic models of segregationrdquo Journal ofMathematical Sociology vol 1 pp 143ndash186 1971

[23] R Axelrod ldquoThe dissemination of culture a model withlocal convergence and global polarizationrdquo Journal of ConflictResolution vol 41 no 2 pp 203ndash226 1997

[24] M McPherson L Smith-Lovin and J M Cook ldquoBirds of aFeather Homophily in Social Networksrdquo Annual Review ofSociology vol 27 pp 415ndash444 2001

[25] T Antal P L Krapivsky and S Redner ldquoDynamics of socialbalance on networksrdquo Physical Review E vol 72 Article ID036121 2005

[26] S Marvel J Jon Kleinbergb D Robert R D Kleinberg andS Strogatz ldquoContinuous-Time Model of Structural BalancerdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 108 no 5 pp 1771ndash1776 2011

[27] C Gracia-Lazaro L M Flora and Y Moreno ldquoSelectiveAdvantage of Tolerant Cultural Traits in the Axelrod-SchellingModelrdquo Physical Review E vol 83 Article ID 056103 2011

[28] F Aguiar and A Parravano ldquoTolerating the IntolerantHomophily Intolerance and Segregation in Social BalancedNetworksrdquo Journal of Conflict Resolution vol 59 no 1 pp 29ndash50 2015

[29] M Lim R Metzler and Y Bar-Yam ldquoGlobal pattern formationand ethniccultural violencerdquo Science vol 317 no 5844 pp1540ndash1544 2007

[30] R M Dancygier Immigration and Conflict in Europe Studies inComparative Politics Cambridge University Press 2010

[31] R M Dancygier and D D Laitin ldquoImmigration into EuropeEconomic Discrimination Violence and Public Policyrdquo AnnualReview of Political Science vol 17 pp 43ndash64 2014

[32] S Galam and M A Javarone ldquoModeling Radicalization Phe-nomena in Heterogeneous Populationsrdquo PLoS ONE vol 11Article ID e0155407 2016

Complexity 11

of course is a metaphor for the demise of globalizationironically at the very hands of the progressive system (iedemocracy) thatmade globalization possible in the first place

Disclosure

A version of this work was presented in Warsaw at ldquoEcono-physics Colloquium 2017rdquo

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

The authors are grateful to S Galam and T Lipic forhelpful suggestions Boris Podobnik received the supportfrom Slovenian Research Agency (ARRS) via Project J5-8236and from the University of Rijeka Boris Podobnik and H EStanley received support from the Defense Threat ReductionAgency (DTRA) theOffice ofNaval Research (ONR) and theNational Science Foundation (NSF) (Grant CMMI 1125290)Marko Jusup was supported by the Japan Science andTechnology Agency (JST) Program to Disseminate TenureTracking System

References

[1] H Grubel and A Scott ldquoThe international flow of humancapitalrdquo American Economic Review vol 56 pp 268ndash274 1966

[2] G C Hufbauer and K Suominen Globalization at Risk Chal-lenges to Finance and Trade Yale University Press 2010

[3] F Docquier and H Rapoport ldquoGlobalization Brain Drain andDevelopmentrdquo Journal of Economic Literature vol 50 pp 681ndash730 2012

[4] D Rodrik The Globalization Paradox Democracy and theFuture of the World Economy W W Norton amp Company NewYork USA 2011

[5] B R Chiswick ldquoThe Effect of Americanization on the Earningsof Foreign-born Menrdquo Journal of Political Economy vol 86 no5 pp 897ndash921 1978

[6] G J BorjasHeavenrsquosDoor Immigration Policy and theAmericanEconomy Princeton University Press 1999

[7] G J Borjas ldquoThe economics of immigrationrdquo Journal ofEconomic Literature vol 32 pp 1667ndash1717 1994

[8] G J Borjas ldquoAssimilation Changes in Cohort Quality and theEarnings of Immigrantsrdquo Journal of Labor Economics vol 3 no4 pp 463ndash489 1985

[9] D Swank and H G Betz ldquoGlobalization the welfare stateand right-wing populism in Western Europerdquo Socio-EconomicReview vol 1 no 2 pp 215ndash245 2003

[10] J Gibson and D McKenzie ldquoEight Questions about BrainDrainsrdquo Journal of Economic Perspectives vol 25 no 3 pp 107ndash128 2011

[11] H Betz ldquoPolitics of Resentment Right-Wing Radicalism inWest Germanyrdquo Comparative Politics vol 23 no 1 pp 45ndash601990

[12] R Knight ldquoHaider the freedom party and the extreme right inAustriardquo Parliamentary Affairs vol 45 no 3 pp 285ndash299 1992

[13] P Fysh and J Wolfreys ldquoLe pen the National Front and theextreme right in Francerdquo Parliamentary Affairs vol 45 no 3pp 309ndash326 1992

[14] G Voerman and P Lucardie ldquoThe extreme right in the Nether-lands The centrists and their radical rivalsrdquo European Journalof Political Research vol 22 no 1 pp 35ndash54 1992

[15] WD Chapin ldquoExplaining the electoral success of the new rightThe German caserdquoWest European Politics vol 20 no 2 pp 53ndash72 1997

[16] J M Smith ldquoDoes crime pay Issue ownership political oppor-tunity and the populist right in Western Europerdquo ComparativePolitical Studies vol 43 no 11 pp 1471ndash1498 2010

[17] B Bonikowski and PDiMaggio ldquoVarieties of American PopularNationalismrdquo American Sociological Review vol 81 no 5 pp949ndash980 2016

[18] D Rodrik ldquoHow far will international economic integrationgordquo Journal of Economic Perspectives vol 14 no 1 pp 177ndash1862000

[19] B Davis ldquoDespite his heritage prominent economist backs im-migration cutrdquo The Wall Street Journal 1996 httpswwwhksharvardedufsgborjaspublicationspopularWSJ042696pdf

[20] B Podobnik M Jusup Z Wang and H E Stanley ldquoHow fearof future outcomes affects social dynamicsrdquo Journal of StatisticalPhysics vol 167 no 3-4 pp 1007ndash1019 2017

[21] R Reuveny ldquoClimate change-induced migration and violentconflictrdquo Political Geography vol 26 no 6 pp 656ndash673 2007

[22] T C Schelling ldquoDynamic models of segregationrdquo Journal ofMathematical Sociology vol 1 pp 143ndash186 1971

[23] R Axelrod ldquoThe dissemination of culture a model withlocal convergence and global polarizationrdquo Journal of ConflictResolution vol 41 no 2 pp 203ndash226 1997

[24] M McPherson L Smith-Lovin and J M Cook ldquoBirds of aFeather Homophily in Social Networksrdquo Annual Review ofSociology vol 27 pp 415ndash444 2001

[25] T Antal P L Krapivsky and S Redner ldquoDynamics of socialbalance on networksrdquo Physical Review E vol 72 Article ID036121 2005

[26] S Marvel J Jon Kleinbergb D Robert R D Kleinberg andS Strogatz ldquoContinuous-Time Model of Structural BalancerdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 108 no 5 pp 1771ndash1776 2011

[27] C Gracia-Lazaro L M Flora and Y Moreno ldquoSelectiveAdvantage of Tolerant Cultural Traits in the Axelrod-SchellingModelrdquo Physical Review E vol 83 Article ID 056103 2011

[28] F Aguiar and A Parravano ldquoTolerating the IntolerantHomophily Intolerance and Segregation in Social BalancedNetworksrdquo Journal of Conflict Resolution vol 59 no 1 pp 29ndash50 2015

[29] M Lim R Metzler and Y Bar-Yam ldquoGlobal pattern formationand ethniccultural violencerdquo Science vol 317 no 5844 pp1540ndash1544 2007

[30] R M Dancygier Immigration and Conflict in Europe Studies inComparative Politics Cambridge University Press 2010

[31] R M Dancygier and D D Laitin ldquoImmigration into EuropeEconomic Discrimination Violence and Public Policyrdquo AnnualReview of Political Science vol 17 pp 43ndash64 2014

[32] S Galam and M A Javarone ldquoModeling Radicalization Phe-nomena in Heterogeneous Populationsrdquo PLoS ONE vol 11Article ID e0155407 2016

12 Complexity

[33] R J Sampson S W Raudenbush and F Earls ldquoNeighborhoodsand violent crime a multilevel study of collective efficacyrdquoScience vol 277 no 5328 pp 918ndash924 1997

[34] httpwwwunhcrorg[35] httpwwwlucifycomthe-flow-towards-europe[36] httpwwwelectographcom[37] C Castellano MMarsilli and A Vespignani ldquoNonequilibrium

Phase Transition in a Model for Social Influencerdquo PhysicalReview Letters vol 85 p 3536 2000

[38] D Card A Mas and J Rothstein ldquoTipping and the Dynamicsof SegragationrdquoTheQuarterly Journal of Economics vol 123 no1 pp 177ndash218 2008

[39] httpsenwikipediaorgwikiList of non-state terrorist incidents[40] httpwwwmipexeuwhat-is-mipex[41] H Betz ldquoThe New Politics of Resentment Radical Right-Wing

Populist Parties in Western Europerdquo Comparative Politics vol25 no 4 pp 413ndash427 1993

[42] K PopperThe Open Society and Its Enemies The Spell of Platovol 1 Routledge UK 1945

[43] A Majdanzic B Podobnik S V Buldyrev Y Kenett SHavlin and H E Stanley ldquoSpontaneous recovery in dynamicalnetworksrdquo Nature Physics vol 10 pp 34ndash38 2014

[44] B Podobnik V Vukovic and H E Stanley ldquoDoes the wagegap between private and public sectors encourage politicalcorruptionrdquoPLoSONE vol 10 no 10 Article ID e0141211 2015

[45] J-H Lee M Jusup B Podobnik and Y Iwasa ldquoAgent-basedmapping of credit risk for sustainablemicrofinancerdquo PLoSONEvol 10 no 5 Article ID e0126447 2015

[46] D J Watts ldquoA simple model of global cascades on randomnetworksrdquo Proceedings of the National Academy of Sciences ofthe United States of America vol 99 no 9 pp 5766ndash5771 2002

[47] httptheconversationcomhard-evidence-how-areas-with-low-immigration-voted-mainly-for-brexit-62138

[48] httpwwwdwcomengolden-dawn-trial-begins-in-greecea-18393111

[49] B Podobnik D Horvatic T Lipic M Perc J M Buldu and HE Stanley ldquoTheCost ofAttack inCompetingNetworksrdquo Journalof The Royal Society Interface vol 12 Article ID 20150770 2015

[50] R Kanai T Feilden C Firth and G Rees ldquoPolitical Orien-tations Are Correlated with Brain Structure in Young AdultsrdquoCurrent Biology vol 21 no 8 pp 677ndash680 2011

[51] G Zamboni M Gozzi F Krueger J-R Duhamel A Siriguand J Grafman ldquoIndividualism conservatism and radicalismas criteria for processing political beliefs a parametric fMRIstudyrdquo Social Neuroscience vol 4 no 5 pp 367ndash383 2009

[52] D R Oxley K B Smith J R Alford et al ldquoPolitical attitudesvary with physiological traitsrdquo Science vol 321 no 5896 pp1667ndash1670 2008

[53] E Agliari A Barra A Galluzzi M A Javarone A Pizzoferratoand D Tantari ldquoEmerging heterogeneities in Italian customsand comparison with nearby countriesrdquo PLoS ONE vol 10 no12 Article ID e0144643 2015

[54] S Galam ldquoThe Trump phenomenon an explanation fromsociophysicsrdquo International Journal of Modern Physics B vol 31no 10 17 pages 2017

[55] R Kennedy S Wojcik and D Lazer ldquoImproving electionprediction internationallyrdquo Science vol 355 no 6324 pp 515ndash520 2017

[56] M Jusup T Matsuo and Y Iwasa ldquoBarriers to CooperationAid Ideological Rigidity and Threaten Societal Collapserdquo PLoSComputational Biology vol 10 no 5 Article ID e1003618 2014

Submit your manuscripts athttpswwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MathematicsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Mathematical Problems in Engineering

Hindawi Publishing Corporationhttpwwwhindawicom

Differential EquationsInternational Journal of

Volume 2014

Applied MathematicsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Mathematical PhysicsAdvances in

Complex AnalysisJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OptimizationJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Operations ResearchAdvances in

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Function Spaces

Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of Mathematics and Mathematical Sciences

Hindawi Publishing Corporationhttpwwwhindawicom Volume 201

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Algebra

Discrete Dynamics in Nature and Society

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Decision SciencesAdvances in

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom

Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Stochastic AnalysisInternational Journal of