Come Together: Far-Right Parties and Mainstream Coalitions

While far-right parties tend to receive a small minority of votes in national elections, their presence in ruling coalitions is becoming much more common. In this article, I ask under what conditions mainstream parties are willing and interested in forming a coalition with a far-right party, given the potentially high costs associated with having such a partner in government. I characterize such moves as the co-optation of a growing political rival in an effort to minimize electoral threat. That is, as far-right parties become more threatening to the electoral success of a mainstream party, they will invite the party into their government, in an effort to stave off said threat. This characterization borrows from the literature of authoritarian co-optation to build on our current understanding of parliamentary coalition-building. Quantitative analysis utilizing cross-national, survey and spatial data is employed to support this theory.

governing with such partners. For one, there may be a backlash from betrayed voters (see e.g. BBC News 2018) who baulk at the idea of their mainstream party uniting with such radical partners. Second, there are costs relating to governing: radical parties are typically inexperienced in government and may damage the performance of their coalition partner. Third, there are international costs, as allies and intergovernmental organizations (IGOs) fear they pose a potential threat to liberal democracy. In the aftermath of the far-right Freedom Party of Austria (FPÖ) and mainstream Austrian People's Party (ÖVP) 1999 coalition, Austria immediately faced economic sanctions and reduced communications from European allies, with the European Union proclaiming 'the admission of the FPÖ into a coalition government legitimizes the extreme right in Europe' (Meret 2010).
Much of the literature on coalition formation has characterized the process spatially, as parties selecting partners closest in ideological space (Maravall 2010). Yet this alone does not completely explain recent events (De Lange 2012). There has been, at the same time, efforts by mainstream parties to 'co-opt' growing far-right opponents in an effort to minimize the electoral threat they pose (Meguid 2005;Van Spanje and De Graaf 2018). In addition, many have noticed that government inclusion can bring a 'taming' effect on far-right party behaviour (Minkenberg 2001(Minkenberg , 2013, often leading to toned-down rhetoric and a renewed focus on realistic policies (Akkerman et al. 2016;Heinisch 2003;Van Spanje and Van Der Brug 2007). Others have noted that such moves may be a strategic response to the growing far right as an alternative to the more widely understood strategies of cordons and policy co-optation (Minkenberg 2013). Thus I argue that a crucial element to our understanding of these far-right coalitions involves the desire of mainstream parties to co-opt their far-right rivals. When radical parties pose no electoral threat there is little reason to pay them any attention. If, however, they begin rising in popularity and pose a real threat, inviting them into coalition may become less costly than allowing them to continue as an opponent. This comes about as farright parties can be largely characterized as protest parties, with their time in opposition spent railing against the government and dominant parties. Once they become part of the government, however, we often see the protesting stop, meaning the radicals' damage can be diminished. While this effect of taming farright parties via coalition has been noticed by others, this article serves as the first to formally theorize and empirically test its utilization as a strategic method of co-optation.
Utilizing evidence from over 200 European coalitions, I find that as conditions change to increase the threat of the far right's ongoing attacks on the mainstream, and thus are prime for an imminent rise in their support, they are much more likely to be included in a county's governing coalition, even while their seat share and ideological distance to the ruling party is unchanged. In an attempt to home in on this concept of 'threat', I provide evidence from three analyses utilizing macro-level, survey and spatial data. The results are consistent with this theory of co-optation.

Coalition formation in Europe
Generally, research on coalition formation can be broken down into two paradigms: office-seeking (concerned with gaining political power) and policy-seeking (concerned with enacting policy change). Thus, the two most important factors in predicting coalition formation are seat share and ideology (Müller and Strøm 2003). Office-seeking theories argue that politicians simply have a goal of getting into office. William Riker (1962) argues that politicians will form minimum winning coalitions (MWC)a coalition that passes the threshold for a majority by as small a margin as possible. This gives the coalition a majority control of government but divides the rewards of maintaining control among the fewest officeholders, maximizing individual reward. However, simple office-seeking approaches have not been well supported empirically (Müller and Strøm 2003). Rather, party ideology must be considered as well.
Thus, many scholars have advanced policy-seeking theories of coalition formation. Robert Axelrod (1970) and Abram de Swaan (1973) add ideological constraints to the MWC theory of formation, changing our conception instead to a minimum connected winning coalition (MCWC). In this way, coalition formation is understood spatially, describing parties as ideologically placed on a left-right dimension, where coalition formation depends on the Euclidean distance between parties (Maravall 2010). Parties are not simply interested in forming coalitions to hold majority power but have policy preferences as well, and thus will form coalitions with parties that have similar preferences (Budge and Laver 1993). More recent research has continued to find party size and ideology to be dominant in predicting coalition formation (Back and Dumont 2008;Debus and Gross 2016;Martin and Stevenson 2001).

The costs of radical parties in coalition government
Forming coalitions with a far-right party is surprising, then, not only due to the parties' extreme ideology, which decreases their chances of coalition inclusion (Döring and Hellström 2013), but also due to the strong threats that exist for the mainstream partner. These threats can be broken down into two categories: domestic and international. In terms of domestic threats, high costs may present themselves for the mainstream partner. It is likely mainstream party voters will feel betrayed by the inclusion of such an ideologically radical party in government. Coalition formation is influenced in part by voters' preferences for certain ruling partners (Debus and Muller 2013;Falco-Gimeno 2012), and likewise, vote choice is partially determined by prospective ideas on possible emergent coalitions (Bowler et al. 2010;Herrmann 2014). Therefore, parties should expect some backlash when these preferences are ignored. Further, research has demonstrated that coalition partnership is often utilized as a heuristic by voters in determining a party's ideology (Fortunato and Stevenson 2013): thus partnering with a far-right party may lead voters to infer the mainstream partner has become much more radical ideologically, and perceptions of drastic ideological shifts tend to be quite costly to parties.
Far-right parties also tend to be inexperienced in government. They are generally characterized as 'protest parties', thus their transition to governing brings with it many difficulties that the mainstream party will have to endure. Their governance is correlated with high levels of executive turnover, and an elevated rate of cabinet defections (Powell 1981), both likely unwanted by governing partners. For example, during the 1999 Austrian FPÖ-ÖVP coalition, the FPÖ was plagued by party infighting, organizational troubles and other setbacks after entering government (Luther 2007;Meret 2010). Due to these troubles, several prominent ministers resigned, and an early election was called.
Finally, there are clear international costs to the government when a party on the radical fringes joins a coalition. Largely, this is due to radical parties' tenuous relationship with liberal democracy. Far-right parties, who tend to be quite populist in nature (Rooduijn and Akkerman 2015), threaten liberal democracy (Houle and Kenny 2018;Kendall-Taylor and Frantz 2016). As international partners and IGOs have a clear interest in maintaining democracy, costs will likely be borne by governments that include a far-right party, as previous sanctions within the European Union have shown (Luther 2007;Meret 2010).
In sum, existing literature suggests coalition partnerships with far-right parties may be unlikely. Yet, this is exactly what has been transpiring lately. Why would mainstream parties agree to form these coalitions given the difficulties that would present themselves? It is easier to understand the calculus for the radical party. For one, entering into coalition aids in granting legitimacy, and likewise helps remove the so-called stigma of being merely a 'protest party' (Dunphy and Bale 2011). However, for the mainstream party, the reasoning seems less clear.
Of course, there are benefits to the mainstream partner in forming these coalitions as well. As I discuss in the next section, it offers them a reprieve from far-right attacks, while signalling to voters they are taking the issues at hand seriously. While some previous literature has argued this is largely a process of 'politics as usual' (e.g. De Lange 2012), others have suggested this possibility of mainstream parties capitalizing on the far right's novelty by trying to cooperate (Bale 2003;Wagner and Meyer 2017;Zaslove 2012)but this premise has never been formally theorized or empirically evaluated. Thus, in the remaining sections I put forth a theory of coalition as co-optation, and provide the first empirical assessment of it.
A theory of radical co-optation This study empirically evaluates a new logic for understanding mainstream and farright coalition partnerships, borrowing in a way from the literature on authoritarian governments' co-optation of political rivals by admitting them to national legislatures. Without doubt, mainstream parties have taken notice of the recent rise in farright success and have begun changing their behaviour to remain competitive against the growing threat: altering policy positions (Abou-Chadi and Krause 2018; Williams and Ishiyama 2018) and emphasizing different issues (Abou-Chadi 2016). It is apparent that the far right has grown immensely as a political threat in recent years, and centre-right parties in particular are most vulnerable to this threat (Akkerman 2012).
Importantly, far-right parties thrive as protest parties (Abedi 2004;Mayer and Perrineau 1992), often voicing loud, 'inappropriate' opinions aimed at the political establishment (Akkerman et al. 2016). Characterized as 'anti-system' (De Bromhead et al. 2012;Jackman and Volpert 1996), they make a name for themselves by railing against the system of government, dominant parties and national conditions. They are eager to assign blame for all the country's problems on the political elite and ruling parties (Van Der Brug and Fennema 2007). Understandably this is not ideal for mainstream parties, who rely on the current government system and the people's trust in said system to maintain power, and would rather that an entire party devoted to attacking them did not receive much attention.
Generally, a mainstream party has two tools at its disposal: ostracization and accommodation (Meguid 2005;Van Spanje and De Graaf 2018). To ostracize the party, in effect, is to invoke a cordon sanitaire, refuse to cooperate, and publicly label them as illegitimate and radical. While ostracization is beneficial to a point, Bonnie Meguid (2008) argues that as niche parties become more threatening, more accommodating strategies are needed to minimize electoral losses. The most common accommodation strategy is to co-opt policies from the far-right agenda. Often, this means mainstream parties developing more radical stances on immigration (Abou-Chadi and Krause 2018). And yet, shifting policy positions to more closely mirror far-right opponents does not appear enough to stave off electoral threat completely. Evidence shows that a more restrictive 'immigration climate'when more parties have anti-immigration policiesstill leads to more success for the far right (Lubbers et al. 2002).
At the same time, research suggests that coalition formation with far-right parties may have a 'taming' effect (e.g. Heinze 2018;Minkenberg 2001Minkenberg , 2013, noting that in many cases government participation leads to toned-down party rhetoric and a renewed focus on realistic policies (Akkermanet al. 2016;Heinisch 2003;Van Spanje and Van Der Brug 2007). Others still have noted that, while MCWC theories are still quite important in predicting the far right's participation in government, the utilization of these coalitions to 'neutralize' far-right success may well be on the mind of mainstream parties. And yet, this potential motivation has never been analysed (De Lange 2012).
Importantly, this moderation largely seems to happen only after cooperation begins (Akkerman et al. 2016). Thus, I theorize that mainstream parties may be strategically utilizing these coalitions as another tool to stop the criticism, attacks and assignments of blame, for fear of the harm these attacks may bring to their own popularity and future electoral success. It is quite difficult for a far-right party to criticize so vehemently the government administration when it is itself a member of that administration. What was once a source of such criticism and electoral threat is disarmed in government, as the far-right party tones down its critiques and focuses on more moderate policy.
In Finland for example, the decision by the largest party, Centre, to include the far-right Finns Party was largely viewed as a move to minimize the threat of the far right, with the Guardian referring to the remaining party leaders' agreement as the best option, lest they allow them to 'gain ground in opposition' (Chastand 2015), while The Economist (2019) similarly characterized Centre's move as an effort to 'tame' the Finns Party. Indeed, evidence demonstrates that once a far-right party is cooperated with, its incentive to maintain radical rhetoric is attenuated and it moderates ideologically (Van Spanje and Van Der Brug 2007). Thus, given that more accommodating strategies are needed as the far right becomes more threatening, we should expect to see the likelihood of their participation in government coalitions to increase as the level of threat they pose rises, and mainstream parties look beyond previous strategies such as ostracization to protect themselves.
Beyond this, a second mechanism may be that mainstream parties signal to voters that they take their preferences seriously. These radical parties' 'anti-system' rhetoric takes hold as some voters view themselves as 'forgotten': they were left out of the monetary spoils of modernization (Betz 1994) and increasingly feel left out of society at large (Gidron and Hall 2020). As a result, they grasp the far right's promise as an alternative to mainstream political elites (Loch and Norocel 2015). Thus, by engaging and agreeing to work with these radical parties, mainstream parties can signal to voters that they hear and understand their concerns. Including these far-right parties in government often has the effect of restoring voters' confidence in the system they once decried as broken and unfair (Muis et al. 2022). This may prevent voters from fleeing mainstream parties in favour of the far right, as their rising discontent subsides and they see the government incorporating these alternative ideas.
In sum, I argue that, when national conditions begin to shift such that the far-right platform can better take hold, and the threat of the far right (which I define in the succeeding section) increases, we will see far-right parties more likely to enter government coalitions. In these cases, the barrage of attacks is most likely to pose the greatest threat to mainstream ruling parties, and be most persuasive to voters. When this happens, the costs of ruling with the far-right partner become outweighed by the costs of allowing them to continue in opposition.

Coalition invitation and acceptance
Up until now, the process of the radical right's participation in government has been characterized as one-sided. While the larger partner generally 'invites' the smaller to form a coalition (Lees 2001), it is not simply that a government is formed once one party extends an invitation to another party. Rather, the second party must accept the offer.
Much evidence points to the idea that far-right parties do benefit from serving in government coalitions. First, serving in a government affords them some form of legitimacy they did not have (cf. Dunphy and Bale 2011) and increases voters' perception of them as a mainstream party (Wagner and Meyer 2017). As simple protest parties, it may be hard for far-right parties to make the case to voters that they are able to govern effectively. In government, however, they gain an 'official' status. They are designated by a mainstream political party as a worthy partner, capable of overseeing government ministries. Beyond granting legitimacy, this aids in making their policy impact more visible, expanding their electoral base.
While some scholars argue that this search for legitimacy only began after 2000, it is important to consider that opportunities for government participation for farright parties rose greatly after 2000, as far-right parties became more electorally successful (Akkerman et al. 2016). Thus, it is not likely that these parties had no interest in serving in government, but rather that they had first to build up vote share in order to be in a position to enter government, where they could then exert influence over policy (De Lange 2012). This is consistent with the co-optation theory, as only once they begin posing a threat to mainstream parties are far-right parties invited to join a coalition government.
This idea brings up the second reason far-right parties generally benefit from serving in coalition government: like any party, they are policy-seeking. That is, they do not simply seek to hold political office, rather they seek to implement their agenda. Unquestionably, it is easier to implement change from a place of power. Indeed, far-right parties' influence on policy while in opposition is sometimes described as limited. While the parties have succeeded in pulling mainstream party positions to the right in areas such as immigration policy, most policies ultimately passed by legislatures remain mainly moderate, and not altogether representative of the radical agenda (Mudde 2013). In government, however, we have seen widespread changes to immigration law in countries such as Austria and Italy (Fallend 2012;Zaslove 2012).
Thus, for the far-right party, the benefits of joining a government coalition are not a function of external conditions as they are for the mainstream partner. Rather, they exist as a forum to garner legitimacy and help achieve their platform. Thus, I assume the calculus remains constant: accept invitations to join coalitions, as the benefits almost always outweigh the costs. Conceptualized as a strategic game, the dominant strategy of the far-right party remains the same regardless of external factors, while for the mainstream party the payoffs depend on the level of threat posed by the radical opponent. I present the extensive form of this theoretical game in Figure A1 of the online Supplementary Material.
Measuring the threat to mainstream parties Following this theory regarding coalition membership and threat, the next step is to operationalize the threat posed by the far right. One straightforward option would be to utilize vote share won by far-right partiesyet, vote share and seat share tend to be nearly perfectly correlated. Given the simple fact that parties are more likely to enter government as their seat share grows, such a measure is inappropriate. Instead, the main mechanism of this co-optation theory argues that, given the same representation in the legislature, a far-right party is more likely to enter a governing coalition due to the future threat it poses to the dominant party system. Thus, a more abstract measure of threat is required.
To capture this conceptualization of threat more robustly, I rely on three separate operationalizations. The first two attempt to provide a latent measure for the salience of these far-right attacks, while the third offers a wholly exogenous proxy of the parties' potential future success. First, I utilize macro-level data, conceptualizing threat as the national conditions that influence the salience of far-right parties' criticisms and arguments. Second, I utilize survey data to measure congruence between public opinion and the far-right platform. Third, I utilize spatial data to measure the success of far-right parties in neighbouring countries. I discuss each of these tests in the next section.

Study 1: Macro-level conditions
A major feature of far-right parties is that they tend to find relevance in times of pronounced economic crisis (De Bromhead et al. 2012;Jackman and Volpert 1996). As the national economy worsens and becomes a more salient issue to voters (Singer 2011), the very tenets of the radical platform (e.g. immigration, unemployment) become more persuasive, and thus the far-right party may become much more threatening to the mainstream. Further, in the aftermath of an economic crisis the rhetoric of the far right, namely blaming these macro-level conditions on minorities and foreigners, dramatically increases their attractiveness to voters (Funke et al. 2016), increasing competition between the native and immigrant groups (e.g. Lubbers et al. 2002). This should serve to increase the salience of the far right's attacks, as voters are more likely to hold opinions now in line with these views. Importantly, much work has found that far-right parties in particular benefit from these economic downturns, and not all niche parties more generally (Bichay 2020;Brückner and Grüner 2010;Funke et al. 2016). Even while some scholars refute a correlation between economic decline and far-right vote share, it is hard to attack an economy that is growing at a healthy rate. Further, such arguments remain quite unconvincing to voters (and thus not threatening to the mainstream) when times are good. Given that the far right focuses much of their time on attacking mainstream politics and national-level conditions, these attacks can be much more persuasive when conditions are poor. Additionally, it is the case that these periods of economic crisis provide a greater opportunity for far-right parties to experience a performance surge, or at the very least for mainstream parties to feel threatened by these far-right attacks, as support for mainstream parties tends to falter (Downes and Loveless 2018). As this potential threat grows and eventually surpasses the costs of co-governance, the strategic response would be to invite the far right into coalition where they are no longer able to criticize the government and further threaten the incumbent party system. Thus, my first expectation is that a far-right party is much more likely to find itself in a governing coalition in times of prolonged economic crisis, when they pose a bigger threat to mainstream politics due to the enhanced salience of their attacks, and voters' increased disposition to support them.
Hypothesis 1: In times of prolonged economic crisis, the far right is more likely to enter government.
In addition, far-right parties tend to be characterized by their nativismthat is, their belief that their nation should be made up only of members of the native group (Mudde 2007). The overwhelming majority of far-right voters support them due to these anti-immigration considerations (Akkerman 2012;Arzheimer 2018;Van Der Brug et al. 2000). Centre-right parties no doubt have caught on to this connection, as far-right success has driven them to adopt a more antiimmigration stance (Abou-Chadi and Krause 2018). And yet, more restrictive immigration climates lead to more success for the far right (Lubbers et al. 2002). In other words, changing policy positions to remain competitive is not enough.
There appears to be a direct link between immigration issue salience and foreign inflows, as increases in both immigration and asylum flows increase the salience of immigration as an issue (Dennison 2020;Hatton 2021). That is, as inflows grow, citizens increasingly hold the opinion that immigration is an important threat facing the country. In times of low inflows, then, far-right attacks regarding immigration should not pose much of a threat to mainstream parties. Thus, as is the case in a poor economy, far-right parties are able to use periods of high inflows as a convincing line of attack against their mainstream rivals. Given that mainstream parties cannot simply change their policy positions to remain electorally secure, in response to this threat I expect them to invite the far-right party to join them in coalition so that they appear to voters to be taking the issue seriously, and to stave off the electoral threat of the far-right party remaining in opposition and continuing to decry the current political system's handling of the issue.
Hypothesis 2: As foreign inflows grow in a country, the far right is more likely to enter government.
Admittedly, these are rather indirect tests. However, the literature has consistently shown that centre-right parties recognize the benefit both economic crisis and increased foreign inflows bring to the far right, and thus the threat they bring them (Downes and Loveless 2018;Downes et al. 2021). Therefore, using these macro-level measures of far-right popularity allows for an approximation of the threat mainstream parties are feeling, while sidestepping any methodological concern that our independent variable may have direct effects (i.e. effects not related to mainstream party threat) on the dependent variable.

Study 2: Public opinion
While I argue the indirectness of Study 1 is beneficial and necessary due to the nature of the question, here I provide a more direct test of the theory. Rather than measure the threat of a far-right party via the contextual factors of a nation, I instead measure the degree to which voters hold opinions in line with the far-right platform by utilizing survey data over time. Survey data form a powerful alternative test as it is clear that attacks against mainstream parties are more salient when the audience of those attacks (voters) agrees with their message. As voters agree more with far-right policies, mainstream parties should feel more threatened by them and the attacks they make.
The largest motivator of a vote for a far-right party remains concerns regarding foreign immigration (Akkerman 2012;Arzheimer 2018;Van Der Brug et al. 2000). Further, evidence suggests a large component of this motivation is due to 'ideational concerns', or that immigration is a threat to the identity and culture of the nation (Golder 2003). Thus, as a more direct measure of threat I utilize a question from the European Social Survey (ESS) that asks respondents whether immigration makes their country a better or worse place to live. I argue that, as the electorate becomes more hostile to the idea of immigration, the more likely a far-right party is to enter a coalition: Hypothesis 3: As public opinion towards immigration becomes more hostile, the far right is more likely to enter government.

Study 3: Geographical electoral success
Finally, I conceptualize threat to mainstream parties via the geographical success of far-right parties. This test is especially beneficial as it is highly unlikely success in neighbouring countries has any direct effect on seat share in the country under analysis. It is, however, likely that politicians view the rise of the far right around them as a portent of potential political changes. While it is still not clear whether any direct diffusion exists, far-right success does seem to come in 'waves'. Thus, leaders in one country may feel threatened by successes in other countries, fearing the farright momentum will soon be at their door. In this way, I conceptualize the threat felt by mainstream parties as the success of far-right parties in nations close by.
For this study I define diffusion via the mechanism of 'neighbour emulation' (Brinks and Coppedge 2006;Most and Starr 1990), as this tends to be the avenue of diffusion predominantly tested in literatures such as democratic diffusion. The main justification for this model is that countries generally receive benefits from having similar regimes to their immediate neighbours (Brinks and Coppedge 2006) and are simultaneously the most at threat to spillover and domino effects. Thus, I argue a country sharing a border with another country who recently experienced a rise in far-right success will feel most threatened. Beyond this diffusion of ideals, countries who neighbour one another tend to experience common shocks in terms of the economy and immigration, which often leads to similar political outcomes (e.g. Houle et al. 2016).
Thus, I utilize spatial data and measure domestic threat as the electoral support of far-right parties in a given country's neighbours. In this case, I argue mainstream parties see the success of far-right parties around them as a signal of potential threat in their own country: Hypothesis 4: As far-right parties have more success in a country's neighbours, the far right is more likely to enter government.
No doubt, each of the above conceptualizations of threat has its own flaws, as the concept of 'threat' itself is quite impossible to measure in tangible terms without some methodological concerns. But, when taken together, these proxies provide evidence for the theorized mechanism in favour of rival explanationsof which I discuss several below. First, it may be that the factors leading to a rise in far-right support are simultaneously pulling mainstream parties further to the right, leading to ideological convergence. This closing of the ideological gap between the two may simply make them more likely to cooperate, as they now share common views on how to legislate (e.g. Abou-Chadi 2016; Abou-Chadi and Krause 2018). I account for this by including a control for party ideological range. Further, while this may be a shortcoming of Study 1, and perhaps Study 2, ideological convergence makes little sense as an explanation for the third study.
Second, mainstream parties may be partnering with a far-right party not necessarily in an attempt to prevent vote loss, but in an attempt to profit from these same conditions that benefit the far right. That is, by forming an alliance, they allow themselves to ride the same wave of popularity by gaining the partnership of a now popular political ally. In this way it is not simply that they hope to put the party in government to silence it, but rather to attempt to gain a comfortable majority and effectively govern with it on the issues salient to the public. At its core, however, this is essentially a similar motivation to avoiding electoral threat, as mainstream parties seek an effective partner to legislate these salient issues so as not to lose the support of voters. Additionally, empirical evidence suggests these mainstream governments actually make quite moderate changes to issues such as immigration reform, not altogether representative of the far right's policy platform (Mudde 2013).
Finally, we may be seeing mainstream parties choosing to form coalitions with the far right in an attempt to demonstrate national unity in the face of a crisis (e.g. economic recession or immigration shock). In this way, inviting the far right into government attempts to display a united front to deal with crises at hand (e.g. the Draghi cabinet in Italy during COVID-19, or the Papademos government of Greece during their debt crisis). Yet again, while this may well explain alternative explanations for the results in Study 1, it cannot explain the evidence found in Studies 2 and 3.
In this way, while any one of the aforementioned studies on its own may be uncompelling, together they provide persuasive evidence for this idea of co-optation.

Data and methodology
In this section I outline the empirical strategies used to test this theory of co-optation. First, I identify parties that qualify as far right via Matthijs Rooduijn et al. (2019). Their definition follows from Cas Mudde (2007), who identifies two characteristics: far-right parties are nativist and authoritarian. That is, they believe: (1) their country should be ethnically homogenous, consisting only of the native group; and (2) laws are strictly enforced, with any transgression heavily penalized. Parties identified by Rooduijn et al. (2019) are coded for Europe dating back to 1989 by country-experts and have been peer-reviewed by over 80 scholars. This umbrella term 'far right' includes both populist radical-right parties, as well as the more anti-democratic extreme right (e.g. Pirro 2022). Table A1 in the online Supplementary Material lists all far-right parties by country.
I code every governing coalition for 30 European countries as to whether or not they include a far-right party for the time period of 1989-2018. I only consider cases where far-right parties provided cabinet members, and did not simply support a government that excluded them (i.e. confidence and supply agreements). This choice is made for two reasons: first, these agreements are much less common than actual coalition membership, with two clear cases occurring in Denmark and the Netherlands (Mudde 2019). But more importantly, it is unlikely such arrangements would be caused by similar conditions, as research suggests it is membership in the governing coalition that leads to any legitimizing or taming effects we may see.
This totals 449 cabinets coded over the sample, of which roughly 12% include a far-right party. Given that the quantity of interest is the probability a far-right party enters coalition, I only include elections where these parties won at least one seat in the legislature, leaving 255 observations. Changes in cabinets are identified by the Parliaments and Governments (ParlGov) dataset (Döring and Manow 2019) and defined as any time when either (1) party membership in the governing coalition changes; or (2) the prime minister changes; or (3) any parliamentary election occurs. Table A2 in the online Supplementary Material lists all countries and the respective number of coalitions included. The table also displays the number of governing coalitions that include a far-right party, the average seat share won by far-right parties, and the number of far-right prime ministers who have led the country. In sum, 14 of the included countries have separately experienced the inclusion of a far-right partner in government. Switzerland presents an interesting case as members are appointed to the cabinet not through coalition agreements, but through consensual agreement by the main parties dating back to 1959 (Wolff and Karagök 2012). As it thus is impossible for these same considerations to be taken into account, Swiss cases are dropped from the analyses.
To test Hypothesis 1, I include a measure of GDP growth. Studies have shown that long-term growth is more likely to have an effect on far-right success than the rate of change in one given year (e.g. De Bromhead et al. 2012). In this way, it is likely that mainstream party leaders will be more threatened by a prolonged economic crisis than by a sudden slow-down. I follow these findings and use a threeyear moving average of economic growth. Beyond being more theoretically relevant, this has the added benefit of reducing outlier sensitivity, especially important in an unbalanced panel such as this where observations come every few years, only when a cabinet turns over. In the online Supplementary Material I alternatively test this hypothesis using a three-year moving average of unemployment. The results remain consistent.
For Hypothesis 2, there are several possibilities in terms of operationalization. Namely, the two obvious choices most relevant to Europe are foreign inflows and asylum inflows. Many have noted that recent anti-immigration sentiment is provoked by the ongoing refugee crisis in Europe (e.g. Arzheimer and Berning 2019), with exposure to refugees shown to be a critical driver in support for the far right (Dinas et al. 2019). Thus, I choose to test this hypothesis by utilizing a measure of asylum inflows, as others have done (Swank and Betz 2003), normalized as a percentage of total population. This measure, beyond being most theoretically appropriate and relevant, also suffers least from missing data problems that often arise when utilizing immigration data. Utilizing foreign inflows or foreign population stock instead, for example, leads to a dramatically reduced sample size and unreliable results.
To test Hypothesis 3, I utilize a relevant question in the ESS that asks respondents to rank on a scale of 1 to 10 whether immigration makes the country a worse (0) or a better (10) place to live. Using all nine rounds of the ESS for a time period of 2002-2018, I code any response of 5 or less as 'hostile' and any response above 5 as 'not hostile'. Observations are then collapsed incorporating respondent probability weights to arrive at a country-year level measure of the population's feelings towards immigrants. Because the ESS is conducted biannually in even-number years, I use linear interpolation to estimate values for coalitions that were formed in odd-number years. This leaves 80 coalition observations for the 16-year time period.
Finally, to test Hypothesis 4 I calculate a measure of success of nearby far-right parties. For every year a coalition was formed in a given nation, I calculate the average seat share of far-right parties in the legislatures of countries that share a border with that nation. 1 This represents how successful neighbouring far-right parties are during the election and coalition-formation period of each country. I include the same host of controls in each of these tests. First, it is clear that as the far right rises in popularity, it is likely to win more seats, and that greater seat share in turn has a direct effect on the probability of entering government. Thus, the first control measures the seat share of far-right parties in the election that led to the coalition. If the theoretical expectation is far-right parties will be more likely to enter government when they are more threatening to mainstream parties, this of course must be the case given constant seat share. It is also evident that institutional factors have a large role to play (Müller and Strøm 2003), thus I include a control for whether the country utilizes a parliamentary or presidential system, and a measure of the effective number of parties in parliament to capture the party system (Laakso and Taagepera 1979). I include the country's level of democracy from Polity as a control, as less democratic governance may engender increased support for the far right as well as an increased prevalence of economic crisis or hostility towards foreigners. I also include the country's GDP per capita (logged), and an Eastern Europe regional dummy variable, as important regional differences likely exist.
Finally, as we know from theories on MCWCs (Axelrod 1970;De Swaan 1973), parties who are ideologically closer are more likely to form a coalition. Therefore I include a measure of the difference in left-right ideology between the majority coalition party and the far-right party contesting the election. If more than one radical party contests an election, I utilize the seat share weighted average of their ideological scores. This control is especially important to rule out alternative mechanisms. Specifically, it may not simply be co-optation, but rather ideological convergence driving these results. We are able to rule this out as a mechanism by controlling for said ideological range between parties. Ideology measures for parties come from Holger Döring and Johan Hellström (2013). Each party's ideology is measured on a 0-10 scale, where higher numbers indicate ideology further to the right.
The modelling strategy takes the form of multilevel probit regression, 2 conceptualized as coalitions nested within countries. I include cubic restricted time splines to account for temporal trends in the popularity of the far right.

Results
For ease of presentation, I first present the results of the macro-level study (Hypotheses 1 and 2) and then the results of the remaining studies (Hypotheses 3 and 4). Table 1 presents results from Study 1. Column 1 presents a simple model with no economic or political controls, while column 2 presents the full model. First, we see as expected that a far-right party is much more likely to enter a coalition as its seat share increases. Indeed, the average seat share of a farright party in coalition tops 24%, compared to just 13% when out of government. Similarly, the closer ideologically the far-right party is to the prime minister's party, the more likely it is to form a coalition. In this case, the average ideological distance between the two when they do form a coalition successfully is less than half (1.14) of the distance when a coalition is not formed (2.37).
I begin with evidence for Hypothesis 1, that even as seat share and ideological distance are accounted for, a far-right party is more likely to join the governing coalition as economic conditions deteriorate. This hypothesis is supported by the data. I calculate substantive effects of GDP growth by randomly simulating 1,000 draws of model coefficients based on their variance. I simulate along a GDP growth rate sequence of −6% to 6%, and hold all other variables at their observed values.
Results of the simulation are displayed in Figure 1. Overall, a one standarddeviation change in GDP growth corresponds with a nearly 7 percentage-point change in the probability of a far-right party in coalition. 3 More generally, going from a very strong economy to very poor economic conditions nearly triples the probability of a far-right party entering a coalition, from 13% to 36%.
Hypothesis 2 sees support as well. As the inflow of asylum seekers increases, farright parties are significantly more likely to enter government. Similar to economic growth, a one standard-deviation increase in inflows corresponds to a roughly 6 percentage-point increase in the probability of a far-right party in coalition. Here, a transition from zero asylum seekers to a high rate of 0.5% of population increases their probability of entering government from roughly 15% to over 41%. These results are illustrated in Figure 1.
Importantly, we see these results as the context of the election are held constant. That is, such dramatic increases in the probability of a far-right party's inclusion in coalition transpire even as seat share and ideological distance are controlled for. This implies an important question: are macro-level conditions alone associated with an increase in the probability of far-right entry into government even with very low seat share? I compute a simulation holding seat share constant at a very low 5% and simulate a realistic scenario: the simultaneous increase of asylum inflows and decrease of economic growth, conditions fairly typical of many European countries during the debt crisis. I dub the contextual salience of the farright platform to be low when immigration is low and GDP growth high, and conversely high when immigration is high and GDP growth low. I utilize the 10th and 90th percentiles of GDP growth and immigration for the limits of this simulation. Results are displayed in Figure 2. Overall, the change in contextual salience is associated with an increase in the probability of a far-right party in coalition from 4% to 23% (nearly a sixfold increase). Next, I present results from Studies 2 and 3. The results of Study 2 are presented in column 1 of Table 2. Given the very small sample size of the survey data, I also indicate statistical significance at the p < 0.10 level. Indeed, even the seat share of the party is only significant at such a level. The proportion of citizens with hostile views of immigration is significant at the p < 0.10 level, and a first-difference test indicates that a change from the 25th percentile to the 75th percentile of hostility is significant at the p < 0.05 level. In column 2 I present the results of Study 3, utilizing spatial data. Here too we see significant results: as far-right parties are more successful in a country's neighbourhood, we are more likely to see the far right join that country's ruling coalition.
I simulate the substantive effects from these studies in the same manner as previously described. In terms of public opinion, a nation's electorate shifting opinion on immigration from very welcoming to very hostile has dramatic effects on the probability of a far-right coalition. Going from the 10th percentile (0.48) to the 90th percentile (0.75) roughly triples the probability of their inclusion, from 11% to 37%. We see similar results in terms of far-right proximity. A country surrounded by neighbours with no far-right representation has a relatively low chance of seeing a far-right party enter government, roughly 11%. By the time the surrounding area sees an average seat share of 20% (the 90th percentile), that probability more than doubles to 26% (Figure 3).
These three studies paint a cohesive picture. When conditions lead to the far right imposing a greater threat on mainstream parties, the far right is much more likely to participate in the governing coalition. This holds true in all three of the studies performed in this article: whether looking at the degree to which contextual factors in a nation are ripe for the far right, how closely public opinion  begins to line up with the far-right platform, or even the growth of far-right parties in neighbouring countries. In every case the implication is clear: as conditions change to represent a more threatening environment for mainstream parties vis-à-vis the far right, the probability of inviting them into coalition is significantly higher. Rather than allow the far right to continue to gain ground in opposition and turn voters against them, mainstream parties may be strategically inviting the far right into coalition as a way to co-opt it, and decrease the threat it poses.

Conclusion
These results suggest that as conditions that signify a more electorally threatening atmosphere to mainstream parties by far-right competitors rise, the mainstream may be more likely to form a governing coalition with that competitor. These results build on the work of previous co-optation research (e.g. Heinze 2018;Meguid 2005;Minkenberg 2001; Van Spanje and De Graaf 2018) by demonstrating that coalition formation does indeed seem to be more common in times of potential increased threat from far-right opponents. Indeed, there sometimes appears to be more involved in the calculus of government formation than previously understood. In this article I demonstrate how the level of far-right threat felt by the winner of the election may influence the resulting coalition. There will be costs to forming a government with a far-right party, as history shows. Allies are scared by the potential democratic decline, voters are angered by the betrayal, and the potential for governmental instability increases. And yet, when conditions in the country shift to better foster attacks from the far right, the benefits begin to outweigh these costs. For one, governments send a signal to voters that they take these issues seriously and are interested in working with farright parties to resolve them. Second, when in government the far-right party is no longer able to focus all its time and energy on criticizing the dominant parties and state of national affairs. It no longer gets to sit by and protest, but must attempt to govern. For the far-right party, this offers legitimacy, while for the mainstream it offers insurance. Even as the seat share of the far-right party remains constant, as conditions favouring its platform arise, public opinion shifts so that these criticisms may better take hold, or, as mainstream parties see the far right gain ground nearby, mainstream parties are much more likely to invite the far right into government, in an attempt to quell a growing electoral threat.
There is evidence of this mechanism working. Following the Finnish election of 2015, the Centre Party formed a governing coalition with the far-right Finns Party. Many in the media then characterized this move as a way to attenuate the far right's recent growth in popularity, while some called it an effort to 'tame' it. 'Taming a far-right party' by inviting it into coalition is similarly how the ÖVP-FPÖ partnership following the Austrian election of 1999 has been described, as some noticed the FPÖ's abandonment of many radical proposals, adoption of neoliberal economic policies, and a newfound commitment to democratic values after entering government (Minkenberg 2001).
Substantively, these dynamics have far-reaching effects on government. Coalitions made up of ideologically diverse partners tend not to last nearly as long as ideologically homogenous ones (Axelrod 1970;De Swaan 1973), in part due to the fact that radical parties have a measurable impact in increasing government instability (Powell 1981). Overall, the characteristics that allow them to gain popularity in opposition serve to doom them once in office, and these problems tend to be greatly exacerbated when they serve in coalition with more mainstream partners (Duncan 2010;Heinisch 2003). Interestingly, in the end, mainstream conservative parties have generally been the largest beneficiaries of such processes (Heinisch 2003), often emerging as an option to provide capable government alone with an agenda similar, yet more reasonable, than the far-right party's agenda that had been in such high demand.
Still, given our previous understanding of coalition bargaining, and the long history of the cordon sanitaire, it seems somewhat strange to see so many mainstream parties agreeing to form governments with such radical partners, especially given the high costs associated with such a move, both to the party and nation. In explaining the logic underlying the inclusion of far-right parties in governing coalitions, this study sheds light on this recent trend that has been transpiring across Europe. The rise of far-right parties in governing coalitions suggests that cases such as Finland are not aberrations, but rather indicative of a larger calculus mainstream parties utilize to pick their ruling partners.
No doubt, each of the tests conducted in this analysis carries with it a set of flaws, largely owing to the fact that this abstract concept of 'threat' itself is quite impossible to measure in tangible terms. However, it seems that, when taken together, these separate proxies provide some evidence for the theorized mechanism over rival explanations. Future research would be well served to further conceptualize methods of operationalizing the 'threat' posed by political rivals. Of additional benefit would be conceptualizations of the costs of working with such radical partners, which would allow for more direct measures of the costs vis-à-vis the benefits of such coalition-formation processes.