Why do Democracies Experience Antisystem Movements: Statistical Analysis from Interwar Europe (1919-1939)

Why do some democracies experience antisystem movements whereas others do not? Enlisting novel crossnational data from Varieties of Democracy (V-Dem) project and empirically focusing on interwar Europe, this article statistically tests widely-accepted theoretical claims in the literature on political authoritarianism. Logistic regression models applied on 22 interwar European democracies reveal that economic crises are a major cause behind antisystem movement emergence whereas the evidence for attributing causal effect to party institutionalization is not strong enough. Furthermore, the results show that in polarized party systems, the support of center-left/right parties is a crucial factor that leads to antisystem movement emergence in democratic polities.

I make my argument in four parts. First, I define what I mean by antisystem movements and "antisystemness" in general. I also clarify the concept of ideologically bordering party and explain the significance of choosing the interwar period as contextual basis for looking at antisystem movement emergence. The second section goes through the literature on political authoritarianism and puts forward three theoretically-driven, empirically-falsifiable hypotheses. The third section presents the statistical analysis and discusses the findings. The fourth and last section concludes by considering the implications of the findings for present day European politics.

Definitions:
I adopt the concept of "antisystemness" from the work of the distinguished Italian political scientist Giovanni Sartori. Sartori distinguished between two meanings of "antisystemness" in order to avoid what he aptly calls conceptual "stretching" 6 The first meaning refers to "relational" antisystemness and it applies to the political parties operating within a political system, denoting the ideological difference a party has from the others. In this case, such parties would be the extreme left and right parties of the political spectrum. Yet, the first meaning of "antisystemness" applies strictly to political parties and parties do not concern me here. My dependent variable revolves around antisystem movements and not parties. Thus, the second meaning Sartori attaches to "antisystemness" applies to my case here. It is referred to as "ideological" antisystemness and is used not for an analysis of party dynamics but rather for an analysis of democratic regimes. As such, it refers to those political formations that would alter "if it could, not the government, but the system of government … their primary strategy is to make the system collapse" 7 The system of government here stands for two things: (1) the fundamental principles of democracy and (2) the territorial integrity of the state.
As far as the fundamental principles of democracy are concerned, to avoid the conceptual confusion generated by the many proposed definitions of democracy in the literature 8 I opt for a minimalist, procedural definition of democracy. Accordingly, the presence of free and fair elections and the civil/political freedoms that accompany them are what sets apart a democracy from an authoritarian regime. Three criteria distinguish free and fair elections: "ex ante uncertainty (the real possibility that incumbents can lose), ex post irreversibility (the winners assume office), and repeatability."" 9 The second referent of a democratic system is state"s territorial integrity. This is the major finding of two of the leading democratization scholars, Juan Linz and Alfred Stepan, who argue that the presence of a independent, sovereign state with clearly demarcated borders is a prerequisite for democracy. Their words are worth quoting "Without a state no modern democracy is possible … Modern democratic government is inevitably linked to stateness. Without a state, there can be no citizenship; without citizenship, there can be no democracy." 10 Using these two referents of the democratic system, I can proceed to define antisystem movements as those political formations that on the basis of their ideological orientations and goals are either against the fundamental principles of democracy or the territorial integrity of the state, or both. In the interwar As such, an examination of case studies for each of the twenty-two interwar European countries was undertaken in order to identify the ideologically bordering parties of the left and right. 12 Table II shows the results. Coming to the hypothesized linkage between ideologically bordering parties and antisystem movement emergence which I put forward, it seems logical to think of this linkage as a two-way trade: parties providing support to the antisystem movements and/or antisystem movements themselves seeking support from the parties. In this article, I focus on the former only, leaving the latter to subsequent research. Accordingly, this brings forth the question: Why would ideologically bordering parties provide support to antisystem movements? To answer this question, I first have to note that interwar European party systems were polarized ones. "Polarized party systems are those characterized mainly by the presence of relevant antisystem parties and bilateral oppositions, which are mutually exclusive and far apart ideologically." 13 This polarization owes much to the rising influence of communist parties in interwar Europe in the immediate aftermath of the October 1917 Bolshevik Revolution. The rising influence of the communist parties in itself was an important catalyst to the emergence of the extreme right in Europe 14 What this electorally meant was that bordering parties had now to compete with the extremist parties for the votes of the left/right electorate. As Capoccia nicely puts it, "in polarized party systems there is an built-in tendency to "centrifugal competition" in which strong antisystem parties compete in such a way as to force all others, and in particular precisely those bordering them along that space, towards extreme positions." 15 Yet, Capoccia leaves the argument here without further probing into what these "extreme positions" border parties resort to actually mean. To some extent unsurprisingly so, since Capoccia focuses only on strategies within the party system, leaving out the extraparliamentary arena. Trying to fill in this gap, I contend that what "extreme positions" actually means here is giving support to antisystem movements. This support might take the form of financial, armed or rhetorical support (public or private) To conclude, Capoccia still might be useful in remarking that "The very ""systemic propensities"" of party competition thus push the border parties towards the extremes in order to regain the electors that they have lost (from the electoral growth of their antisystem competitors), thereby nurturing the overall polarizing trend." 16 Last, two factors stand out for choosing the interwar period as contextual basis for analyzing antisystem movements, one methodological and the other practical. Methodologically speaking, one can generate valuable findings if the selected cases are subject to common structural characteristics and at the same time present relevant variation on the variables of interest. Interwar period in Europe provides such a setting. The economic, political, and social development of European countries during this time period was clearly delineated by common events. They experienced the two world wars. They all suffered from an external Practically, scholars nowadays are comparing the political ramifications of the 2008 global economic crisis with the 1929 Great Depression. 17 It is widely shared in scholarly circles that the Great Depression in Europe was one of the main causes leading to the rise in influence of both extreme left and right. In the fourth section of this article, I am going to see whether this scholarly consensus holds to the statistical tests or not. Last but not least, the interwar European context increasingly is becoming a scholarly referent given the present rise of the far right in European countries such as Austrisa, Netherlands, Switzerland and so on. 18 Thus, studying what led to the emergence of antisystemness previously in Europe (i.e. during interwar period) may help to extrapolate for present development in Europe.

Literature Review and Hypotheses:
Seymour Martin Lipset in his magisterial Political Man: The Social Bases of Politics maintained that citizens in modern democratic polities and more broadly "each major social stratum has both democratic and extremist political expressions." 19 Accordingly, one could plausibly argue that the social bases of modern mass movements can potentially include any citizen. The empirical puzzle here becomes when will citizens join the ranks of democratic movements and when will they participate in anti-democratic ones?
Lipset is also one of the first to provide an answer to this puzzle. Specifically, a "crisis" or feeling of "displacement" is needed to ignite the authoritarian tendencies in individuals. 20 Although not explicitly stated, what Lipset actually meant by "crisis" or "displacement" were the hard conditions produced by any economic recession. These arguments echo the "relative deprivation" thesis famously put forward by Ted Gurr, who argued that Societal conditions that decrease men"s average value position without decreasing their value expectations similarly increase deprivation, hence the intensity of discontent. The inflexibility of value stocks in a society, short-term deterioration in a group"s condition of life, and limitations of its structural opportunities have such effects. 21 This point is aptly captured by Nancy Bermeo more than four decades after Lipset: "If citizens experience severe material scarcities in new democracies, they don't just get mad, they go mad. They abandon the political center and actively support groups and movements that will destroy the democracy that gave them the freedom to make demands in the first place." 22 Another leading scholar of democracy and authoritarianism, Juan Linz, argues along the same lines. Accordingly, citizens abandon the political arena and participate in anti-democratic movements "as a result of a crisis of legitimacy, efficacy or effectiveness." 23 Linz also focuses on the notion of "crisis" as instigator of authoritarian tendencies, yet his conceptualization of it seems different from Lipset. When talking about a "crisis of legitimacy, efficacy and effectiveness" Linz has in mind a crisis of political institutions in a democratic polity. A leading contemporary institutionalist, John Campbell, reminds us that political institutions "channel and regulate conflict and thus ensure stability in society. Without stable institutions, life becomes chaotic and arduous." 24 Up until now, what emerges out of the literature in question is that citizens" attachment and commitment to democracy is conditional. "In times of crisis they cannot be trusted to resist the allure of authoritarianism unrestrained." 25 Thus, it is commonly agreed in the literature that under these conditions what is needed is strong political institutions that can placate citizens" demands in times of crisis and effectively channel them during other times. Scholars tackling this specific problem focused their attention mostly on strong and institutionalized political parties as the quintessential democratic institution. 26 "Political parties became (and remain) one of the principal means of controlling the less desirable instincts of a suspect citizenry." 27 In what is probably his best academic study, Political Order in Changing Societies, the highly controversial scholar Samuel Huntington forcefully stresses the point that political stability is contingent upon the ability of its political parties to effectively channel citizens" demands and participation in politics. Thus, "A strong Idlir Lika / Why do Democracies Experience Antisystem Movements: Statistical Analysis from Interwar Europe  International Journal of Contemporary Research and Review, Vol. 9, Issue. 03, Page no: SS 20402-20415 doi: https://doi.org/10.15520/ijcrr/2018/9/03/458 Page | 20406 political party system has the capability, first, to expand participation through the system and thus to preempt or to divert anomic or revolutionary political activity." 28 In sum, institutionalized parties constitute the principal (in some cases even the only) means for citizens to access state institutions and express their demands and interests. When citizens are deprived of this right by unresponsive parties, they may question the legitimacy of the system these parties claim to represent and thus turn to other avenues for making their voices heard. Among the main routes applied for are political movements opposing the system. According to Sydney Tarrow, the defining characteristic of such movements is "contentious challenges." What is important for our purposes here is how Tarrow defines the notion "contentious" He clarifies that "Collective action becomes contentious when it is used by people who lack regular access to representative institutions, who act in the name of new or unaccepted claims, and who behave in ways that fundamentally challenge others or authorities." 29 Sheri Berman further explicates that "When national political institutions and structures proved either unwilling or unable to address their citizens" needs," citizens may "turned away from them and found succor and support in the institutions of civil society instead." 30 Anti-system political movements definitely are an institution of civil society, although an unconventional one judged from the premises of the incumbent political actors. Armony highlights the idea that under conditions of weak political institutionalization, political movements may emerge "an alternative to politics and a realm for the circulation of antisystem ideas (the rejection of the democratic regime)" 31 Adopting two structural variables from the above literature, economic scarcity and political party institutionalization, I statistically assess their causal effect on antisystem movement emergence. A quantitative evaluation of such a specific relation is missing in the literature primarily because of a lack of cross-national data on antisystem movements. This has been remedied by the recent data provided by the Varieties of Democracy (V-dem) project.
I complement these two structural variables with a domestic agentic one: the ideologically bordering parties. Lastly, I do not focus solely on the emergence of antisystem movements in itself. What is equally important, I am concerned with the level of threat such movements pose to the incumbent democratic regime. Accordingly, my hypotheses are as follows: H1: Antisystem movements in democratic polities are more likely to emerge when ideologically bordering parties left/right provide support to them.
H2: Antisystem movements in democratic polities are more likely to emerge under conditions of economic scarcity and weak level of political party institutionalization.
H3: Once emerged, antisystem movements pose more threat to the incumbent democratic regime when ideologically bordering parties left/right continue providing support to them.

Statistical Analysis & Interpretation:
My sample of cases consists of 22 interwar European democratic countries identified based on the recently released Lexical Index of Electoral Democracy 32 I employ time-series-cross-section data, consisting of repeated observations on each democratic country throughout interwar years. Detailed explanation on the dependent, independent and control variables, their operationalization and measurement are provided in the appendix. The do-file and replication file are also available.
I use two main estimation models in this research. First, when I look at antisystem movement emergence, given that this variable is dichotomous, I employ logistic regression models. Judging that the logistic regression coefficients that will come out of these models will not mean much practically, I also compute the odds ratios for my independent variables, giving me the odds that my dependent variable will take on a particular value for a one unit increase in the explanatory/independent variable.
Second, when I look at the level of threat these movements posed, given that this variable is categorical, I run ordinal logistic models. Furthermore, I also calculate the odds ratios for my independent variables, revealing the odds that the dependent variable will take on the values of the higher versus lower categories (or vice versa) for a one unit increase in the independent variable. Table 2 presents the results of the logistic regression models clustered around country code, my unit of analysis, to estimate the likelihood of antisystem movement emergence. Odds Ratios in bold *** p<0.01, ** p<0.05, * p<0.1

Table III Likelihood of antisystem movement emergence in interwar European democratic polities
The empirical results are revealing in several respects. First, both of my central explanatory variables, border party support, are statistically significant and preserve the significance throughout all the models. The signs are in the expected direction for both of them. The logistic regression coefficients give us the change in the log odds of the outcome for a one unit increase in the predictor variable. Thus, for a unit increase in border party support of the left and right, the log odds of antisystem movement emergence (versus nonemergence) increase by 1.91 and 2.63 respectively. The R-squares are satifying enough with the fourth model capturing a huge chunk (% 85) of the variation in the DV.
What is more crucial for our purposes here are the odds ratios of the IVs. For every model, I report them under each IV in bold. (Table III) I do not report for model Log 1-4 since they are too inflated. As can be clearly discerned, the odds ratios for party support is above 1 in all the models and they are quite robust in size, which gives strong empirical evidence to prove H1. Take model Log 1-5 for instance. What the odds ratios actually tell us, for a unit increase in the degree of support and/or signal the border party gives to antisystem movements, the odds that an anti-system movement will form/emerge versus will not increases by a factor of 6.7 and 13.9 respectively. Tellingly, in all the models, the impact of the left parties is smaller than the border parties of the right. This means that support from an ideologically bordering party of the right increases the odds that an antisystem movement will emerge more than support from an ideologically bordering party of the left. This constitutes one major finding. And this result is reasonable enough when one reflects upon the political environment in interwar Europe where the spectre of "red subversivism" or "red threat" loomed much larger than any other threat, including the nazi or fascist one. Thus, all in all, H1 is statistically proven to hold.
Coming to what I call the Huntington's thesis, namely that antisystemness is more likely to emerge under conditions of weak party institutionalization, the results seem to be mixed. Two of the proxies I used to capture party institutionalization (party organization and party linkages) produce statistically significant results in most of the models, with signs in the directions expected and odds ratios smaller than 1. Take for instance party organization in model Log 1-3. With an odds ratio smaller than 1 and given the fact that it is a continuous variable, I use the command listcoef to calculate the factor change in odds in party organization. The results generate a b coefficient of -1.43 that in practice means that, for a unit increase in the degree of party organization in an interwar European country, the odds that an antisystem movements will emerge will decrease by a factor of 1.43, very much in line with Huntington"s thesis. Yet, the remaining proxy, party branches, is also significant in one of the models (Log 1-4) defying our expectations. Thus, I conclude that the evidence to support Huntington"s thesis seems to be mixed, not conclusive.
Finally the "Get Mad, Go Mad" scenario seems to hold to empirical scrutiny since the variable depression index (the logged version of it) is statistically significant in both of the models where it is included. Indeed, according to the odds ratio results of model Log 1-5, for a unit increase in the depression index, the odds that an antisystem movement will emerge will increase by a factor of 2,30. Empirically it seems the case to argue that when interwar European citizens experienced economic scarcity, they did not just get mad, they went mad. To conclude, the first part of H2 (i.e. Antisystem movements in democratic polities are more likely to emerge under conditions of economic scarcity) is empirically substantiated. Now I turn to H3. Table 3 presents the results of the ordinal logistic regression models clustered around country code, my unit of analysis, to estimate the likelihood that antisystem movements will pose higher versus lower levels of threat to the incumbent democratic regime. is controlled for in our model, the ideologically bordering party of the left loses its statistical significance. To help us explain this occurrence, let"s take a look at the odds ratios of the relevant IVs for model 2-4 reported in bold in Table IV.

Table IV Likelihood that antisystem movements will pose higher versus lower levels of threat
As the odds ratio show us, the border party of the left does not have any significant effect on the level of threat. Yet, once an antisystem movement emerges, for a unit increase in the degree of support and/or signal the right border party gives, the odds that an antisystem movement will pose higher versus lower levels of threat to the incumbent democratic regime will increase by a factor of 26,3. Let"s not forget that what is special about model 2-4 is that it controls for times of economic scarcity (i.e. depression index) What this may actually point to is that in times of economic hardships (consider for example the aftermath of Great Depression most notably) support from parties of the right may make antisystem movements more dangerous because it is expected that antisystem activity from the left will increase much more under such circumstances.
What about when depression index is removed from the model? Let us take a look at the odds ratios for model 2-5. Here interestingly we see that once emerged, any support from the left border parties makes antisystem movements much more dangerous than any support from parties of the right. This is roughly the opposite of the empirical results we found for our first dependent variable (i.e. the formation/emergence of antisystem movements) In other words, while support from parties of the right makes the emergence of antisystem movements more likely; once antisystem movements have emerged, support from parties of the left makes them more dangerous. And this is the other major empirical finding of the present analysis.
And again let"s not forget that these findings are applicable to the interwar European setting, where, as emphasized above, the spectre of communist threat loomed larger than anything else. Thus, it is against this backdrop that support from parties of the right made more likely the emergence of antisystem movements. Yet, once emerged, excluding times of financial hardships, support from the left made antisystem movements more dangerous.
Among other variables, we see that party organization behaves the opposite of what might be expected, as it increases the likelihood for the movement to pose lowers versus higher levels of threat. Whereas the signs of the variable depression index are at the expected direction and significant. Yet, I have not hypothesized anything about these variables here. I confined myself only to border party support. Thus, to conclude, H3 is statistically proven.

Conclusion:
This article statistically tested widely-accepted theoretical claims in the literature on political authoritarianism. I found clear evidence to support the argument by leading scholars such as Lipset, Gurr, Linz, Bermeo and others that economic difficulties lead masses to join antisystem movements in attempts to overthrow the democratic system, viewed as the embodiment of the socioeconomic order in crisis. The evidence to support Sam Huntington"s thesis was mixed. While weak party institutionalization unable to effectively channel public participation generally led people antisystemic, there were also instances where it did not.
Perhaps the major finding of this research is that, in interwar Europe, ideologically bordering party support seems to have constituted a major cause leading to antisystem movement formation and the level of threat they posed once emerged. While support from parties of the right made the emergence of antisystem movements more likely; once antisystem movements emerged, support from parties of the left made them more dangerous. Thus, the major conclusion is that challenges to democracy in interwar Europe seem to have been both internal and external to the party system. In this respect, I totally agree with Capoccia when arguing that "the vanguard of the antisystem movement may sit in parliament while the rank and file commit violent acts in the streets." 33 Last but not least, a major implication of this research for present European politics is that under conditions of polarized party systems, the behavior and democratic commitment of border parties from both sides of the spectrum is crucial to the viability of the democratic system. Confronted by the concerning rise of the far