First, we would like to thank Mihail Chiru for the generous and thorough analysis of our book that complements well the reflections of other published reviews (e.g. Van der Eijk 2021; Vittori 2021; Puleo 2022; Yovcheva 2022; Rodríguez-Teruel 2022). We particularly appreciate that Chiru proposes new venues for research. While we are not (yet) able to solve all the issues raised by our discussant, below we make an attempt at responding to his most important critical remarks.
A first group of comments/suggestions refer to the determinants of party system closure, the central phenomenon of our book. Building on Rubabshi-Shitrit and Hasson’s (2022) recent work on the positive impact of the constructive vote of no confidence on cabinet durability, Chiru raises the question of whether “this very high threshold for cabinet removal” (Chiru 2022) may contribute to high levels of closure. Then he partly answers this question by pointing out that only half [1] of the six European countries with such constitutional options can be considered to be closed. We also think that the relationship is either non-existent or very weak. None of the inchoate party systems (that is, party systems whose closure score is 85.9 or smaller) had such provision, but none of the over-institutionalized party systems (≥96) had it either. The non-European cases with constructive vote of non-confidence suggest a weak or non-existent relationship too. For example, Lesotho, especially before 2012 (when they had a dominant party system, cf. Pelizzo and Nwokora 2016) had high closure, while Israel is characterized by partial alternations and rather open access, indicators of relatively low closure.[2]
[1] The Belgian party system still remains open, at least according to our closure threshold of 91 percent.
[2] In the case of Israel, one may just think of Bennett-Lapid’s government composed of parties in the previous government (Blue and White, Yamina, the Labor Party), first time partners (Yesh Atid, Yisrael Beiteinu, Meretz) and totally new governing parties (New Hope and United Arab List).
Nevertheless, Chiru is right that the relationship should be, theoretically, positive. A similar argument could be made for fixed-terms of legislatures, as well as the responsibility of the cabinet to the president in both presidential (e.g. Brazil, Cyprus) and president-premier semi-presidential (e.g. Weimar Republic, Russia) regimes. The linkage between these institutional solutions and the predictability of party systems requires further investigation. This leads us to the question on the effects of institutional design on party system closure. We do not find much direct effect, but perhaps considerable indirect impact exists. As a matter of fact, and building on both Duverger’s (1954) and Lijphart’s (1994) seminal work, it is possible to think of “[an indirect causal chain] going from the electoral system through party system format to institutionalization” (Casal Bértoa 2019: 580). And the same can be said regarding (semi-)presidential regimes, whose negative effects on two of the main determinants of party system openness (i.e. high fragmentation and low party institutionalization) have been consistently demonstrated (see Casal Bértoa 2021).
A third point within this first cluster of comments refers to the role economic crises play in fostering the closure of unstable party systems and in the opening of institutionalized ones. While an answer to this very interesting question would require a larger analysis, using Casal Bértoa & Weber’s (2019) definition of crisis and simply looking at figures 3.2 to 3.6 as well as 4.3 to 4.5 of our book, the impression is that the 2008 Great Recession did not have much impact on the levels of closure, not even in those cases where democracy later collapsed (e.g. Erdogan’s Turkey and Yanukovych’s Ukraine). In some instances, like Andorra, post-colonels Greece, Italy, Slovenia, post-Francoist Spain, Bulgaria, Czechia and, to a lesser extent, Moldova, economic downturns may have played some role in destabilizing party politics. The results are also mixed concerning the 1929 Great Depression (see figures 3.2 and 4.2-4.5).
A second cluster of comments refers to the consequences of closure, either for democracy or for the institutionalization of party systems in countries that manage to return to the democratic path after its collapse. Regarding the former, Chiru questions if the rather robust results showing the positive impact of party system closure (chapter 11) would hold if other variables (e.g. years since transition, democratic wave, exposure to economic recession or war) were controlled for. A proper analysis of this issue would require running complex models, but a simple look at Table 1 below, which presents data for all those four variables in all 22 European democracies that collapsed, suggests that the impact of the first two variables was probably limited: democracies collapsed at varying levels of experience (consider pre-WWI Greece vs the Yugoslav Kingdom), originating in various waves (e.g. French 2nd Republic, Austrian First Republic, post-WWII Greece, Erdogan’s Turkey, post-communist Russia). There are also examples of cases that collapsed without being exposed to economic recession (e.g. French Fourth Republic, Restoration Spain). The linkage may be strongest to wars. Discussing specific cases (e.g. the collapse of the Hellenic democratic monarchy after the National Schism in 1915, see page 247), in the book we also point to the destabilizing role of certain wars.
Table 1. Collapsed democracies by length of democratic regime, democratic wave, exposure to economic recession and exposure to war (PDF)
Along similar lines, Chiru directs the attention to the link between survival of democracy and conditions that are accidental, not related to the degree of democratic consolidation. While testing this would require not only a robust statistical model but also conceptual clarification (i.e. what exactly we mean by crisis or by consolidated democracy), the support given to anti-establishment parties could give us a first approach to the solution. In the book we take the percentage of votes going to parties that oppose the political establishment as the indicator of polarization, but it can also be interpreted as the indicator of democratic consolidation. Dividing the cases into two groups, depending on whether the support for such parties was above or below 15%, and considering the latter group as the consolidated one, we can see some relationship: in the “consolidated” group 29%, in the “non-consolidated” group 41% collapsed. But the link is indeed not very strong.
Chiru also wonders about the educative effects collapses might have on elites. The descriptive accounts included in chapters 3 and 4 suggest a mixed picture. Thus, while political leaders seem to have learned the lesson in countries like post-WWII Germany, post-Salazar Portugal or post-Francoist Spain, to name just a few, that was not the case, for example, in the Spanish Second Republic, or even in the French Fourth Republic. In these two cases political elites just repeated some of the same mistakes (e.g. type of head of state, electoral system design, ideological polarization, party personalization) that had led to the failure of previous democratic experiences.
Elites, as suggested by Chiru, often responded to challenges by implementing constitutional reforms. But these reforms typically failed to increase the level of party system closure, consider the cases of inter-war Greece (1927 Constitution), post-1960 coup Turkey (1961 Constitution), France (1946 Constitution) and Spain (1931 Constitution). In some cases, an increase in the closure scores was possible without major changes of the constitution, take Austria, Finland, San Marino and, to a lesser extent, Latvia.
Chiru also questions the use of anti-political-establishment voter support as a proxy to capture polarisation. Some of the other reviews raised the same point. In the book (p. 193) we acknowledge the limitations of our proxy. We point out the lack of data for the alternative indices, especially for the pre-1945 period but also for the micro-states. We would also like to remind the readers that the number of indicators used to measure polarization is immense and has changed considerably over the years (Enyedi & Casal Bértoa, forthcoming 2022), no consensual standard exists. Still, as we also point out in the book (p. 194), correlation between our index and Dalton’s (2008) frequently used measure is rather high (0.7).
A fourth group of comments have a methodological character, addressing the use of aggregate level data. The bulk of the analyses indeed rely on aggregate closure scores that have the close-to-impossible task of summarizing the nature of the party systems across many decades. We try to increase the robustness of our conclusions by some analyses based on individual years as well as on within case-studies. In Mölder, Enyedi & Casal Bértoa (2022), the latest spin-off of the party system closure project, we focus not on the overall closure index but on its three components. There we also provide an answer to Chiru’s reasonable doubts about the negative impact of party system closure on the quality of democracy once economic development is controlled for.
Not least, Chiru also presents some intriguing recommendations for using party system closure in future research to examine how it impacts policy change or the adoption of non-partisan cabinet types. On the former our current knowledge is very limited. While a congruent degree of stability at both levels appears to be intuitively more likely, it is also possible that in bipolar closed systems government changes lead to particularly radical change in policies. Such a configuration would amount to a trade-off between personnel change (accountability) and policy change (representation). The second suggestion also merits further investigation, especially given the negative correlation (-.48, significant at 0.01 level) at the aggregate level between party system closure and the percentage of non-partisan ministers. A new version of the dataset, focused on governments and ministers, rather than party systems, was recently presented in Casal Bértoa & Rodríguez-Teruel (2022).
Finally, we recognise that the party system typology advanced in the book has been under-utilized. We use this typology to grade party system types according to their level of closure (pp.123-124) and to show how closure trajectories and party systems are related (p. 125). But indeed, more could be done, especially using the within-country periodization of party system types (tables 3.1 and 4.1 in the book). Based on these tables one could identify instances of “systemic change” in a Sartorian (1976) / Mairian (1997) way (i.e. moves from one particular category to the other), and then one could search for the causes and the consequences of these changes.
Working with so many countries and with so many years often feels like riding a dragon. Even though we started the project more than seven years ago, we feel we barely scratched the surface of the topic of party system closure. Many sub-topics are waiting to be explored (not just in Europe but in other – less researched – parts of the world). One first attempt can be found, for instance, in Casal Bértoa and Lee (2021). To conclude with yet another metaphor, the harvest is plentiful, we just need more laborers.
