![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
Abstract
Recent elections in Europe have shown that a context of increasing citizen distrust towards democratic institutions may lead to very high levels of electoral volatility and to the emergence of new parties. On the other hand, institutional reforms are sometimes presented as a solution to citizens’ discontent with political institutions. Focusing on a specific type of political institution ‒ electoral systems ‒ the question addressed in this study is whether high levels of electoral volatility may trigger electoral reforms. The article investigates the conditions under which reforms affecting the electoral system’s degree of openness to new parties were enacted in 25 European countries between 1945 and 2012. The findings demonstrate that volatility due to the emergence of new parties is the most powerful explanation to account for the introduction of electoral reforms, particularly those that hinder the entry of new parties into the system.
Acknowledgements
Replication data is available at http://www.electoralsystemchanges.eu/. Previous versions of this article have been presented at the ECPR Joint Sessions of Workshops, Salamanca 2014; the EPSA Annual Conference, Edinburgh 2014; and the ECPR General Conference in Glasgow, 2014. We thank the participants of these panels for their valuable feedback.
Notes
1. For a good overview of the literature, see Rahat (Citation2011).
2. Data availability limitations on several countries made it impossible to include in the analysis all of the countries included in the ESCE dataset.
3. The variables reflecting those reforms that open and those that close the system only capture reforms having a clear direction. The variable reflecting the overall number of reforms is not equal to the sum of the other two dependent variables because the former includes cases of reforms with no clear direction of their effects (see explanation in the next page).
4. The classification of the formulas has followed standard conventions: majoritarian formulas lie at one end of the continuum from plurality, majority to run-off systems. On the other side of the continuum, we find proportional systems: there is some consensus on the view that Hare, Sainte-Laguë and Droop formulas are the most proportional and that Imperiali (largest remainders), d’Hondt HA and Imperiali (highest averages) are the least (Benoit Citation2000; Lijphart Citation1994).
5. This approach has the advantage of not assigning any arbitrary threshold to qualify a party as new. On the contrary, Powell and Tucker (Citation2014), who proposed one of the most complete accounts of the calculation of within- and extra-system volatility, considered a party to be in the system when it had achieved at least 2 per cent of the vote share, which may be problematic when considering this variable as an explanatory factor for electoral reform.
6. The data was downloaded from Carey data archive (http://sites.dartmouth.edu/jcarey/). Note that the district magnitude data for Sweden is modified according to information gathered in the framework of the ESCE project.
7. The use of regime duration offers more detailed information than regional dummies because it accounts both for the geographical area (Central and Eastern European democracies are much younger) and for the degree of system institutionalisation, which has been said to decrease the likelihood of reform. Long-lasting democracies tend to have more settled structures that are more difficult to change (e.g. see the impact of path dependence mechanisms on institutional change).
8. We have opted for random effects models because we assume that variation across countries is random and uncorrelated with the independent variables. Appropriate time-invariant variables, such as those belonging to Central and Eastern Europe, have been included to control for a possible omitted variable bias (see Appendix). Hausman tests have been run for each model to check whether random effects models give consistent results compared to fixed effects models. Results confirm the suitability of random effect models.
9. These analyses are included in the Appendix (Tables A1, A2 and A3).
10. There is a high correlation between new-party volatility and the Pedersen index (p = 0.79, sig = 0.000). A similar problem occurs with other volatility components: between within-system volatility and the Pedersen index the correlation is very high (p = 0.77, sig = 0.000). Consequently, the Pedersen index cannot be included in the same model as the other two sub-components of this index.
11. New-party volatility and within-system volatility are significantly correlated (p = 0.23, sig = 0.000), but to a much lesser extent than in the case of the Pedersen index (see note 9). We have therefore re-run the analysis testing the impact of these two types of volatility jointly. The results of this analysis are shown in the Appendix (Tables A2 and A3).
12. In order to test for the robustness of our findings we have performed additional analyses with alternative explanations of electoral reform, namely the existence of economic crises, veto players (number of parties in government), citizens’ dissatisfaction with democracy, populist parties’ share of votes and ideological composition of governments. None of these alternative explanations has a significant effect, but the inclusion of these variables in the models does not produce different results for the volatility measures to the ones reported here. Data and results of these analyses are available from the authors upon request.
13. For clarity we do not report the results for the variable accounting for within-system volatility. In agreement with the results shown in Table 2, this variable does not have any significant effects in the models for which the dependent variable accounted for more inclusive or restrictive reforms (Tables 4 and 5).
14. We have run additional models in which the volatility due to the entry of new parties is treated as an ordinal variable. We do not find strong evidence of the existence of a linear effect: only higher levels of volatility have a positive, though not statistically significant effect.
15. We have reproduced the analyses considering new-party volatility and within-system volatility in single models. The direction and the size of the effects are consistent with the results shown here.
16. According to Model 5 in Table 5, new-party volatility has an odds ratio of 1.052 (sig: 0.011; standard error: 0.02). The Pedersen index displays in Model 4 in Table 5 an odds ratio of 1.037 (sig: 0.016; standard error: 0.015).