Seat enlargements in mixed-member proportional systems: evidence from the German Länder

Abstract

The German mixed-member proportional (MMP) system is considered a role model worldwide. Nevertheless, it has a neglected side-effect: it may produce greatly enlarged parliaments, like the 2017 Bundestag with 111 additional seats. Thus, it is highly relevant to know under which conditions MMP systems lead to such seat enlargements. The article explores this question for the German Länder that have used various MMP versions and seen occasional parliamentary inflations. The analysis demonstrates that a two-stage model of party-system features and institutional factors explain enlargements under MMP systems in 156 Länder elections from 1947 to 2019. Concerning the party system features, enlargements are driven by high seat concentrations in single-member districts and low list-vote shares of the largest party. Institutionally, high ratios of SMD seats and full levelling of surplus seats affect parliamentary oversize. These results have important implications for MMP systems in Germany and other countries.

Keywords:

• Electoral system
• mixed-member proportional system
• surplus seats
• Germany

Acknowledgements

We are grateful for helpful comments from the anonymous reviewers, Jared Sonnicksen, Arndt Leininger, and the participants and audiences of the panel ‘Preferences for Electoral Systems’ at the 2018 ECPR General Conference in Hamburg. We also thank Maren Bestehorn and Atanas Stoilov for their valuable research assistance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Other – in part historical – MMP cases include Albania, Bolivia, Lesotho and Venezuela (Nohlen et al. 2000).

2 This is the common interpretation of the German electoral system that is considered the ‘archetype of a mixed-member proportional system’ (Saalfeld, 2005: 209).

3 The exceptions are Hamburg (multiple-tier PR; Trefs 2008), Bremen and Saarland (both PR with party lists).

4 In (West) Berlin since 1989 and Rhineland-Palatinate, parties may choose between Land lists and regional lists. For our analysis, we classify the (West) Berlin cases as two-level and those in Rhineland-Palatinate as at-large in accordance with the list form predominantly used by the largest parties.

5 For instance, if 6 percent of the votes are cast for parties that do not meet the threshold, the qualified votes taken as allocation basis would be 94 percent of the original vote.

6 Theoretically, the allocation formula also matters in this context as, for instance, the D’Hondt method tends to favour larger parties and may thus slightly reduce the number of surplus seats compared to more proportional formulas. However, this effect is presumably very small in one at-large constituency. Therefore, we do not include the allocation formula in our parsimonious model.

7 De facto only integers are possible. Moreover, not fully proportional formulas like D’Hondt tend to favour larger parties and might slightly reduce the number of surplus seats. However, these distortion effects are presumably negligibly small.

8 Two further cases deviate slightly: Lower Saxony 1951 and 1955. These can be explained by an idiosyncratic procedure for surplus seats at that time (Table A.1 in the online appendix). If the enlargements for those two cases are calculated without this special provision, both deviate less.

9 There are also a few rather homogenous results in West Berlin. The most striking is 1985, when the CDU won nearly all districts in all boroughs (Bezirke). In this case, our at-large approximation explains the enlargement very well.

10 Two Berlin cases are slightly overestimated. This might be a consequence of the idiosyncratic provision of Berlin’s electoral law that the parties may choose if they run with Land lists or regional lists. While the two largest parties employ regional lists – which led us to categorise the MMP system as two-level – the other parties employ Land lists which tends to reduce the number of surplus seats.

11 For example, it might be that a party with one surplus seat would receive the next proportional seat anyway if parliament is to be increased. Then even with a full levelling rule, no levelling seats are assigned. In general, only divisor methods like D’Hondt and Sainte-Laguë establish a clear order of seat distribution while levelling rules for LR Hare have to cope with the well-known paradoxes (Balinski and Young 2001; Pukelsheim 2017).

12 We thereby assume that the largest party receives most surplus seats in relation to its QLV share. If a smaller party also wins surplus seats but less than the largest party, it could nonetheless be the decisive party for the enlargement procedure. On the other hand, as only the maximum enlargement demand matters, we could even err in assuming only one party wins surplus seats as long as other surplus-seat-winning parties are not decisive for the total enlargement.

13 This is rather improbable because as long as the largest party’s QLV share is below 50 percent, every surplus seat beyond the first would trigger levelling seats (see previous section). In Lower Saxony and Saxony-Anhalt, election laws even prescribe that parliament is not to be increased incrementally until the limit is reached, but instead the size resulting from doubling the surplus seats is distributed in one definite step.

14 Another notable outlier is Schleswig-Holstein in 2009 when the limited number of levelling seats was not enough and the remaining surplus seats was above the predicted number. This caused the unusual situation under MMP systems that the parties forming the government had gained less votes than the opposition parties, which led to a judgement by the state constitutional court requiring early elections (Linhart and Schoen 2010).

15 This approximation estimates the surplus seats of every party in every regional constituency, aggregates them over all regions (for each party separately), calculates the enlargement demand by every party and takes the maximum of those.

16 The SMD seat shares vary largely following Cube Rule functions (Manow 2011) and are therefore hardly suitable for choosing few scenarios.

17 See, for instance, the joint bill by the Liberals (FDP), the Left, and the Greens for a change of the federal electoral law from 6 November 2019 (Deutscher Bundestag, Drs. 19/14672). It proposes to decrease the number of SMDs from 299 to 250 and increase the regular Bundestag size from 598 to 630, which lowers the SMD ratio from 50 to 40 percent.

Additional information

Notes on contributors

Philipp Weinmann

Philipp Weinmann is a Postdoctoral Lecturer for Political Science Research Methods at the Helmut Schmidt University, Hamburg. His research focuses on electoral systems, direct democracy, and quantitative methods. [[email protected]]

Florian Grotz

Florian Grotz is Professor of Comparative Government at the Helmut Schmidt University, Hamburg. His research interests include political institutions and party governments in European democracies. His work has been published in journals such as Administration & Society, Democratization, European Political Science Review and World Politics. [[email protected]]