![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
This article analyses how the presence of a dominant group of voters within the electorate affects voter turnout. Theoretically, we argue that its absolute size affects turnout via increased free-riding incentives and reduced social pressure to vote within a larger dominant group. Its relative size compared to other groups within the electorate influences turnout through instrumental and expressive responses – in both the dominant and dominated groups – to the degree of electoral competition between groups. Empirical evidence from a large cross section of German municipalities is in line with these theoretical predictions. The observed effects should be taken into account when redesigning electoral jurisdictions through, for instance, municipal mergers or gerrymandering.
Acknowledgment
The authors are grateful to the editor and two anonymous referees for insightful comments and suggestions that significantly improved their manuscript.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. The absolute size of a group does not necessarily equal its power in the political process. Rather, it is the distribution of group sizes – i.e., each group’s size relative to all other groups – which determines each group’s power (Shapley and Shubik Citation1954; Banzhaf Citation1965; Laruelle and Valenciano Citation2001). As such, we will often use the terms ‘relative size’ and ‘(political) power’ interchangeably in the remainder of this article.
2. Voters naturally may also feel part of non-geographic groups, such as those based on socio-economic characteristics. The exact nature of the groups is irrelevant for our theoretical argument, but we discuss a geographic criterion for groups here as this will be central to our empirical analysis. Anecdotal evidence suggests that such geographical ‘community’ considerations can be very strong, and often are connected to distinctive policy preferences (Wille and Deschouwer Citation2012; see also note 4 below). For our empirical analysis, it is sufficient that geographical group considerations have some relevance for individuals – even if they play a minor role compared to socio-economic ones.
3. Note that an increased relative size of a dominated group benefits the closeness of the election.
4. One recent example in Belgium is a 2015 petition against the planned introduction of one single postal code for the municipality of Beringen – almost 40 years after Beringen merged with three other communities in 1976. The main reason for the opposition was that the unified postal code represents ‘an assault on the individuality of our communities’ (Royackers Citation2015; own translation).
5. If municipal amalgamations are in some way connected to factors related to voter turnout in elections, this may induce endogeneity concerns in our analysis. From this perspective, it is important to note that we have no knowledge of any studies in the vast literature on voter turnout nor that on municipal amalgamations suggesting – or empirically verifying – turnout-related drivers of amalgamation processes.
6. For a more detailed description of Germany’s federal system, see, for example, Biehl (Citation1994).
7. In some states, so-called ‘Ortschaftsverfassungen’ allow for some degree of local autonomy. This setting hands back some competences in administering and decision-making on local public good provision to communities. In practice, this opportunity is only rarely applied and limited to a small number of municipal tasks.
8. The overall population and the number of eligible voters in the largest residential area are correlated by nature. Including both variables – as suggested by theoretical considerations – comes at the costs of larger standard errors, which implies that the inferences drawn from our estimation will be less precise. Even so, we find significant coefficients for both variables in all specifications (see below).
9. Unfortunately, ex ante data on election closeness from, for instance, pre-election polls are not available. Also, significant changes in the municipal structure prevent us from using historical election outcomes as a proxy (note that this would in itself be an imperfect proxy given the long time period between consecutive local elections). The ex-post measure of closeness we rely on is, however, potentially endogenous as it may be affected by turnout levels. Still, excluding it from the analysis leaves our key results unaffected.
10. As no data exists for the education level of the entire population, we use data from employment statistics.
11. Consider, for instance, a merger between two municipalities. If both are small, relative group size will be small and voter turnout high (since small pre-merger size would be associated with higher turnout). However, if one municipality is very large and one very small, relative group size will be high and turnout low (since large pre-merger size would be associated with lower turnout). Based on this example, relative group size would be negatively associated with voter turnout due to such selection issues. This implies that our estimates reported below might be biased downwards.
12. In our setting, this unexpected closeness effect may simply reflect that the (ex post) difference between winner and runner-up is a highly imperfect measure of the competitiveness of an election in a multi-party setting. Moreover, the (ex ante) expected closeness of an election may be better captured by the ‘power’ variable in our model (which measures the relative size of different residential areas within the municipality).
13. Excluding our interaction term or running models with either absolute or relative group size (but not the other) indicates that both aspects have an overall negative relation to voter turnout. When adding non-linear terms to the model, we lose statistical significance for the coefficient estimates of relative group size, but overall the key results remain qualitatively unaffected.
Additional information
Funding
Notes on contributors
Peter Bönisch
Peter Bönisch is Senior Research Economist at Zalando SE, Berlin, Germany, and has a doctoral degree in applied econometrics from Martin-Luther-University Halle-Wittenberg. His research interests cover a wide range of topics ranging from applied statistics, political economy, industrial organisation and overcharge estimation to platform economics and online ad auctions.
Benny Geys
Benny Geys is Professor of Economics at the Norwegian Business School (BI) in Oslo, Norway and Research Professor at Vrije Universiteit Brussel (VUB) in Belgium. His research focuses on political accountability, (local) public policy, and civic engagement. His work has appeared in journals including Economic Journal, Journal of the European Economic Association, Journal of Public Economics, Journal of Urban Economics, Organization Studies, Leadership Quarterly, Public Opinion Quarterly, European Journal of Political Research and Public Administration.
Claus Michelsen
Claus Michelsen is Head of Forecasting and Economic Policy at the German Institute for Economic Research (DIW Berlin), having previously worked at the Halle Institute for Economic Research. He has a doctoral degree in economics from Martin-Luther University in Halle-Wittenberg (Germany). His research interests cover topics in applied empirical economics, political economy and urban economics.