Up close and personal with Mr GDPR
30 May 2018 – 12:20 | No Comment

Government Gazette’s Janani Krishnaswamy caught up with the European Data Protection Supervisor, Giovanni Buttarelli for an exclusive interview to discuss everything relating to the new regulation on personal data protection – the General Data Protection …

Read the full story »

Energy & Environment

Circular Economy

Climate Change


Home » Elections and Governance

Decentralization and Voter Turnout

Submitted by on 25 Nov 2013 – 11:45

By Benny Geys, Vrije Universiteit Brussel & Norwegian Business School BI

The question of why people participate in elections has triggered much debate among both political scientists and economists. A significant share of the academic debate thereby focuses on how a jurisdiction’s institutional setting affects voters’ cost-benefit calculation. Evidently, political institutions – such as compulsory voting, registration systems, quorum rules, concurrent elections and computerized voting – receive most attention. However, other institutional factors also play a role. The institutional design of public good provision is one example. Since this is likely to influence voters’ expected benefits of voting, it might well affect turnout decisions.

To see how this works, it is fruitful to consider how public good provision can be institutionally organized. Although a vast array of different institutional designs are employed throughout the world, most of these reflect three ‘prototypes’ differing with respect to who is elected to decide if, how and how much public goods should be provided. The first type – henceforth ‘centralized provision’ – concentrates all competencies for public good provision at one central level of government, such that the differing geographic reaches of various public goods are simply ignored. That is, all decisions on the provision of different public goods – independent of their geographical reach – are taken and administered under central control. The second type – henceforth ‘federal provision’ – divests all competencies for public good provision in line with the geographical reach of different public goods. As such, there are independently elected parliaments to decide upon the amount of public goods with a broad reach (e.g. roads) and a narrow reach (e.g. child care facilities). Evidently, an extreme interpretation of this view becomes unpractical, since there would be a different elected body responsible for each public good. Finally, the third type – henceforth ‘confederal provision’ – concentrates political authority over public good provision at one level of government, but introduces co-ordination mechanisms for public goods with a broader geographical reach. Hence, this type corresponds with centralized provision in the sense that voters only have one vote (i.e. there is only one directly elected government body), but resembles federal provision by divesting actual decision authority on public goods in line with their geographical reach.

In terms of their electoral consequences, two important effects of these institutional designs can be distinguished. The first relates the benefit of voting, whereas the second is related to the probability of an individual’s decisiveness in the election (which implies him/her affecting the outcome of public good provision). Starting with the former, we have argued above that under confederal and centralized provision, there is only one elected body, such that voters only have one vote to cast. Hence, they have to optimize their expected payoff by choosing the party offering the best policy bundle across all different types of public goods. Under federal provision, however, voters have multiple votes, and can therefore optimize by choosing (the party/parties offering) their preferred policy for each public good. It can be shown that the benefit of voting (and thus the incentive to turn out to vote) under federal provision can therefore never be less than that under confederal or centralized provision.

Turning to the probability of being decisive in the election, the three prototypes of public good provision influence the number of eligible voters deciding upon public good provision. Specifically, under centralized provision, elections are held only at the highest level of government, while under confederal and federal provision elections also (or only) take place at lowers levels of government. Given that the size of the electorate therefore is largest under centralized provision, any individual voter becomes more powerful (and thus perceives an increased incentive to turn out to vote) in federal and confederal settings compared to a centralized setting.

Taking both effects together, one can hypothesize that voter turnout is likely to be highest under federal provision and lowest under fully centralized provision. The intuition for the first hypothesis is that federal provision allows voters to express a distinct preference for a certain set of policies (possibly offered by different parties), rather than have to choose a single party that may implement their preferred policy for some, but not necessarily all, public goods. This increases the benefit of voting under federal provision. The intuition for the second hypothesis is that voters’ probability of being pivotal in bringing their optimal outcome about is lowest under fully centralized provision, which reduces their willingness to incur the cost of voting.

A recent study by Claus Michelsen, Peter Boenisch and Benny Geys has tested these hypotheses using a dataset on local-level elections in Germany, where a substantial variety of institutional designs can be observed in the provision of the same broad set of public goods. Their results show that the institutional design of public good provision has a substantial influence on voter turnout. Indeed, and in line with theoretical expectations, voter turnout is highest under federal provision and lowest under centralized provision. Particularly, controlling for other potentially confounding influences (including population size, density and mobility, election closeness, income, education, unemployment and concurrent elections), a shift of the institutional setting from confederal (the reference group) to federal, is associated with a 2.5% higher turnout rate on average. This supports the idea that voters under federal public good provision perceive a larger net benefit of voting, which is driven by the fact that they can express a distinct preference for a set of policies (rather than a party). Similarly, all other things being equal, a shift of the institutional setting from confederal to centralized, is linked to a 1.9% lower turnout rate on average. This is due to voters’ relatively small probabilities of being pivotal in favor of a desired policy, which reduces their incentive to vote.

Although it should obviously be taken into account that these results are obtained using municipal data (where the mean turnout rate is just under 55%) and have not (yet) been verified at higher levels of government, they do point towards the importance of maintaining a close link between the geographic reach of public goods and the decision authority responsible for their provision. Assigning responsibility to the ‘right’ level of government appears to affect not only the efficiency of public good provision (as often argued by economists), but also on voter turnout.


Further reading

Michelsen, C., P. Boenisch and B. Geys (2013), (De)Centralization and Voter Turnout: Theory and Evidence from German Municipalities, Public Choice, forthcoming.

Geys, B. (2006b). Explaining Voter Turnout: A Review of Aggregate-Level Research. Electoral Studies, 25 (4), 637–663.

Oates, W.E. (1999). An Essay on Fiscal Federalism. Journal of Economic Literature, 37 (3), 1120–1149.