Elsevier

Electoral Studies

Volume 34, June 2014, Pages 261-265
Electoral Studies

A note on electoral competition and turnout in run-off electoral systems: Taking into account both endogeneity and attenuation bias

https://doi.org/10.1016/j.electstud.2013.11.005Get rights and content

Highlights

  • I estimate the causal effect of expected closeness on electoral turnout.

  • I develop an identification strategy that takes into account both endogeneity and attenuation bias.

  • I find that higher expected closeness significantly increases turnout.

  • Not addressing attenuation bias yields estimates that are biased by up to 50%.

Abstract

In this short note, I propose an identification strategy to estimate the causal effect of expected electoral competition on voter turnout in run-off systems taking into account both endogeneity and attenuation bias. I find that electoral competition significantly raises turnout. Not addressing attenuation bias yields estimates that are biased by up to 50%.

Introduction

The standard model of voting participation (Dhillon and Peralta, 2002) regards a single voter's decision to participate in an election as a trade-off between the expected benefits and costs of the participation. While these costs consist for example of the time invested for the electoral act, the expected benefit depends on the expected utility difference between the political alternatives and the expected probability that the single voter's vote is pivotal to change the electoral outcome to his preferred alternative. This expected probability increases in the expected competition in the election (for example the margin of votes between two candidates in the case of a two-candidate election). Therefore, the likelihood of electoral participation should be higher with more competition. The model thus predicts a positive causal relationship between electoral competition and voter turnout.

The contribution of this short note is to measure the causal effect of expected electoral competition on aggregate voter turnout empirically. Measuring this causal relationship empirically is difficult because the estimation problem is subject to two types of biases: First, (expected) electoral competition is endogenous to voter turnout. Second, voters decide on participating at an election not by taking into account the ex-post electoral competition (as this variable is unknown at the time of the decision whether to vote), but by forming ex-ante expectations about the degree of electoral competition. As this expectation is not observable to the econometrician and proxies for electoral competition therefore measure this expectation only with a bias, one has to use appropriate econometric methods to estimate the causal relationship of interest. Otherwise, the estimates are subject to an attenuation bias that drives the slope coefficient of the regression towards zero and thus (regarded in absolute terms) underestimates the coefficient of interest (see Greene, 2012).1 While the existing literature has found convincing solutions to address the endogeneity concern, it has completely ignored the problem of attenuation bias. In this note, I therefore propose an identification strategy that takes both possible types of biases into account.

I find that political competition significantly increases electoral turnout. Moreover, I find that the attenuation bias is much larger than the endogeneity bias. To be precise, not addressing attenuation bias yields estimates that are biased by up to 50% which undermines the importance of accounting for the dynamic nature of the estimation problem.

The remainder is structured as follows. Section 2 presents the empirical strategy and the data set used. Section 3 presents the results. Finally, Section 4 concludes.

Section snippets

Empirical strategy and data

Let E(Ci,t+1|Ii,t) denote the expected competition in an election held in period t + 1 in political unit i. This expectation is formed by voters in the pre-election period t given the available information set at this point in time, Ii,t. If one observed the expectation formed by the voters and if this expectation was exogenous to voter turnout, one could estimate the causal effect of interest directly byTi,t+1=α+βE(Ci,t+1|It)+εi,t+1with Ti,t+1 as the aggregate turnout.2

Results

Table 3 shows the results for valid ballots. In column 1, I include the realized ex-post electoral competition as a proxy for the expected competition. As explained above, this proxy leads both to problems of endogeneity and attenuation bias. In column 2, I use the electoral competition from the first round as a proxy for the competition in the second round, such that the endogeneity bias gets eliminated. Both estimates are negative and significantly different from zero, but the estimate in

Conclusion

This note estimates the causal effect of political competition on electoral turnout taking the dynamic nature of the policy problem into account. I find that ignoring the dynamic nature of the problem, as commonly done in the literature, yields estimates that are biased by up to 50%. The attenuation bias is larger than the endogeneity bias which demonstrates the importance of taking the measurement error in the explanatory variable into account. Overall, I find that increasing political

Acknowledgments

I would like to thank the two anonymous referees for helpful comments and suggestions. Moreover, I would like to thank Ilona Kari for providing excellent research assistance.

References (16)

There are more references available in the full text version of this article.

Cited by (12)

  • Fresh evidence on the relationship between market power and default risk of Indian banks

    2022, Finance Research Letters
    Citation Excerpt :

    However, gnpa becomes weakly significant (at 10 percent level) and insignificant for DD and DC, respectively. The reason for this could be the bias in the level variables' estimates in panel A, owing to their endogenous nature because of simultaneity with the dependent variables (see Garmann, 2014; Wooldridge, 2015). The results of the dynamic model using the system GMM estimator are presented in Panel C.

  • Election frequency, choice fatigue, and voter turnout

    2017, European Journal of Political Economy
    Citation Excerpt :

    Omitted variable bias is thus a serious estimation challenge.7 A prime example is the expected closeness of the election results, from the voters’ viewpoint—something that is, according to the standard calculus of voting (Downs, 1957; Dhillon and Peralta, 2002), an important determinant of turnout, but in general is unobservable to the researcher in real-world elections (Duffy and Tavits, 2008; Garmann, 2014). The purpose of my empirical approach is to filter out such contest-specific but unobservable determinants of turnout.

  • Local candidates and voter mobilization: Evidence from historical two-round elections in Norway

    2017, Electoral Studies
    Citation Excerpt :

    Existing empirical studies on single-round SMD elections have regularly found that turnout is indeed higher in close elections, in part because elite actors tend to increase campaign spending and effort in close elections, which increases voter mobilization (Dawson and Zinser, 1976; Caldeira and Patterson, 1982; Cox and Munger, 1989; Denver et al., 2003). Studies on two-round SMD elections similarly find that the closeness in competition in the first round has a positive effect on voter turnout in the second round (Fauvelle-Aymar and François, 2006; Indridason, 2008; Simonovits, 2012; De Paola and Scoppa, 2014; Garmann, 2014). The second strand in the elite mobilization literature focuses on who is mobilized.

View all citing articles on Scopus
View full text