Site Loader
Rock Street, San Francisco

Turnout rates tend to show considerable levels of variation over different geographical areas (over countries as well as states and municipalities within these countries). This has led to a vast amount of research effort in explaining these variations. Most of the literature concerning turnout and its determinants is concentrated in the United Kingdom and the United States. In this paper, we will investigate turnout variation among the Flemish municipalities and try to infer whether the same elements influence turnout in the same way under compulsory voting than when voting is not compulsory.

Figure 1 shows the turnout percentages for the 307 of the 308 Flemish municipalities studied for the municipal elections of 1982, 1988, 1994 and 20001. On the horizontal axis are the turnout percentages, while on the vertical axis we represent the number of municipalities with a certain percentage turnout. It can be easily seen that there was a reasonable amount of variation between the turnout rates of the Flemish municipalities in all of the years under investigation – even though voting is compulsory in all municipalities.

Figure 1: Variation in turnout rates in 307 Flemish municipalities Our dependent variable – percentage turnout in the municipality – is measured as the total number of votes cast (valid as well as invalid and blank) divided by the number of registered voters. Following Eagles and Erfle (1989) and Shachar and Nalebuff (1999), we use a logistic transformation of our dependent variable: Logistic Turnout = log (Turnout/(1-Turnout)). This is necessary as the range of turnout is limited to the 0 to 100 percent interval.

It is important to realise that in our analysis we focus on the Flemish municipalities themselves. In other words, the empirical analysis provided in this work is not based on individuals, but on socio-spatial units (municipalities) using aggregate data. This holds as a direct consequence that we do not mean to answer the question ‘why some people are more likely to turn out than others (individual-level analysis)’. Instead, we will try to infer ‘why turnout rates are higher in some municipalities compared to others (aggregate-level analysis)’.

2. Theory The basic positive model of voter turnout is the “expected utility model” of Downs (1957). He argues that the voter will, in deciding whether to vote or abstain, calculate the expected utility from either possible action. The rational individual will thereby vote only if the benefits of doing so outweigh the costs. Using the notation introduced by Riker and Ordeshook (1968, 25), this can be represented as follows R = BP – C + D > 0 (1)

The instrumental benefits (BP) from voting consist of the difference in expected utilities from the policies of the two candidates (B)2. This has to be weighed with the probability (P) that one’s vote will influence the outcome and bring about the victory of the desired candidate. C stands for the different costs of voting. Finally, the D-term refers to Riker and Ordeshook’s (1968) “civic duty” concept. It is the benefit the voter receives from the act of voting itself, from compliance with the ethics of voting.

The central observation of the calculus of voting model is that single votes do not really matter (P very close to 0). When the electorate reaches a certain size, it is unlikely that one single vote will either break or make a tie in favour of ones preferred candidate. The obvious conclusion then is that one should not ‘rationally’ vote to affect the outcome. The discrepancy between this conclusion and the observation of high turnout rates has been labelled “the paradox of not voting”.

One caveat has to be mentioned: Whereas this theory refers to the choice of individuals to vote or abstain, our empirical analysis has reference to a more aggregated level. It is beforehand not obvious whether this model also holds at the municipal level, so ideally we should adapt the theory to better suit our data. However, to the best of our knowledge, there does not exist a likewise model at the aggregate level. Constructing a model ourselves lies beyond the scope of this study. Anyway, all aggregate-level empirical work we know of makes abstract of this problem and assumes the inference to be the same at the aggregate and the individual level (cfr. Filer, 1977; Foster, 1984; Davis, 1991).

3. Empirical Analysis 3.1. Explanatory Variables Concerning the explanatory variables of our model, we look at a number of socio-economic and other variables that are most frequently used in the literature or that can be expected to have a significant influence in our setting.4 This holds that we analyse the effects of Income per capita, % Unemployment, % older than 65, Total population, Mobility of the population, Municipal taxes, Number of pre-1976 municipalities, Electronic voting and Geographical location. The exact “definition” of these variables will be explained below. We will thereby also group the different variables according to the element of equation (1) they refer to. Some variables may refer to several elements in equation (1), which will then be mentioned.

a) Benefits (B) We took up two variables that relate to the benefits an individual has of turning out, the first being taxation levels in the community. Higher taxation means that the municipal government will have more financial resources, which increases the stakes of the election. “Taxation” here consists of two distinct variables referring to the two main elements in the municipal taxation structure: ‘opcentiemen op de onroerende voorheffing’ (Tax_A) and ‘aanvullende personenbelasting’ (Tax_B). Both are expected to be positively correlated with turnout.

Unemployment (Unem) is measured by the number of unemployed in the community divided by the population of working age (18-65). This is also expected to be positively correlated with turnout. The general level of satisfaction with the policy of the current government is likely to be lower in times of economic distress in general and unemployment in particular (Durden and Gaynor, 1987, 234). People have a higher benefit of voting when unemployment is high – in order to protect their employment.

Post Author: admin

Leave a Reply

Your email address will not be published. Required fields are marked *