Citizens as veto players: climate change policy and the constraints of direct democracy

Abstract

The search for and implementation of effective climate change policies is one of the crucial challenges of policy-makers. One strand of literature argues that domestic factors and in particular institutional prerequisites or veto points strongly influence the quality and pace of a country's policy innovation and adaptation. Focusing on a particular institutional veto point – direct democracy – how does direct democracy influence a country's adaptive capacity in the areas of climate change and, more precisely, what kind of climate change policies have the best chances to be accepted in citizen's direct decision-making? The analyses demonstrate that direct democracy makes it difficult to implement far-reaching, but probably most effective climate change policies, while its direct and indirect impact on the policy-making process rather produces politics of small steps that are supported by a broad political elite.

Keywords:

• climate change policy
• direct democracy
• voting behaviour
• multilevel analysis
• Bayesian estimation
• Switzerland

Notes

1. Given that most countries make little or no use of direct democratic instruments, international comparisons of direct democracy's effects on environmental policy are difficult. Since 1980, out of 45 ballot measures worldwide on this issue at the national level only four did not take place in Switzerland (Centre for Research on Direct Democracy, www.c2d.ch).

2. ‘Public Choice’ is here used quite broadly and mainly as a label for different approaches and arguments applying a rational choice/cost–benefit perspective.

3. The softest means of political steering are persuasive policies that try to achieve behavioural changes by education and information (Windhoff-Héritier 1987, p. 32). In the present context these measures are not relevant because they are normally not subject to decisions on the ballot.

4. As a measure of individual voting behaviour, recall data tend to over-report actual turnout rates. Moreover, we cannot be sure whether individuals reveal their true voting behaviour. While no better, objective data on voting behaviour is available, the comparison between official and reported yes-shares shows that this problem should not question the accuracy of the presented results: in the majority of votes the reported and official results are very close. There are two outliers: the popular initiative ‘Stop the construction of nuclear power plants (moratorium)’ (1990) and the ‘Constitutional amendment for an environmental incentive tax’ (2000), for which we observe a deviation from the official results of around 10%. In both cases however the data still predict the right final result.

5. Further analyses revealed that mandatory and facultative referenda do not differ regarding individual voting behaviour in environmental ballot measures. Similarly, counter proposals to popular initiatives do not substantially differ from initiatives themselves.

6. For more information on the individual determinants of voter decision in environmental ballot proposals the reader is referred to Thalmann (2004).

7. It could be expected that over the years the chances of success for climate change proposals has increased. However, further analyses show that there is no time trend in the acceptance rate of climate change proposals.

8. The variance proportion at the level of ballot measures can only be estimated, for individual level variance depends on the values of the explanatory variables (Goldstein et al. 2002). The different approaches to estimate the proportion of contextual variance lead to the conclusion that the share of contextual variance is around 10%.

9. I refrain from modelling anti-nuclear attitudes as a quadratic term as this model specification leads to severe multicollinearity and convergence problems. Moreover, the introduction of this variable does not lead to different conclusions regarding the other predictors in the model.