Individual Risk Attitudes and Democratic Representation.

Abstract

Although often ignored, the variation of risk attitudes across individuals may have strong implications for democratic representation. I investigate the effect of risk attitudes on the behavior of voters and representatives. This work is divided into four papers.

The first paper investigates the effect of risk attitudes on the likelihood of voting for a new party in multiparty systems. Using the Swiss case, where two new parties appeared on the political scene, I show that risk aversion decreases the likelihood of voting for a new party. Given that risk aversion is negatively correlated with right ideology, new party entry may be more difficult on the right. Consequently, rightist parties may feel less pressure to adapt to changes in voters preferences than leftist parties.

In the second paper, I argue that existing work presents gaps in empirical evidence of strategic voting driven by the wasted vote logic. Testing if risk attitudes affect strategic voting can bring additional evidence of the wasted vote logic. Analyzing survey data from the UK, Canada, and Germany, I find that risk-averse voters are less likely to vote strategically than risk-acceptant ones. Thus, given the correlation between risk and ideology, strategic voting may be stronger on the left, meaning that representation may be more proportional on the right than on the left.

In the third paper, I use a new dataset that combines a survey measuring risk propensity of Swiss legislators and their votes, and find that risk propensity has an effect on legislative voting. However, as this effect is mediated by ideology, risk does not have particular implications for representation here.

In the fourth paper, I discuss if the heterogeneity of risk preferences should be taken into account in spatial models of parliamentary voting. I suggest three different ways to do so. First, legislators' utility function can have different shapes. Second, one can allow for the status quo bias to be stronger among more risk-averse legislators. Third, risk aversion could affect the predictability of legislators' behavior. The results suggest that relaxing the assumption of similar risk preferences does not appreciably impact the estimation of ideal points.