CeDEx Seminar - Geoffrey Castillo (University of Vienna)

How does the way we represent lotteries affect risk preferences?

When setting up a risk preference elicitation experiment, most of the discussion revolves around choosing the best elicitation method. What usually receives less attention is how to actually describe lotteries to subjects. Different ways of representing lotteries have appeared in the literature and we do not know whether they have an effect on the elicited risk preferences.

In an experiment, we fix the elicitation method and manipulate the way we represent lotteries to our subjects. We move from a baseline, alphanumeric representation of lotteries, to graphical representations that allow us to manipulate which lottery attribute we represent—probabilities, money amounts, or both at the same time.

We find that the representation does not have a statistically significant effect on the certainty equivalents. When we move to structural estimations, however, we find that the representation matters. In particular, we find that graphically representing probabilities and payoffs at the same time reduces risk aversion. The choice of the representation also affects decision noise.