Dominated choices in Risk and Time Elicitation

Abstract

Many risk and time elicitation designs rely on choice lists that aim to capture a switch point. A choice list for a respondent typically contains two switch point defining choices; the other responses are dominated in the sense that the preferred option could be inferred from the switch point. While these dominant choices may be argued necessary in the data collection process, it is less evident that they should be included at an equal footing with switch point defining choices in the subsequent analysis. We illustrate this using the same data set and model framework as in the seminal paper Andersen et al. (2008). The inclusion of dominated choices has a significant effect on both discount rate and risk aversion estimates. In the case of discount rate estimation, including the near (far) future-dominated choices give higher (lower) discount rates. In the case of risk aversion estimates, including more dominated save option choices tend to give more risk aversion, but the picture is more mixed than in the discount rate case.