Experimental Seminar [Online]: Filip Fidanoski (UNSW)
The School of Economics invites you to an Experimental seminar by Filip Fidanoski (University of New South Wales).
On the consistency of latent risk preferences across contexts
There are presently more than 100 ways how to elicit risk preferences. Unfortunately, the results of the different elicitation mechanisms [EMs], often differ (and occasionally even contradict each other), making it difficult to identify latent risk preferences. Many EMs have not yet been compared directly to each other. Also, the comparisons that have been made have been afflicted by problematic design and implementation choices, as well as questionable estimation techniques.
Specifically, we are interested in understanding why various EMs, or contexts, provide risk-preference estimates that often are inconsistent across EMs even if the domain (another potential contextual variant) is the same. Robust elicitation of latent risk-preferences remains a challenge for scholars and policymakers alike, given their importance for (the calibration of) many economic models.
Our research complements the literature by providing response on the following: 1) What is the degree of consistency and convergence within, and across revealed EMs [Random Lottery Pair Design, Multiple Price List, Random Compound Lotteries, Dynamic Experimental Estimation of Preferences, Portfolio Choice, Portfolio Choice in Two Stages, Bomb Risk Elicitation Task] and stated EMs [Domain-Specific Risk-Taking and Socio-Economic Panel Study]? 2) Is there a difference between risk-preference estimates obtained through various groups of EMs such as those encompassing static and dynamic decision-making models i.e. does the dynamic consistency exists and how is it related to elicited risk-preference estimates? 3) Is there a logical choice consistency in decision-making under risk and what is the relevance of multiple switching behaviour [MSB]? 4) How does noise specification affect risk-preference estimates and which utility model and parametric specifications fit best our data? 5) What is the difference in the risk attitudes elicited through revealed and stated or hypothetical EMs, i.e. what’s the evidence for hypothetical bias? 6) How do socio-demographic and personal characteristics, state and life events moderate risk-preference estimates across contexts, i.e. what are determinants of conditional context consistency?
Note from Filip: I shall present as part of my Annual Progress Review the motivation and design of our study and would be most grateful for constructive feedback; this (that is before pre-registration) is just the right moment to give it. The project is joint with Andreas Ortmann (UNSW) and Vinayak Dixit (UNSW).