Methods for measuring risk-aversion: problems and solutions

Philip J Thomas

Research output: Chapter in Book/Report/Conference proceedingConference Contribution (Conference Proceeding)

1 Citation (Scopus)
228 Downloads (Pure)


Risk-aversion is a fundamental parameter determining how humans act when required to operate in situations of risk. Its general applicability has been discussed in a companion presentation, and this paper examines methods that have been used in the past to measure it and their attendant problems. It needs to be borne in mind that risk-aversion varies with the size of the possible loss, growing strongly as the possible loss becomes comparable with the decision maker's assets. Hence measuring risk-aversion when the potential loss or gain is small will produce values close to the risk-neutral value of zero, irrespective of who the decision maker is. It will also be shown how the generally accepted practice of basing a measurement on the results of a three-term Taylor series will estimate a limiting value, minimum or maximum, rather than the value utilised in the decision. A solution is to match the correct utility function to the results instead.
Original languageEnglish
Title of host publication2013 Joint IMEKO (International Measurement Confederation) TC1-TC7-TC13 Symposium: Measurement Across Physical and Behavioural Sciences
Subtitle of host publication4–6 September 2013, Genoa, Italy
PublisherIOP Publishing
Number of pages6
Publication statusPublished - 2013
Event2013 Joint IMEKO TC1-TC7-TC13 Symposium Measurement across physical and behavioural sciences - Genoa, Italy
Duration: 4 Sept 20136 Sept 2013
Conference number: 13

Publication series

NameJournal of Physics: Conference Series
PublisherIOP Publishing
ISSN (Print)1742-6588
ISSN (Electronic)1742-6596


Conference2013 Joint IMEKO TC1-TC7-TC13 Symposium Measurement across physical and behavioural sciences


Dive into the research topics of 'Methods for measuring risk-aversion: problems and solutions'. Together they form a unique fingerprint.

Cite this