Modeling of Risk Perception in Engineering Decision Analysis

Authors

Michael H. Faber and Marc A. Maes

Abstract

Within the framework of risk based engineering decision making one is often
faced with the problem of integrating risk perception into the formulation of the utility function.
Often it is felt that the standard linear utility function is inappropriate as it fails to express the
increasing undesirability of large consequences. In practical risk assessment various proposals
for non-linear utility functions have been suggested and applied. But the question remains
which type of non-linear utility function would be appropriate in a given situation and whether
it would lead to a rational decision.
The basic premise of this paper is that the risk aversion described by non-linear utility functions
can almost always be explained by the non-inclusion of certain "follow-up" consequences.
"Follow-up" consequences are, generally speaking, triggered by extreme losses, such as excessive
business losses, loss of reputation or other indirect or so-called intangible losses. The noninclusion
of such losses occurs either voluntarily or involuntarily. Although in principle the use
of appropriate non-linear utility functions and the inclusion of "follow-up" consequences are
mathematically equivalent and may lead to identical decisions, only the latter approach leads to
risk-consistent rational decision making. An illustrative example is given in the paper, highlighting
the significance of the findings.

Published in/by

Proceedings 11th IFIP WG7.5 Working Conference on Reliability and Optimization of Structural Systems, eds. M.A. Maes and L. Huyse, Banff, Canada, November 2-5, 2003.