Intended Actions: Risk Is Conflicting Incentives

摘要

Most methods for risk analysis take the view that risk is a combination of consequence and likelihood. Often, this is translated to an expert elicitation activity where likelihood is interpreted as (qualitative/ subjective) probabilities or rates. However, for cases where there is little data to validate probability or rate claims, this approach breaks down. In our Conflicting Incentives Risk Analysis (CIRA) method, we model risks in terms of conflicting incentives where risk analyst subjective probabilities are traded for stakeholder perceived incentives. The objective of CIRA is to provide an approach in which the input parameters can be audited more easily. The main contribution of this paper is to show how ideas from game theory, economics, psychology, and decision theory can be combined to yield a risk analysis process. In CIRA, risk magnitude is related to the magnitude of changes to perceived utility caused by potential state changes. This setting can be modeled by a one shot game where we investigate the degree of desirability the players perceive potential changes to have.