Using cross-impact analysis for probabilistic risk assessment

Ahti Salo, Edoardo Tosoni, Juho Roponen, Derek W. Bunn

Research output: Contribution to journalArticleScientificpeer-review


Cross-impact analysis is widely employed to inform management and policy decisions based on the formulation of scenarios, defined as combinations of outcomes of relevant uncertainty factors. In this paper, we argue that the use of nonprobabilistic variants of cross-impact analysis is problematic in the context of risk assessment where the usual aim is to produce conservative risk estimates which may exceed but are not smaller than the actual risk level. Then, building on the characterization of probabilistic dependencies, we develop an approach to probabilistic cross-impact analysis which (i) admits several kinds of probabilistic statements about the outcomes of relevant uncertainty factors and their dependencies; (ii) maps such statements into constraints on the joint probability distribution over all possible scenarios; (iii) provides support for preserving the consistency of elicited statements; and (iv) uses mathematical optimization to compute lower and upper bounds on the overall risk level. This approach—which is illustrated with an example from the context of nuclear waste repositories—is useful in that it retains the informativeness of cross-impact statements while ensuring that these statements are interpreted within the coherent framework of probability theory.
Original languageEnglish
Number of pages14
JournalFutures and Foresight Science
Publication statusE-pub ahead of print - 20 Sep 2021
MoE publication typeA1 Journal article-refereed


Dive into the research topics of 'Using cross-impact analysis for probabilistic risk assessment'. Together they form a unique fingerprint.

Cite this