Escalation instability in probability-impact matrices: evidence from capital projects

Purpose Probability-impact matrices are widely embedded in project governance systems and used to prioritise risks, allocate mitigation resources and trigger management escalation. This study examines whether commonly used qualitative risk matrices preserve relative risk prioritisation when compared to quantitative assessments of the same risks. Design/methodology/approach The study analyses 1,144 paired qualitative and quantitative risk assessments from 91 mining capital projects across 43 organisations. Paired nonparametric tests, distributional comparisons, Wasserstein distance metrics and colour-band reclassification analysis are used to evaluate differences in probability ratings, impact ratings and composite risk scores. Findings Qualitative and quantitative probability ratings align at the ordinal level. In contrast, qualitative impact ratings and composite risk scores are systematically higher than their quantitative counterparts. Quantitative reassessment altered escalation colour bands in 73.3% of risks, with 59.4% moving to lower bands and 37.2% shifting by two or more categories, indicating material changes in prioritisation. These findings suggest that qualitative matrices do not reliably preserve escalation stability under alternative elicitation formats. Practical implications When escalation outcomes depend on representational format, risk-informed governance becomes structurally unstable. Hybrid configurations that separate qualitative communication from quantitative decision support can enhance decision quality and improve the robustness of prioritisation and contingency allocation. Originality/value The study introduces escalation stability as a practical criterion for evaluating risk assessment tools and provides large-sample empirical evidence that elicitation format materially influences risk prioritisation in capital projects.