A comprehensive review of uncertainty in multi-regional input–output modelling

Environmentally extended multi-regional input–output (EE-MRIO) models are now central to consumption-based environmental accounting and to the assessment of global supply-chain impacts. Yet their policy relevance is constrained by a complex and only partly systematised uncertainty structure. This article provides a comprehensive review of uncertainty in EE-MRIO modelling and organises the dispersed literature into six categories: aggregation error; MRIO balancing and assumptions; valuation, price, currency and inflation uncertainty; data source uncertainty; stochastic uncertainty; and inter-model variation. For each category, we synthesise how uncertainty enters the modelling workflow, how it propagates through the Leontief system, and what is known about its quantitative magnitude. The review shows that aggregation choices, data source selection, and assumption-based decisions in balancing and environmental extensions are the dominant drivers of variation, while stochastic measurement error tends to be attenuated by the aggregative structure of input–output models. Inter-model comparison studies confirm that MRIO results are broadly consistent at macro level, but can diverge substantially for specific countries, sectors and supply-chain stages. We argue that capturing the joint effect of multiple, interacting uncertainty sources remains an open research frontier and outline priorities for integrated, multi-dimensional uncertainty analysis. The review concludes with practical recommendations for statistical offices, MRIO developers and users, emphasising the need for better underlying data, more harmonised classifications, and routine quantitative reporting of uncertainty to support cautious but confident use of EE-MRIO indicators in policy.