Market-Average Data Can Misrepresent Supply Chain Environmental Impacts and Their Uncertainty

Abstract Market-average data sets are often used in life cycle assessment and carbon accounting when the production route of a procured good is unknown. The uncertainty this introduces is commonly characterized using generic pedigree matrix factors, which do not reflect how variability across production geographies and technologies differs between goods. Here, we develop a hybrid method that combines pedigree matrix uncertainty characterization within individual production routes with market-share-weighted Monte Carlo sampling between known production routes to represent uncertainty about which route supplied the good. We compare it with conventional market-average uncertainty analysis for 30 economically important goods. For climate impacts, conventional 5th–95th interpercentile intervals are narrower than route-weighted sampling intervals for 25 goods, at least 50% narrower for 14 goods, and up to 82% narrower. Conventional intervals exclude the central climate impact estimates of 44% of possible production routes. Route-weighted sampling captures these outcomes and reveals multimodality and impact-category-specific uncertainty profiles that conventional analysis does not. Variance decomposition distinguishes geographical and technological differences between routes from uncertainty within routes, helping prioritize supplier data collection. These differences show that conventional uncertainty estimates can mislead and may not capture how substantially initial estimates could change when supplier-specific data become available.