Uncertainties in Constructing Environmental Multiregional Input-Output Models

With increasing interest in the modeling of global trade, international supply chains, and multi-scale environmental impacts of global production chains, the method of multiregional input-output (MRIO) modeling is enjoying substantial recent research interest. Models ranging from simple two-region models to expansive world models with more than 100 countries have been constructed. However, relatively little attention has been given to the uncertainties inherent in the method and data typically used for constructing these models. This paper examines three of the greatest uncertainties in the method through a series of models built using input-output data from the United States and several of its largest trading partners (Canada, China, Mexico, Japan, Korea, Germany, and the UK). The three major uncertainties, relating to aggregation and concordance to a common sectoral scheme, treatment of the rest-of-world (ROW) region, and monetary exchange rate issues for factor embodiments, are all shown to be major problems for MRIO models. In fact, while MRIO proponents often claim the method limits uncertainty compared to single-region models, it is likely that these inherent uncertainties often end up raising total uncertainty beyond the levels of a detailed (ie, >200 sector) single-region model, though the relative lack of understanding of underlying uncertainty in standard IO models prevents a strong comparison. Nonetheless, MRIO models offer border-specific emissions inventories and supply