ABSTRACT
In practice, environmental impact assessment and emissions accounting are generally conducted at the rm level, project level, or product level, where the rst hand data is available. Since the early 1990s, input-output analysis (IOA) has been widely used for environmental accounting, such as carbon footprint assessment and the calculations of embodied emissions and virtual water at the sectoral level for a nation, a region, or multiple regions, depending on the purpose of the research. One of the advantages of using IOA is that not only direct emissions, but also indirect emissions from various upstream processes in a supply chain can be taken into account systematically. Every empirical input-output (IO) table rests on the denition of industries, which is usually given by aggregating similar products into one sector. For each dened industry, rm-level data is collected and the input coecients in an IO table can represent an average of the production functions of many dierent sample rms. erefore, the usefulness of published IO tables for practical environmental assessment is challenged by the robustness of the results, which can be inuenced by dierent classications of rms, products, or processes into sectors (Gibbons et al., 1982; Wiedmann, 2009a).
