The role of services and capital in footprint modelling

Purpose System incompleteness is an outstanding issue in footprint studies, causing systemic truncation errors and misestimation of results. This issue has many implications for analysts, from misleading conclusions in comparative assessments to hampering effective data exchange and comparability between models. A key element of system incompleteness is the treatment of services and capital, which are, respectively, often misrepresented in life cycle assessment (LCA, due to being largely missing in process-based databases) and input–output analysis (IOA, due to being exogenous to the intermediate uses). To gain insight into both the magnitude of such truncation errors and how to mitigate these, this paper analyses the impact of systematically including both services and capital in the system descriptions used in footprint analysis. Methods Manufactured capital is endogenised into the input–output table (IOT) by using capital use information from growth and productivity accounts. Comprehensive service inputs are included in life cycle inventories (LCIs) by means of integrated hybrid LCA. For illustration purposes, the method is applied on two popular LCI and IOT databases—ecoinvent and EXIOBASE—and four common modelling applications of LCA and IOA: LCA- and IOA-based footprints, comparison between IOA and LCA footprints, and a case study using hybrid LCA. Results and discussion The results suggest that the inclusion of both services and capital, either individually or in combination, leads to overall notable differences in footprint results, for example, median relative changes in carbon footprints of 41% and 12%, respectively, for IOA- and LCA-based footprints. Such differences can have notable implications, such as redefining environmental ‘hotspots’ and reversing the results of comparative analyses. Results, however, vary greatly across applications, impact categories and industry/product types, and so specific implications will depend on the research question and scope of analysis. Overall, endogenising capital has a larger impact than including missing services. Conclusions This exercise highlights two fundamental aspects for footprint modelling: the trade-offs between external and internal consistency and the facilitation of model integration. First, the proposed method increases system completeness of LCA (external consistency with the subject of study, namely economic systems) at the expense of internal inconsistencies stemming from ontological discrepancies between input–output and LCI systems (e.g. system completeness). This discrepancy can be mitigated by exploiting the potential of integrated hybrid LCA to create a highly interconnected hybrid system. Second, this approach shows how footprint models can complement each other towards more comprehensive and consistent descriptions of the socio-economic metabolism.