A review of methods to trace material flows into final products in dynamic material flow analysis: Comparative application of six methods to the United States and EXIOBASE3 regions, Part 2

Abstract

Modeling pathways toward sustainable production and consumption requires improved spatio‐temporal and material coverage of end‐use product stocks. Momentarily, studies on inflow‐driven, dynamic material flow analysis (dMFA) extrapolate scarce information on material end‐use shares (i.e., ratios that split economy‐wide material consumption to different end‐use products) for single countries and years across longer time periods and global regions. Therefore, in part 1 of this work, we reviewed five methods to derive material end‐use shares which use industry shipment data in physical units and monetary input–output tables (MIOTs). Herein, we comparatively apply these methods to the United States, drawing on detailed national data, as well as the multi‐regional input–output model EXIOBASE3. To better match MIOT and dMFA system definitions, we propose the end‐use transfer method, which re‐routes specific intermediate outputs to final demand in MIOTs. In closing, we conclude on 12 points for improved end‐use shares. We find mixed results regarding the fit between end‐use shares derived from industry shipments and MIOTs: for detailed national data, we find good fit for some materials (e.g., aluminum), while others deviate strongly (e.g., steel). In many cases, the temporal trend of MIOT‐derived end‐use shares roughly agrees with industry shipments. For EXIOBASE3, we find good fit for some countries and materials, but substantial mismatches for others. Despite mixed results, combining MIOT‐based end‐use shares with industry shipments and auxiliary country‐level data could enable improved temporal, geographical, and end‐use resolution. However, the scarcity, documentation, and quality of input data are key limitations for more accurate and detailed end‐use shares. This article met the requirements for a gold‐gold data openness badge described at http://jie.click/badges.

A review of methods to trace material flows into final products in dynamic material flow analysis: From industry shipments in physical units to monetary input–output tables, Part 1

“…Also at high MIOT product resolution, some end‐use product stocks can show large differences to physical accounts (Chen, 2017), which merits the question how reliable the interpretation of results for individual MIOT sectors is. In part 2 of this review, we empirically investigate these issues (Streeck et al., 2022).…”

“…The definition of filter matrices requires expert knowledge and is to some degree up to assumptions (e.g., when is a transaction non‐physical). Specifically, the decision on excluding transactions with service sectors (see compartment in Figure 2), that is, whether to only filter service sector outputs (non‐physical assumption likely) or also service‐sector inputs (non‐physical assumption precarious), can give wrong results, for example, when large material flows to service sectors like repair are ignored (Streeck et al., 2022). Additionally, filter matrix compilation can be tedious and filters are hardly available in published works, which complicates comparing different studies.…”

“…">Third, the MIOT system boundaries present a challenge for tracking material use to particular end‐use products. The functionality of products that are end‐use products in the sense of dMFA, but are intermediate products in the definition of MIOTs (e.g., packaging), is lost during the calculation of the Ghosh/Leontief inverse (e.g., plastic in computer packaging identified as plastic in a computer; see Section 2.1.1 in Streeck et al., 2022). This problem applies to all reviewed methods except for the partial Ghosh‐IO, where end‐uses are defined by the practitioner (see point (1) above).In theory, the correct end‐use can be re‐identified via secondary calculations.…”

“…for individual MIOT sectors is. 10 In part 2 of this review, we empirically investigate these issues (Streeck et al, 2022).…”

“…In Section 3 we conclude on industry shipments versus MIOTs, and on the different MIOT‐based methods, and suggest potential methodological improvements. In part 2 of this work, we apply the five identified methods, including the suggested improvements, to the data‐rich case of the United States, as well as to major regions of a multi‐regional input‐output (MRIO) model (Streeck et al., 2022).…”