As mentioned in Section 1
, mass-based material flow indicators are assigned a strong communication function in an information hierarchy as indices or key indicators to draw attention and possibly represent other natural resources [24
]. The more these indicators are aggregated, the more they are subject to a target conflict between their information content and their effectiveness in communication [85
]. This is illustrated by the information pyramid in Figure 4
. By choosing and limiting the analysis to one or a few indicators, other key figures of a system that may also be important are disregarded. In the lower part of the pyramid, material flow indicators are analytically far more interesting for the analysis of the socio-economic metabolism. In contrast to the use of highly aggregated mass-based indicators the concept of the socio-economic metabolism also established as anthropogenic [86
] or industrial metabolism [87
] incorporates MFA-indicators in a systemic way and puts emphasis on the actual transformation of materials [26
]. For complex systems such as national economies, material flow indicators, in their functional aggregation and classification according to different materials and uses, enable considerable insights into the system conditions, changes, and driving forces [89
]. They provide basic structural information about the economy’s material basis. Thus, input indicators linked to output indicators provide important systemic information and control variables for exports and inputs into the environment, stock accounting balances and domestic use, which can also be interpreted by the comparison of different economies and their development. The analytical instruments are essentially based on material flow analyses, accounting and modeling for comprehensive analysis.
3.2.1. Representation Function for Environmental Impacts
In the environmental policy relevant DPSIR model (Driver-Pressure-State-Impact-Response), mass-based material flow indicators represent environmental pressures that can cause undesirable effects and condition changes (Figure 5
). The DPSIR model is suitable for locating material flows in the human–environment set of interactions [91
]. This creates an interdependency between socioeconomic drivers, environmental pressures and impacts, the state of the natural environment and the ecological effects or actual damage. Furthermore, it enables an integrated view of the societal, political, and economic measures that have an impact on the overall set of interactions.
Translated into the model logic of DPSIR, a dematerialization strategy now suggests that fewer raw materials are used when modified technologies or changed demand profiles in the economic system are employed; consequently, not only adverse, degrading changes in the environmental status but above all negative environmental impacts are avoided. As a result, fewer responses and adaptive measures would be required. The approach of dematerializing at all can also be interpreted as a response to an already experienced or expected over-use of natural resources, as suggested by Factor X approaches (see Section 2.3
). If this is supposed to be promising, mass-based indicators would have to correlate with the actual environmental effects, in spite of the lack of effectivity characterization, i.e., they must be ecologically significant. The ecological significance of raw materials extends to the assessment of further use of natural resources through pressures on the environment (land use and conversion, energy demand, fresh water use, and consumption) and, in particular, environmental impacts (use of the environmental media sink function provided as an ecosystem service and impairments of biodiversity).
In the following, five scopes are to be systematically differentiated in order to correlate the raw material use of System S 1 with environmental impacts (Figure 6
). Opposed to Scopes 1, 3, and 5, Scopes 2 and 4 are based on a consideration of the RME; i.e., cumulative raw material requirements up to the actual extraction and thus the physical pre-chains are taken into account. On the other hand, a kilogram of gold and a kilogram of steel would be included in equal measure in a correlation analysis in approaches 1, 3, and 5. However, the actual specific raw material requirement per ton of material (primary raw material intensity) as defined by the indicator CRD (see Section 3.1
) of basic materials and homogeneous semi-finished and finished products increases in the following order, as an initial approximation: construction minerals < fuels < biomass < industrial minerals < basic chemicals and plastics < ferrous and non-ferrous metals < special metals < precious metals [93
]. In detail, however, there is a special rank sequence for each environmental pressure or impact category (e.g., land use, energy use, and greenhouse gas emissions) at the impact level of the respective materials [94
]. With regard to environmental pressures and impacts, Figure 6
shows Scopes 1 and 2 as well as Scopes 3 and 4 forming pairs. The crux is whether material or primary raw material use in System S 1 is only correlated to the impacts in the upstream chains (for example, by the extraction, refining, and casting of a semi-finished metal product outside System S 1, under consideration) or additionally with the effects which are described in the following material use in System S 2 (for example, by a car in which said semi-finished metal product has been installed). The dematerialization logic of the factor X concept, “any input would become an output”, suggests the latter (see Section 2.3
The narrow Scope 5 according to Figure 6
was pursued in a study for the EU 27 that was aimed at identifying threshold indicators in seven environmental action fields [96
]. The extent to which direct abiotic domestic materials consumption (DMCabiot
) can serve as a proxy indicator for national emission limiting values for different air pollutants was investigated, and it was compared with the emissions reported according to the territorial principle of the NEC Directive. However, no linearity could be found which would indicate that DMCabiot
could represent achieving the national emission limiting values. Neither DMCabiot
nor emission data take account of processes abroad. Therefore, they methodically correspond to each other, although they allow only a limited perspective of the overall resource consumption.
Numerous studies have performed regression analyses according to Scopes 1 and 2 in Figure 6
to systematically check the representation function of mass-based and further resource indicators for other natural resources (e.g., representativeness of (raw) material input for land-take or the emission of greenhouse gases). In their study, based on 130 materials and products, Giegrich et al. [97
] concluded that pure mass-based material flow indicators (RMI and TMR) are not suitable as representative resource indicators. A general representation function of individual indicators for the other resource indicators has not been established. This applies neither to other input resources (especially energy, land, and water) nor to the use of the sink function. Other studies that performed similar correlation analyses on up to 100 materials based on their life cycle inventories have arrived at comparable results [98
]. In addition to raw materials, the physical input resources such as land, water, and energy must be dealt with separately in order to map the use of physical resources with sufficient reliability [97
]. If a single environmental impact indicator is supposed to represent environmental impacts, the most appropriate indicator is the primary energy demand according to the results. The latter is also the result of a very complex regression analyses on up to 1200 products and materials. Huijbregts et al. [100
] found that non-renewable energy demand represents many environmental impacts fairly well. Use of non-renewable fossil and nuclear fuels with unambiguous qualitative interactions can be regarded as a key driver for many well-known negative environmental impacts. Conclusions about an inadequate representation function of mass-based indicators apply at a virtual economic system level that contains all materials investigated. A great mix of materials already results in a certain leveling of various environmental characteristics since outliers lose their relevance [99
]. This is valid far less at a micro level of product or process systems, e.g., in the case of paired material substitutions in processes or products, since the heterogeneity of environmental profiles of more similar materials is even more pronounced. An excellent example is the similarity in masses between copper and aluminum: copper meets multiple raw material requirements (in terms of CRD) and acidification potential, while aluminum has high energy demand and global warming potential (GWP)—which can be retraced to both mineralogical and technological factors. When substituting copper with aluminum, a decrease in raw material input and acidification potential comes at the expense of energy demand and greenhouse gas emissions.
In addition, mass-based input indicators do not reflect the environmental burdens or benefit of abatement technologies used within the economy. For example, very specifically mass-intensive materials in terms of raw material intensity may be accompanied by chemico-physical functionalities which can lead to significant resource efficiency gains and additional benefits in their material life cycle compared to technologies lacking these materials. Examples to be mentioned include platinum, palladium, and rhodium catalysts, which can significantly reduce the emission of nitric oxides and volatile hydrocarbons in combustion engines, although they themselves are among the most resource-intensive materials. A selective catalytic reduction system in diesel cars also represents an additional expense during the entire life cycle, but this leads to significant reductions in nitrogen oxides in exhaust gases. Scopes 3 and 4 should be selected according to Figure 6
as the most elaborate approaches in order to take account of the effects in the use phase. However, apart from the availability of data for such analyses, there is a great allocation problem as to which materials of a good we can attribute to the environmental impacts occurring in its further use. A UNEP study for the EU 27 and Turkey took Scope 3 according to Figure 6
using the actual annual material flows in terms of domestic material consumption [102
]. In this case, the consumption phase was taken into account, which generally includes a consuming use of fossil fuels and leads to more intensive environmental pollution profiles. Very different characteristic environmental pollution profiles are obtained for various material categories such as fossil fuels or agricultural products. The respective material flows over different resource consumptions and impact categories do not follow a uniform pattern, which explains the barely existing correlations. This is one explanation that no linearity between material input into the economy and emissions out of the economy could be observed in these studies.
Only if the raw material use were lowered in equal measure for all raw materials, in other words, if a constant distribution of raw material use were assumed, would the overall environmental impact also be consistently reduced. However, this becomes generally invalid as soon as production and consumption patterns lead to deviating raw material input vectors (for example, by a material substitution from bricks to wood materials or simply technology changes regardless of the actual material input of a good). The actual variability of the macroeconomic material use and the cyclical dynamics in individual sectors over time, which can be traced back using environmental-economic accounting data [53
], render this simplistic assumption seem unjustifiable.
Mass-based indicators are not good proxy indicators for relevant environmental impacts according to the studies presented. This makes establishing dematerialization problematic. If mass-based indicators alone are not very representative, the question arises whether they are important in a dashboard of indicators. A set of a small number of indicators can be determined to best describe environmental impacts through a system. This procedure can help reduce effort compared to a comprehensive assessment of all relevant impact categories by eliminating redundancy and thus increase relevance for decisions [104
An equally complex correlation analysis, a so-called principal component analysis of approximately 1000 raw materials, semi-finished, and finished products, plus 135 impact, midpoint, and end point indicators showed that the number of indicators can be reduced substantially while almost never losing significance [104
]. A 92.3% variance between all 135 indicators could be explained by combining only six indicators—GWP, land use, stratospheric ozone depletion, marine ecotoxicity, terrestrial ecotoxicity, acidification, and eutrophication—in a highly representative set. In the case of environmental pressure indicators, 82% of the variance of all indicators can be explained by a combination of four variables: fossil CED, land use, water consumption, and raw material input. Thus, such a set of physical input indicators is not as reliable for use as a set as is a verifiable midpoint (impact) indicator set, but can nonetheless be a notable alternative. The fossil CED already represents 73%, which confirms the above-mentioned study results [97
] as of equal importance to the best possible proxy indicator. As a comparison, the raw material input represents only 54% of the characteristics of all 135 indicators, which underlines its rather poor suitability and the availability of other, more appropriate alternatives. Using one of the proposed indicator sets will be the most robust option at a macroeconomic level. RMI or RMC as the only material flow indicator is of little significance.