Final Report Summary - DESIRE (DEvelopment of a System of Indicators for a Resource efficient Europe (DESIRE))
Executive Summary:
DESIRE is a FP7 project that has developed and applied an optimal set of indicators to monitor European progress towards resource-efficiency. The project ran from September 2012 to February 2016. We proposed a combination of time series of environmentally extended input output data (EE IO) and the DPSIR framework to construct the indicator set. This approach has used a single data set that allows for consistent construction of resource efficiency indicators capturing the EU, country, sector and product group level, and the production and consumption perspective including impacts outside the EU. The project has:
• Improved data availability, particularly by creating EE IO time series and now-casted data
• Improved calculation methods for indicators that currently still lack scientific robustness, most notably in the field of biodiversity/ecosystem services and critical materials. We further have developed novel reference indicators for economic success.
• Explicitly addressed the problem of indicator proliferation and limits in available data that have a ‘statistical stamp’. Via scientific analysis we select the smallest set of indicators giving mutually independent information, and show which shortcuts in (statistical) data inventory can be made without significant loss of quality.
The project comprised further Interactive policy analysis, indicator concept development via ‘brokerage’ activities, Management, and Conclusions and implementation including alignment and hand over of data and indicators to the EU’s Group of Four of EEA, Eurostat, DG ENV and DG JRC.
Partners are:
1. The Netherlands Organisation for Applied Scientific Research (TNO), Delft, Netherlands
2. Wuppertal Institute (WI), Wuppertal, Germany
3. Alpen Adria University - Institute of Social Ecology (UNI-KLU), Vienna, Austria
4. Institute of Environmental Sciences, University of Leiden (UL-CML), Leiden, Netherlands
5. SERI - Sustainable Europe Research Institute, Vienna, Austria
6. Norwegian University of Technology (NTNU), Trondheim, Norway
7. Radboud University (RU), Nijmegen, Netherlands
8. Fundação da Faculdade de Ciências da Universidade de Lisboa (FFCUL), Lisbon, Portugal
9. Vienna University of Economics and Business (WU), Vienna, Austria
10. 2.0 LCA Consultants (2.-0 LCA), Aalborg, Denmark
11. Martin-Luther-Universitaet Halle-Wittenberg (MLU), Halle, Germany
For more information contact the co-ordinator at: arnold.tukker@tno.nl
Project Context and Objectives:
In the past fifty years humans have consumed more goods and services than in all previous generations put together. Various sources estimate that in a business as usual scenario global population and wealth growth until 2050 may lead to an annual resource extraction that is 2 to 4 times higher as in 2000. This is a trajectory that could ‘crash the economy against the Earth’.
The Europe 2020 Strategy establishes resource efficiency as one of its fundamental Flagship Initiatives for ensuring the smart, sustainable and inclusive growth of Europe. The Resource Efficiency Flagship should "help the EU to prosper in a low-carbon, resource constrained world while preventing environmental degradation, biodiversity loss and unsustainable use of resources". In this, a broad definitions of resources is used beyond just e.g. material extraction, including for instance human health and ecosystem health that can be impacted by emissions from economic processes. The Resource Efficiency Flagship has close connections to the Raw Materials Initiative and Innovation Union Flagship. Without access to and smart use of critical materials, European industry will face considerable challenges. Smart resource management and eco-innovation are essential to realise the innovation and competitiveness goals of the Innovation Union Flagship. Resource Efficiency has further relations with other EC policy initiatives, such as the EU Sustainable Development Strategy and the Sustainable Consumption and Production Action Plan. To support this effort, the EU FP7 Environment theme has called for an indicator project in the field of resource efficiency.
This challenge of addressing multiple resource constraints implies that the EU policy agenda in the next decade(s) needs an excellent insight in how resources are used in society, and particularly how access to and efficient use can be fostered. Several key aspects need to be considered:
• A consistent framework of resource indicators should be made available. Indicators for resource efficiency of production processes should be consistent with indicators for the resource efficiency of consumption patterns and these again should be consistent with how the development of resource efficiency is tracked on the national, EU and global level. Such a framework allows one to link consumption to production and to evaluate the contribution of innovations to resource efficiency.
• Material resources, which are preserved in use and recoverable, afterwards must be accounted for taking a consistent life cycle approach. Resource flows must be followed. Indicators should be based on modelling of resource extraction, renewable resources, inputs, flows, within and across industrial sectors, and the final outputs.
• The indicator framework needs to link resource use to the ultimate physical and economic impacts of resource scarcity, ecosystem service or human health and should be developed in a way that it can accommodate new knowledge on mechanisms of action. Interlinkages between different pressures on the state of resources and ecosystem services need to be adequately considered, including environmental impacts of emissions and the like.
• Indicators should look deeper than just the national level. The role of global trade in enhancing resource utilization needs to be adequately considered. An adequate sector breakdown should be conducted, distinguishing in particular the public and the private sector through the use of the national accounts classification (non-financial corporations, financial corporations, general government, households, non-profit institutions serving households and the rest of the world) and also allowing further disaggregation to sections (e.g. manufacturing) and subsections (e.g. manufacture of transport equipment).
• Long time series and now-casting should be considered.
• The reference for calculating resource efficiency (now often economic parameters such as GDP, or population) should be refined to better reflect the benefit of resource use.
The DESIRE project (i.e. Development of a System of Indicators or a Resource efficient Europe) addressed these key aspects. The main goal of DESIRE was to develop and apply an optimal set of indicators and models to analyse and monitor the European progress towards resource-efficiency from both a production as well a consumption perspective. Our key objectives were:
a) to build the indicator system in close connection with the most relevant stakeholders, including DG ENV, DG JRC, EEA and Eurostat, building upon their previous work5.
b) to ensure consistency in the indicator framework. As indicated above this implies consistency between production- and consumption perspectives, national, EU and global level, and industry and product group level. It implies however also aligning various resource-related indicators in different existing indicator frameworks (see Annex 1).
c) to improve data availability, particularly by developing methods that can provide time series and now-casting and create time series and now casted data in the project.
d) to improve calculation methods for indicators that currently still lack scientific robustness, most notably in the field of biodiversity, ecosystem services and critical materials.
e) to explicitly address the problem of indicator proliferation, and demands for indicators that need data that even on the medium term realistically cannot be expected to be supplied by statistical sources. Our aim is to define via scientific analysis the most relevant set of indicators giving mutually independent information, and to show which shortcuts in (statistical) data inventory can be made without significant loss of quality
f) by all of the above, to provide EU policy makers an indicator set that has the highest relevance, is fully accepted, and can be handed over to and implemented by Eurostat, EEA and others with an acceptable effort. This will ensure optimal support to the mentioned policy agenda’s and contribute to a truly sustainable use and management of natural resources (in Europe and worldwide)
A review of the most relevant policy dossiers that should be supported by the DESIRE project tells us that already extensive indicator systems have been developed or are being proposed. Three points stand out from this analysis:
1. There seems almost an overload of indicators related to the different policy agenda’s, also in the field of resources. There seems room for improvement of structure and relations;
2. Most indicator systems seem to select which operational indicators to use rather pragmatically on the basis of data availability;
3. Resource efficiency indicators usually are defined at national level, usually comparing the amount of resources used and the benefits in terms of economic value added or gross domestic product.
DESIRE goes beyond this simple and well-known approach and includes the development of indicators for the impact of resource use on the resource base on one hand and for an extension in measuring the benefit of resource use in terms of alternate measures of welfare on the other hand. In addition, a consistent calculation methodology, based in the framework of national economic accounts, is established and applied to take comprehensively relate resource use to aspects of wellbeing. By this, DESIRE will also ensure that the indicator concept is consistently applicable at the macro, meso and micro level. We propose an indicator framework organised along four dimensions:
1. Use an extended description of the economic system based on UN SEEA. At the core is an extended description of economic processes of resource extraction, transformation, residual production and value added. This description is based on the UN System of Economic and Environmental Accounts (SEEA) and its European implementation (NAMEA). Economic processes of production, consumption and investment are described in monetary terms in input-output tables (IOTs), while satellite accounts track physical processes of resource extraction, land use and emissions by sector. We propose to link these national accounts adding trade information to obtain a picture of the global resource implications of national decisions. We furthermore propose to extend the framework to address the life cycle of small mass flows of valuable ‘critical’ material resources in the economy with tracking their stock and flow.
2. Use the DPSIR concept to relate indicators along the cause-effect chain. The Resource Efficiency Flagship adopts a “wide” definition of resources, including environmental health and ecosystems. The relevant resources considered in DESIRE are ecosystems and their contribution to ecosystem services and biomass production, addressing land, soil quality, water and biodiversity as important preconditions for a healthy state of ecosystems; metals and other valuable minerals; a stable climate, and other environmental conditions conducive for human health. DESIRE adopts the Driving Forces-Pressures-State-Impact-Response (DPSIR) concept to relate different indicators along the chain of cause and effect, and that relates pressures caused by resource use to impacts on the natural resource base (which includes impacts on the environment in broad sense)9. Knowledge of impact mechanisms and ecosystem service provision will be utilized to systematically relate different indicators and select among them. The DPSIR chain also provides guidance for approaches how different indicators can be aggregated, ultimately to single score indicators.
3. Explore alternative reference values for success of the economic system. To assess resource-efficiency, it can be considered to use alternatives for economic references such as value added and Gross Domestic Product. Candidates that may provide more meaningful measures of social objectives, such as prosperity and well-being, include the Human Development Index, perceived happiness, or economic references corrected for external costs etc. At sector and final consumption level one could consider e.g. performance based measures (e.g. for education, health care and elderly care).
4. Use concepts like the Policy Cycle, RACER and scientific cluster/structural analysis to select the minimal needed and most coherent indicator set. There is a need to avoid having too many indicators or indicators that need difficultly available data, whereas at the same time policy needs to be well informed. It hence makes sense to use the Policy Cycle as completeness check for the indicator system by posing the question: does the indicator system – as far as reasonably can be expected - provide the relevant information for different steps in the policy cycle? At the same time, concepts such as the RACER criteria (‘relevant, acceptable, credible, easy and robust’) and scientific cluster and structural analysis can help to identify which indicators are less relevant.
Building the aforementioned, ideal indicator structure for resource efficiency faces a set of challenges. The most important ones are:
1. There is a large variety of indicator systems and suggested list of indicators that relate to the resource efficiency agenda. This needs alignment. Further, quite some indicators (e.g. GDP/DMC, emission data, impact assessment indicators such as GWP) are provided for countries as a whole (rather than industry sectors or final consumption categories) or describe a state of or impact on the environment without that always a clear causal link to economic activities/drivers and pressures is made (e.g. biodiversity indicators). Content wise we propose using the 3 structuring elements: a common data base in EE IO format; linking cause and impact via DPSIR; including alternative reference values for success or ‘outcomes’ of economic activity. We saw however also a need for activities with an important process component to achieve this:
a. Policy-science brokerage
b. Policy analysis
c. (Joint) elaboration of an initial, credible and acceptable indicator framework
2. For a variety of indicators, there is a reasonable agreement how to calculate them, but data availability is still a main problem. This is particularly relevant for institutes like EEA and Eurostat. They are expected to provide authoritative information, which implies considerable quality demands with regard to the data they can use and publish. Creating data sets of ‘statistical quality’ often is difficult. We want to address this point as follows in our project:
a. Gathering high quality data in the project itself (e.g. for time series and nowcasting)
b. Where this is not possible, creating a research database (building upon projects such as EXIOPOL and CREEA) that allows for an analysis which shortcuts in data inventory could be considered without significant quality loss.
c. Performing a scientific correlation analysis and structural analysis to single out the minimum set of indicators for impacts, pressures and drivers with most ‘saying power’, which again can reduce the effort in data inventory.
3. For other indicators there is still a need to improve the scientific basis for their calculation, particularly if we want impact indicators to reflect a causal link with economic activities and drivers at sector or product level. This is particularly true for indicators that assess biodiversity impacts and impacts on ecosystem services, and to a lesser extent indicators related to the analysis of criticality of small resource flows.
4. Finally, uncertainty is a point to consider. Whereas this project has as primary mission to build an indicator system for resource efficiency and not uncertainty analysis per se, some level of attention needs to be given to this issue.
DESIRE created progress beyond the state of the art in the following main areas:
1. Time series and now casting of data, particularly related to a global Multi regional Energy/Physical/Monetary input-output model
2. Indicators for security of supply of critical materials.
3. Indicators for impacts on biodiversity and ecosystem services
4. Novel reference indicators (‘Beyond GDP and value added’)
5. Insight in the optimal ‘resource-efficiency indicator dashboard’ to define the best mix of policy options, and to analyse the progress towards resource-efficiency over time.
Project Results:
1 DESIRE indicator framework
1.1 DESIRE’s conceptual framework
The DESIRE project aims to develop an optimal indicator set to monitor progress towards Resource Efficiency in Europe, yet taking into account the relevant global perspective. Resource efficiency is about using natural resources efficiently, either in a technical sense (i.e. less physical input per physical output) or economic/welfare sense (i.e. economic or societal value generated per unit of resource). The latter implies a need to discern all possible interactions between society and the natural system in a coherent and integrated way. The concept of societal or industrial metabolism is helpful in this regard.
This concept refers to the notion that socio-economic systems require resources (materials, energy, water or land) as input in order to produce goods and services or to maintain socio-economic structures. In addition, production and consumption processes, as well as transportation, put a burden on the environment through their (metabolic) outputs such as wastes and emissions to air, water and soil. “Resources” thus address different categories and issues, all with different impacts to “environment”, i.e. climate, biodiversity, ecosystems, health, etc. Resource [use] efficiency indicators therefore need to address complex interactions between society and the environment in order to empower political action; to set meaningful targets; and to adequately monitor the [global] use of resources.
Figure 1.1: Society-nature interactions as metabolism
DESIRE captures these metabolic relations between the natural environment and society, in a framework of Multi-Regional Environmentally Extended [economic] Input-Output relations with production and consumption flows. This [MR EE-IO] framework is referred to as “EXIOBASE”.
With this framework it is then possible to assess how production and consumption impacts the natural system. The DESIRE indicator framework builds on the causal Driver-Pressure-State-Impact-Response (DPSIR) frame, adopted by the European Environmental Agency (EEA), to understand society-nature interactions. This framework helps to structure and organise indicators along a cause and effect chain. That is: drivers for resource use; that put pressure on the environment; resulting in environmental impacts; and causes changes in the state of the natural system. These insights can eventually be a trigger for responses from relevant actors. Responses that feedback on drivers and thus closes the cycle (see figure 1.2). In the DESIRE project, the DPSIR-framework is integrated with the metabolic perspective of society-nature interactions (see figure 1.3)
Figure 1.2: The DPSIR framework
Source: EEA
Drivers
The structure and characteristics of the socio-economic system, its economic processing, and household consumption patterns are considered driving forces (drivers), which are strongly shaped by the cultural, political, and economic context they are embedded in.
Pressures
Resource use and management activities put pressure on and potentially change the natural system, its ecosystems and ecosystem services and thus the underlying natural State.
Impact
Effects of pressures on the natural system are considered environmental impacts. These impacts can, for example, form a threat on human health, human well-being (i.e. a broad concept of welfare) or economic wealth. Environmental impacts could be interpreted or weighted against a certain amount or quality of natural capital stocks, e.g. planetary boundaries.
Responses
Responses are the decisions and choices made within the socio-economic system by individuals or by policy makers as a response to changes in the societal as well as natural systems with the aim to adapt to these. Examples are: tax regulations, legislation or other (thematic) policy response packages.
Figure 1.3: Conceptual framework for DESIRE’s indicator set on resource use
Source: Eisenmenger et al., 2014
The conceptual interactions between socio-economic activities and the natural environment as depicted in figure 1.3 can be integrated in an environmentally extended input-output framework (see figure 1.4).
Input-output (IO) tables originate from economic accounting. Together with its two main building blocks – supply- and use tables (SUTs) – IO-tables form the backbone of the system of National Accounts. Typically, an input-output table includes inter-industry flows in monetary units as well as flows between industries and the final demand categories (e.g. households, government spending, capital investment, exports). By that, IO tables allow for tracing goods from the extraction process through manufacturing and down to final demand and provide information on the inputs needed by an industry to provide the respective industry output. One could say, it supplies the “ingredients” for one unit of output of the manufacturing industries, either derived directly or also indirectly from other industries. From an IO table, the main economic output indicator, Gross Domestic Product (GDP) can be derived. National statistical offices from all around the world provide SUTs and IO-tables, typically at an interval of a few years.
Figure 1.4: Integration of DESIRE’s conceptual frame in a MR EE-IO framework
Source: Eisenmenger et al., 2014
The standard economic input- and output flows can be complemented with environmental extensions such as material extraction, land use and emissions to air, soil or water (i.e. environmental pressure indicators). The environmental or natural resource inputs enter the production process of a certain sector and are than further distributed via inter-sectoral deliveries until they end up in one of the final demand categories. Thus, the environmental extensions represent the resource use indicators and data, i.e. pressure indicators in absolute values. Figure 1.5 shows these relations in a multi-regional set-up.
Figure 1.5: A Multi-Regional Environmentally Extended Input-Output model
Source: Eisenmenger et al., 2014
1.2 Complementing and improving EXIOBASE
The indicator framework of DESIRE builds on previous versions of EXIOBASE that were developed with support from the EU’s Sixth and Seventh Framework Programmes, i.e. the FP6-project EXIOPOL (that delivered EXIOBASE version 1) and FP7-project CREEA (that delivered EXIOBASE version 2).
EXIOBASE version 1 comprised an extensive economic-environmental database that followed the principles of the System of National Accounts [SNA 1993] and System of Environmental-Economic Accounting (SEEA 2003), based on the International Standard Industrial Classification of All Economic Activities (ISIC and its European equivalent NACE ) and Statistical Classification of Products by Activity in the European Economic Community (CPA). EXIOBASE version 1 provides data for base year 2000.
Accordingly, in the CREEA project, the database and indicator framework was refined according to the new accounting approaches of SNA 2008 and those that were proposed for inclusion in SEEA 2012 regarding four priority areas: water, waste and materials, forestry, and climate change issues.
Version 2 of EXIOBASE provided data at an at an unprecedented level of detail in terms of sectors, products, emissions and resources with coverage of 43 countries; 27 EU countries and the largest non-EU economies (summing up to 95% of the global GDP) with over 150 smaller countries combined in 5 ‘Rest of the World‘ groups by continent. EXIOBASE version 2 provides data for base year 2007.
Moving-on from here, DESIRE complemented and improved the database with new statistical information and expanded the indicator framework by adding the 28th EU Member State, Croatia, and by compiling time series, including now casting of recent years for which not all data is available yet from official statistical publications. With these improvements EXIOBASE version 3 can now be used to analyse consumption and production based accounts as well as stressors embodied in imports and exports, including developments over time.
EXIOBASE version 3 has the following characteristics:
EXIOBASE 3
Base-years 1995 – 2011/16 *)
Products 200
Industries 163
Countries 44 (28 EU member plus 16 major economies)
Rest of the world regions 5 (Europe, Asia, Africa, America, Middle East)
Water accounts 194 (Water blue and green per source, including final demand)
Material accounts 189 (Energy products, including final demand)
222 (Used extractions)
222 (Unused extractions)
Land accounts 14 (Including build up land for final demand)
Social accounts 14 (Employment per skill level and gender; vulnerable employment)
Emissions 28 (from combustion including final demand)
410 (non-combustion)
3 (HFC, PFC, SF6)
*) Historic time series for up to 2011, the rest of the years has been now-casted.
Source: Stadler et al., 2016
1.3 DESIRE’S Resource Efficiency indicator framework
DESIRE’s indicator framework differentiates three main categories of indicators: resource use, resource efficiency and environmental impacts (the columns in table 1.1 below). In addition the framework comprises six flow types of resource inputs or (metabolic) outputs (as indicated by the rows in table 1.1). In accordance, figure 1.7 shows how these Resource Efficiency indicators are positioned in the multi-regional EE-IO model.
Table 1.1: Environmental indicators in the DESIRE indicator framework
Figure 1.7: Positioning of Resource Efficiency indicators in the EE-IO model
Source: Eisenmenger et al., 2014
Pressure indicators, related to direct society-nature interactions, i.e. resource use indicators in absolute physical terms, are a good starting point to measuring resource efficiency. These are input flows of materials, energy, water and land to intermediate use and final use in the IO-framework, as well as output flows thereof of waste and emissions.
Accordingly, resource use has to be complemented by indicators that capture the effects on the natural system (impact indicators) as well as effects on the socio-economic system. The latter are commonly referred to as resource efficiency indicators. In DESIRE’s indicator framework two types of resource efficiency indicators are distinguished: resource efficiency indicators in relation to the economy and resource efficiency indicators in relation to services to society.
Resource efficiency indicators in relation to the economy (production and consumption based) can be measured against GDP or value added, and hence directly be linked to economic transactions in the IO-framework. Efficiency indicators in relation to services to society, on the other hand, often need (detailed) auxiliary data from sources outside the IO-framework. Services to society concern functional outputs rather than economic outputs, such as adequate housing, space heating, nutrition, etc. By their nature these outputs have a ‘beyond GDP’ connotation and can sometimes best be expressed on a rather micro level. An example of such a resource efficiency indicator is the required energy consumption for space heating per m2 of dwelling floor space. In the DESIRE project a work package was devoted to test opportunities for using novel reference indicators ‘beyond GDP and value added’.
The third main indicator category in the DESIRE framework, environmental impacts of resource use, put socio-economic pressures in relation to the natural state, and inform about both quantitative and qualitative aspects of natural capital stocks. They can inform about the state of stocks and changes thereof over time, e.g. depletion, degradation, climate change, biodiversity loss. For these type of indicators a satellite account of natural capital stocks need to be connected to the IO-framework. This is the lower block of data indicated in figure 1.7.
Table 1.2 summarises the rationale behind the indicator types and their position in the DPSIR-framework. It covers the pressure indicators in absolute values (columns in yellow) as well as the “resource efficiency” indicators (columns in yellow as well) which result from relating resource use to macro-economic added value (e.g. GDP) or macro-economic well-being indicators. The columns on the left cover the socio-economic system. Additionally to the resource efficiency indicators that link to socio-economic macro indicators, resource efficiency can be analysed as the relation between resource use and specific societal services provided (column in red). This covers all the activities that directly deal with biophysical flows, however no longer structured along the macro-economic IO matrix but along societal services. The socio-political responses (column in pink) cover the social, political, economic, or cultural responses.
Table 1.2: DESIRE indicator types and their position in the DPSIR-frame
Source: Eisenmenger et al., 2014
The green columns on the right side cover the efficiency of resource use in relation to the environmental impacts on the natural system in two dimensions, quantitatively and qualitatively. These environmental impacts are structured along the commonly used environmental threats (boxes on the very right). Thus, the environmental impacts do not follow the IO structure, just as the socio-economic activities at macro level.
The general idea behind DESIRE’s indicator framework is to apply a 2-level system: a limited set of headline indicators that covers resource use, resource efficiency and environmental impacts on the macro (i.e. country) level, and an accompanying second level of indicators addressing specific questions within each category.
1.4 A set of readily available Resource Efficiency indicators
EXIOBASE contains the physical layers energy, water, materials and land, which can be tracked as resource inputs to the economic production process. In addition there are various material extensions that provide information on metabolic outputs of production and consumption processes, such as emissions and waste. EXIOBASE covers:
• Greenhouse gas emissions, in kilograms of CO2, CH4, N2O;
• Polluting emissions: SOx, NOx, NH3, CO, Benzenes, Indeno (1,2,3-cd) pyrene, PAHs, PCBs, PCDD_F, HCB, VOCs, PM10, PM2.5 TSP, As, Cd, Cr, Cu, Hg, Ni, Pb, Se, Zn, SF6, HFCs, PFCs);
• Nitrogen and phosphorous emissions to water;
• Domestic material extraction of various types of crops, wood, metal ores, industrial and construction minerals & fossil fuels (differentiated in used and unused extracted materials);
• Withdrawal of blue water, differentiated by the manufacturing, electricity production and domestic use sector;
• Green and blue water consumption, differentiated by use category, for various types of agriculture, livestock, manufacturing, electricity production and domestic consumption;
• and land use (by different types of arable land, pastures and forests).
Of the full set of possible indicators that are directly calculable from EXIOBASE, some might score differently on criteria that are relevant for their uptake and implementation in policy making or for monitoring purposes, i.e. on so-called RACER-criteria: Relevance, Acceptability, Credibility, Easiness and Robustness. The proposed indicators in the 2-level indicator system are therefore tested along the lines of these RACER-criteria. The results of this exercise are shown in Annex 1, indicating the relevance of indicators and possible needs for further development. Table 1.3 (below) provides a sample of indicators that can directly be calculated.
From IO-tables coefficients of industry requirements in monetary terms can be calculated (i.e. direct requirement coefficients and Leontief multipliers). Similarly, these factors can be calculated for environmental impacts at two levels of detail:
Scope 1: Direct environmental interventions
Direct environmental interventions (for example, air emissions or material extraction) are available for each industry without further calculations from the table of environmental extensions (i.e. the dark-green boxes as shown earlier in figure 1.5 and the “resource use part” in figure 1.7). One can sum the extensions for all industries in the region, plus the interventions reported under final use (= use phase interventions), to derive the country total. Scope 1 is a territorial perspective, expressing environmental consequences, which originate within a country’s or region’s territory.
Table 1.3: Examples of environmental indicators calculable with EXIOBASE
Scope 2: Total (direct plus indirect) environmental interventions
The main advantage of IO-models applied to environmental issues is that they allow calculating the total direct plus indirect effects for all products and all sectors, also those with very complex supply chains, as the whole economic system is included in the calculation system. IO-analysis thus avoids so-called “truncation errors” often occurring in coefficient-based approaches, i.e. errors resulting from the fact that the whole complexity of production chains cannot be fully analysed based on Life Cycle Assessment (LCA) approaches, where as a consequence certain up-stream chains have to be “cut off”. IO-analysis thus avoids imprecise definition of system boundaries, which is one key advantage over other approaches. IO-models also avoid double counting, as different supply chains are clearly distinguished from each other in the monetary input-output tables. Thus, a specific resource input can only be allocated once to final consumption, as the supply and use chains are completely represented.
However, IO-analysis also contains some disadvantages. Whereas LCA-type approaches are able to cover both upstream and downstream environmental effects, IO accounts only for upstream inputs to the production processes and ultimately to final consumption. Environmental consequences from the use-phase are only given in a single table entry, at the intersection of the final use column and the environmental extension row (see the blue square in the lower right corner of figure 1.5). For CO2 emission for example, this single number includes many types of direct emissions like those from private car use or the emissions related to heating our homes and drinking our soft drinks. Hence, typical use-phase oriented indicators, such as the “per capita CO2 emissions from the housing and infrastructure sector” (i.e. part of the EEA core set of indicators) are difficult to derive from IO tables directly. In order to calculate these indicators, the vector of private consumption would need to be split up by consumption categories, for example following the COICOP classification, which disaggregates consumption by purpose (e.g. food, housing, transport, communication, etc.).
It shall be emphasised that EXIOBASE is different from other MRIO databases (such as GTAP or EORA) because it contains physical layers at the industry level. Therefore, EXIOBASE contains data on direct physical imports and exports, which allow calculating material flow-based indicators, such as DMC, which would not be possible to calculate with other MRIO systems without physical layers. It also contains detailed waste data, which allows calculating specific indicators, such as recycling rates.
1.4.1 Results of resource efficiency indicators over time
For a full set of resource efficiency indicator results over time and details on underlying data and documentation of all readily available indicators we refer to the report of Stadler et al, 2016 (Desire deliverable D9.1).
We would like to stress that working with DESIRE’s MR EE-IO framework and indicators involves a large amount of data, such that some programming skills are required to analyse the detailed datasets. For the purpose of easing these analyses an open source tool, Pymrio, has been developed. This tool has been used to calculate production and consumption based accounts for each stressor per product and per country.
These results can for example be used to assess decoupling of economic growth from environmental pressures. Doing such an analysis for the EU 27 countries shows that, taking a production based account, the European Union appears to have achieved decoupling (see figue 1.8 top). Taking, however, the consumption based perspective (figure 1.8 botom), material usage and fuel combustion grow in the same rate as the economy. In addition, CO2 emissions, water use and land use are only recently decoupled from economic growth.
Figure 1.8: Time series of production- (top) and consumption based accounts (bottom) of the European Union.
Indicators show the resource efficiency as impact per capita.
Source: Stadler et al., 2016 (calculations based on EXIOBASE v3.2.3)
2 Indicator development: novel indicators
DESIRE aimed to push the state of the art for EE MRIO compilation and calculations. For that purpose, besides developing approaches for now-casting EE MRIOs (see Stadler et al., 2015), DESIRE’s research efforts also focussed on more experimental indicator development. This included efforts to develop or improve resource efficiency indicators in the domains: Critical Materials, Biodiversity and ecosystem services, and Novel reference indicators ‘beyond GDP’, as well as efforts to linking them to the MR EE-IO framework. In general, linking pressures to state and impacts in these fields proved to be complex sometimes, as it involved manifold or non-linear relations. This made determining causal relations between society-nature interactions and final environmental impacts to the least ‘challenging’. In this chapter we report the key results per each of the three, what we call ‘novel indicator’ domains, and explain the extent to which these results could be linked to the EE-IO framework.
2.1 Critical material indicators
The objectives of DESIRE’s work package on critical material indicators (WP6) were to specify and define a number of relevant critical material indicators and to link these to the IO-framework. Regarding critical material indicators, economy wide aggregated indicators as proposed in DESIRE’s ‘core’ indicator framework, are not suitable as most of the energy, water, land, carbon and other emissions indicators do not refer to the criticality problem of materials. In addition, the indicators on waste and materials in DESIRE’s indicator framework are on a still too aggregated level to signal criticality problems of material substances as fraction of products.
After the first stage of DESIRE’s research activities on critical materials a list of indicators were considered relevant.
It was suggested to pursue an approach based on material flow assessment, linked with input output (IO) tables. Furthermore, it was proposed to develop a link between IO- tables and elaborate waste statistics, which are highly relevant for quantifying recycling options. It was concluded that a hybrid mixed unit (i.e. combined physical- and monetery information) model is in principle suitable for modelling the flows of critical materials in a MR-IO framework, and to determining flow-based resource efficiency indicators. However, at the same time it was acknowledged that availability of data could be problematic to accurately build such a hybrid IO-model. Research efforts in DESIRE therefore also focussed on possibilities to overcome data challenges.
There are numerous challenges involved with connecting material flows and critcality assessments to economic accounts. Some of the most practical challenges are:
• The detailed industrial production, Europroms statistics (PRODCOM), do not distinguish between primary and secondary commodities (or the items are difficult to classify as such). Moreover, the Europroms statistics seem to be poor in monitoring waste and scrap flows.
• Incompatibility between IO-tables from the System of National Accounts and data on material flows. This problem is most apparent when one tries to connect detailed Substance Flow Analysis (SFA) to the product and sector classifications of input-output tables (the latter are often much more aggregated). Other incompatibility issues arise around: unit of measurement (kg of substance vs. monetary units), coverage of stocks (which are absent in IO-tables) and continuity over time of the model (the IO table being strongly bound to an annual set-up).
• A specific challenge with critical materials is to specify quantities of materials in specific products and accordingly to bridge these to the product classification of supply- and use- or input-output tables. In fact, large uncertainties are introduced by the need to have more detailed information on the critical material content in (most often non-ferrous metal) products. Not only is this information rarely available, also several conversions between monetary and physical layers and assumptions of homogenous products have to be made.
As proof of concept for the ‘dynamic Technology-Hybridized Environmental-Economic Model’, several case studies have been conducted within DESIRE’s work package on critical material indicators. The scope of the substance flow analysis case studies was for practical reasons limited to the regions “EU27” and the year “2007”. Case studies are performed for Tantalum (Ta), Indium (In), Neodymium (Nd) and several steel alloying elements Chromium (Cr), Manganese (Mn), Molybdenum (Mo) and Vanadium (V). All these case studies had the same sequence of methodological steps.
In order to derive critical material flows through Europe the EUROPROMS statistics (EUROSTAT Prodcom statistics) are used to first determine the apparent consumption of critical material-containing ores, materials and products. Secondly, a review of critical material concentrations in those products is performed. Combining these data and performing some additional data processing steps yields the apparent consumption of critical material in various European ores, materials and products, which were then categorized into production stages to result in a highly detailed flow-diagram of critical material in raw materials, semi-finished products as well as in products for final consumption (i.e. “Sankey diagrams”). The scope of the case studies within DESIRE’s WP6 was limited to the compilation of Sankey diagrams.
It is acknowledged, however, that as next steps the apparent consumption of end uses of critical materials in products can be used as input for a waste assessment model. With this model it would be possible to assess the expected future critical material recycling potential from consumer wastes by assuming Weilbull life-time distributions.
The results of the case studies (i.e. the Sankey Diagrams) for Indium, Tantalum and Neodymium are described in detail in the deliverables.
2.2 Biodiversity indicators
Ultimately, the aim with DESIRE’s indicator framework is to show the impacts that environmental stressors cause. To do so, the stressor results need to be characterized into various impact categories. One separated set of characterization factors was setup for assessing the impacts on biodiversity and ecosystem service functions. The research efforts in DESIRE focussed on the impacts of land use on biodiversity. In Marques et al. (2015), based on a literature review, it is explained that land use change is currently one of the main drivers of biodiversity loss in terrestrial ecosystems. Habitat loss and habitat degradation affect more than 80% of globally threatened mammals, birds, amphibians and plants.
Within DESIRE’s work package on biodiversity, three indicators have therefore been developed which can be coupled to the EE MR-IO framework:
(1) Bird species lost;
(2) Cumulative extinction risk of carnivorous mammals; and
(3) Carbon sequestration foregone.
The development of these indicators was dependent on spatially explicit information of land use by sectors included in EXIOBASE. That is because impacts on biodiversity and ecosystems can be very location specific (for example clearing a km2 of Amazon forest will represent very different impacts on biodiversity than clearing a km2 of an agricultural field). For the characterization of pressures the production and consumption accounting scheme from the MR EE-IO tables is used.
For each of three indicator types we summarize the conceptualisation, methodology and key results are described in detail in the deliverables.
2.3 Novel reference indicators beyond GDP
DESIRE’s work on novel reference indicators started from the acknowledgement that GDP and value added are not bad indicators for economic output as such, yet not always sufficiently connected to the input of natural resources and outputs in relation to services to society, i.e. a broader perspective on welfare. Resource efficiency ‘beyond GDP’ requires a link between natural resources (as input) to human well-being (as output). A challenge is than that ‘quality of life’ or ‘human well-being’ aspects have no one-to-one relation with production and consumption. There are much more, and more complex, relations to take into account.
Consumption of goods and services is only one of many mechanisms that contribute to human well-being. It is therefore crucial to acknowledge that the backbone of DESIRE’s indicator framework, the MR EE-IO model, focusses on environmental aspects of economic production and consumption transactions only. As for the part consumption is relevant to consider in a broader welfare context, there are no straightforward ways to linking consumption expenditures to human need satisfaction. Moreover, because consumption of a single product or service can satisfy several human needs at the same time. Another limitation of assessments through an IO-framework is that it represents flows in a single year, while past investments or changes in quality and quantity of stocks of different types of capital are important determinants for some aspects of human well-being, such as health or nature’s recreational values.
As alternative, Freyling et al. (2014) therefore proposed a preliminary framework where human needs and quality of life are at the core, and it is assessed how the output of economic activities that use natural resources satisfy human needs. This framework was further developed by Usubiaga et al. (2015), of which we provide a brief summary shortly below. The overall objective of this stream of work within DESIRE was to explore whether using alternative reference indicators (i.e. alternatives to GDP or value added) in the calculation of resource productivity or resource efficiency better reflects how the use of natural resources contributes to services to society, i.e. to human needs satisfaction. And accordingly, to conclude whether or not the use of novel reference indicators leads to different conclusions when compared to resource efficiency measures with relation to the economy.
A framework for ‘beyond GDP’ resource efficiency
The conceptual framework used in DESIRE for the development of ‘novel reference indicators’ depicts the various mechanisms through which natural capital, social capital, human capital, financial capital and manufactured capital contribute to people’s quality of life (or human well-being). This framework integrates elements from Max Neef’s human scale development and human needs (Max Neef et al., 1991; Max Neef, 1992), Ekin’s four capital model (Ekins, 1992), Sen’s capability approach (Sen, 1981, 1985, 1999), as well as the Conference of European Statisticians’ recommendations on measuring sustainable development (UNECE, 2014) that builds on previous work of Smits and Hoekstra (2011). In doing so the logic behind the System of National Accounts (SNA) and the System of Environmental-Economic Accounting (SEEA) is followed. Figure 2.1 gives a schematic overview of the conceptual framework.
Figure 2.1: DESIRE’s conceptual frame for ‘beyond GDP’ resource efficiency
Source: Usubiaga et al., 2015
Max-Neef considers that human needs are finite, few, classifiable, and non-hierarchal (Max-Neef, 1992). Thus, needs are not dependent on time or cultural factors, they remain invariable. In contrast to needs, their satisfiers are infinite and changeable. Satisfiers go beyond the goods and services provided by the economy. They can take many forms, including that of economic goods. Satisfiers can be seen as the forms of being, having, doing and interacting that represent everything that helps us in a specific time and place to meet our needs. Furthermore, needs remain always constant and inalterable, while the ways individuals choose to satisfy them varies across cultures and time.
Max-Neef grouped human needs in two categories: existential (fulfilling a variety of inherent states of activities) and axiological (value-based), which intercept and form the matrix shown in table 2.1. While the existential category refers to activities such as being, having, doing and interacting, the axiological category covers needs such as subsistence, protection, affection, understanding, participation, idleness, creation, identity and freedom.
Table 2.1: Fundamental Human Needs
Need Being (qualities) Having (things) Doing (actions) Interacting (settings)
Subsistence physical and mental health food, shelter, work feed, clothe, rest, work living environment, social setting
Protection care, adaptability, autonomy social security, health systems, work co-operate, plan, take care of, help social environment, dwelling
Affection respect, sense of humour, generosity, sensuality friendships, family, relationships with nature share, take care of, make love, express emotions privacy, intimate spaces of togetherness
Understanding critical capacity, curiosity, intuition literature, teachers, policies, educational analyse, study, meditate, investigate, schools, families, universities, communities,
Participation receptiveness, dedication, sense of humour responsibilities, duties, work, rights cooperate, dissent, express opinions associations, parties, churches, neighbourhoods
Leisure imagination, tranquillity, spontaneity games, parties, peace of mind day-dream, remember, relax, have fun landscapes, intimate spaces, places to be alone
Creation imagination, boldness, inventiveness, curiosity abilities, skills, work, techniques invent, build, design, work, compose, interpret spaces for expression, workshops, audiences
Identity sense of belonging, self-esteem, consistency language, religions, work, customs, values, norms get to know oneself, grow, commit oneself places one belongs to, everyday settings
Freedom autonomy, passion, self-esteem, open-mindedness equal rights,
means of communication dissent, choose, run develop awareness anywhere
Source: (Max-Neef et al., 1991)
For sector-specific ‘alternative’ outcome oriented indicators EXIOBASE’s MR EE-IO model is not sufficient. In order to connect novel reference indicators to human needs, next to economic indicators on production, income and consumption, functional output and outcome-oriented indicators are needed (most often at the meso or even micro level), as well as subjective indicators. Moreover, an ideal (panel) dataset would than provide explicit information on intentions behind actual consumer expenditures, i.e. to learn which needs are satisfied with certain purchases.
With such a dataset it would be possible to calculate the footprint of the households for every need category according to their expenditure and also calculate the footprint of government consumption in those same categories. In that sense we could assess (1) resource efficiency in terms of objective novel output and outcome indicators and also (2) resource efficiency in terms of subjective well-being.
Cross comparison of such indicators would reveal the (1) efficiency of different regions to provide outcomes from resources, (2) the efficiency of different households to achieve high levels of well-being through resource use, and (3) the different narratives that emerge when comparing indicators based on outcome-oriented and subjective well-being indicators.
Satisfaction of human needs in a resource efficiency context: 3 case studies
We have tried to operationalise this framework by means of three case studies that differ in their scope and level of detail: 1.) linking products in EXIOBASE to human needs (as satisfier), which is, for pragmatic reasons, based on expert judgement rather than the abovementioned ‘ideal panel dataset’; 2.) food and nutrition systems (as satisfier for the need subsistence); 3.) housing in a resource efficiency context (as satisfier for the need protection). The first case study proved that the needs of subsistence and protection are among the most environmentally intensive. This finding formed the rationale to explore their most important satisfiers, food and housing, in more detail in the second an third case study.
2.4 Conclusions on the extent to which indicators of the three “novel reference indicators” case studies could be linked to EXIOBASE
None of the three case studies has managed to completely couple all the proposed alternative reference indicators to the EXIOBASE classification. In specific cases within the food and housing case studies, a one-to-one correspondence between EXIOBASE product groups and the indicators was possible. Nonetheless, besides these exceptions, the indicators proposed should be seen as a separate set of metrics that can only be soft-linked to the input-output model.
The first case study has proposed novel reference indicators at the level of human needs, since there is not simple correspondence between the product groups represented in EXIOBASE and Max-Neef’s human needs. Due to the complexity and interlinkages between needs and products, it was not possible to use indicators at product group level that can describe the specific contribution of each product to the need(s) it intends to meet.
In the food case study, indicators at product (food available for human consumption, food intake) and aggregated satisfier level (i.e. food as a whole: food available for human consumption, food intake, net healthy food intake, quality-corrected net healthy food intake) have been proposed.
Strictly speaking the metrics given at product level could be linked to EXIOBASE from the consumption side, but there are two main factors to consider: The first one refers to the correspondence between the primary data source (Food Balance Sheets [FBS]) used to calculate the indicators and EXIOBASE. This correspondence is problematic, since FBS provide information on food made available for human consumption, yet the classification used represents primary products. Before reaching the consumer, most food products are usually processed one way or another, which requires assumptions to link primary products to the manufactured food categories in EXIOBASE.
The robustness of the results depends on these assumptions. Second, food is not only consumed as part of final consumption activities, but also in ‘food-related services’ such as health services, education, etc. This intermediate consumption is ultimately embodied in the final consumption of other goods and services. In the case study we have not assigned the related food consumption to the end product or service in which it is embodied, but to the food product itself. Hence, the product-level indicators need careful interpretation if they are to be related to individual product groups in EXIOBASE.
With regard to the metrics given for food as an aggregated satisfier, the consumption (intermediate and final) of all food products is combined to depict the per-capita level of nutritional well-being achieved by a country’s population. The set of indicators provided at the level of human need are provided separately and could only be soft-linked to the input-output model.
The housing case study has made use of two types of alternative reference indicators: functional output and deprivation. The functional output indicators used as reference for the construction phase (new dwellings built in terms of number and m2 floor space) can be attributed entirely to the construction sector. However, the activities of the construction sector go beyond building new dwellings (e.g. maintenance work and construction of non-residential infrastructure such as industrial facilities or roads). For this reason, there is not a one-to-one correspondence between the alternative indicator and the product group in EXIOBASE’s MRIO model.
As for the use phase, final consumption of energy by households is considered a good reference. In the case study we have used climate-corrected final energy use by purpose (e.g. heating, electrical appliances, etc.) from an external source. In itself indicators on physical amounts of energy use by use purpose are already effective monitoring instruments to assess developments in resource efficiency. There is future research potential to combine these kind of data with the existing EXIOBASE data classification by energy product. In that way footprints of type of energy used can be connected to the use purposes, which would increase the relevance of the functional unit of measurement even more.
The alternative indicators proposed following the deprivation approach relate the functional output indicators to the amount of people who have structural problems or are not able to keep their homes warm. The functional output indicators cover the material footprint or energy consumption of the whole population, while the deprivation indicator refers only to a fraction of the population. Therefore, the latter cannot be linked to the EXIOBASE product groups as an add-on item, but has to be considered separately and interpreted carefully.
2.5 Conclusion on DESIRE’s novel indicators: ready for uptake?
In DESIRE’s more experimental indicator development, just as in the ‘core framework’, we tried to causally link pressures stemming from socio-economic and nature interactions to the state of the environment and environmental impacts. In this subsection we generalize the lessons and conclusions from our development of resource efficiency indicators in the three ‘novel indicator’ domains: critical materials, biodiversity, and novel reference indicators ‘beyond GDP’.
Indicator concepts and methodology
A commonality of DESIRE’s research efforts in the novel indicator domains is that they required highly detailed information and data, beyond the scope of environmental extensions and the economic IO-framework as included in EXIOBASE. All three novel indicator domains therefore had to follow a case study approach, or in case of biodiversity impacts, had to set a very clear scoping boundary (i.e. bird species loss, mammals extinction risk, carbon sequestration forgone related to a maximum of 16 out of the 163 EXIOBASE production sectors.
Biodiversity is the only novel indicator domain where an appropriate link with DESIRE’s EE-IO framework could be made, in particular for ‘land use intensive sectors’, i.e. agricultural production and forestry. Environmental pressures of production and consumption are, through the link with land use and biomes/species habitats in particular geographical locations, characterized with the EE-IO framework. In addition, only for crop growing and harvesting in the year 2000, explorative calculations were executed on carbon sequestration forgone.
Regarding critical materials the required high level of detail for appropriate substance flow analysis made it a labor intensive as well as methodological challenging task to make sound linkages to EXIOBASE’s full EE-IO framework. Rather than full time-series analysis, pragmatically operationalized case studies on EU-level (EU27), for a single year (2007) were performed.
Regarding the research on novel reference indicator ‘beyond GDP’, just as in the critical materials domain, dedicated case studies needed to be executed to address specificities of resource use and environmental impacts from a human well-being and human needs perspective. None of three novel reference indicator case studies was successful in making a full coupling to the EE-IO framework of EXIOBASE. The primary reason being that there is no one-to-one relation between production and consumption and a broader perspective on welfare and human well-being. EXIOBASE as such is therefore not sufficient for resource efficiency assessments based on other references than GDP or value added.
In line with the findings for critical materials and impact on biodiversity, the general conclusion is that DESIRE’s indicator framework could not be complemented with a full set of alternative reference indicators for resource efficiency assessments for all products and sectors included in EXIOBASE. In the next subsection we will elaborate on the results, the involved uncertainty and what this means for indicator uptake.
Results, uncertainties and considerations for indicator uptake
The critical material case studies for Indium, Tantalum and Neodymium resulted in highly detailed flow diagrams of semi-finished products as well as final consumables. The results are especially informative as ‘standalone’ case study output as each critical material has its own specific use applications and hence impacts of supply risks etc. It is more difficult to draw generalized conclusions on overall resource efficiency. Particularly the Tantalum case study delivered a novel insight (i.e. its importance in hard disks). The Neodymium case, on the other hand, seems to highlight the level of uncertainty involved in the analyses’ methodology (a potentially large overestimation of European consumption when compared to absolute numbers from literature sources).
The work on biodiversity impact indicators proofed that the level of detail in which ‘land use intensive’ economic sectors are covered is crucial to appropriately link economic processes to regional specificities such as species rich biomes and species’ habitat preferences. EXIOBASE offers a good framework to track and trace the impacts of production and consumption on biodiversity, however the method depends on availability of good quality spatially explicit auxiliary data on land use changes, biomes and habitats, as well as stocks of vegetation for carbon sequestration. Uncertainties primarily relate to availability and quality of these auxiliary data, in addition to assumptions that had to be made on e.g. the grow back potential of vegetation for carbon sequestration. Especially for the latter, the carbon sequestration results should be considered explorative. For the indicators on bird species loss and extinction risk of mammals, uncertainties arise through averaging impacts on local grid level to country level before ‘footprints’ can be calculated with EXIOBASE.
Albeit the methodological uncertainties, the first results do provide plausible insights in the causal relation between, for example, agricultural production in, and harvesting of wood from, particular species rich biomes and how this affects other country’s ‘footprint’ of species loss or extinction risk through international trade. Such insights are valuable and can be used immediately in narratives. On the other hand, it might be more difficult for a receiving audience to attach a meaningful interpretation to the indicator results. For example, the average European citizen might not easily understand in what way they can counter any negative impact on e.g. bird species or mammal extinction risks abroad. This probably underscores the need for a good accompanying narrative in relation to the envisaged purpose with the indicator (a topic that we will further address in chapter 4).
Although contextualization of novel reference indicators to calculating resource efficiency is needed, the results of the three different case studies have in common that alternatives to monetary references such as GDP or value added were found to provide relevant, sometimes even more meaningful, insights.
In the food case study we have proposed four indicators (food available for human consumption, food intake, net healthy food intake, quality-corrected net healthy food intake) that describe the role of the different components of the food and nutrition system in meeting the need of subsistence. Compared to monetary indicators each of these metrics has advantages and disadvantages from a RACER perspective. Among the four proposed, only food available for human consumption is provided by a recognised source (FAO). Hence, food available for human consumption has more credibility than the rest, since its methodology is well established. Conversely, this metric does not sufficiently capture the contribution of food to nutritional well-being, since it leaves out key factors such as food waste at consumer level, overconsumption or dietary quality.
For instance, food waste is considered in food intake indicators, overconsumption in net healthy food intake and dietary quality in quality corrected net healthy food intake. Therefore, the relevance of the latter is higher than that of the others (and probably more acceptable in this context), yet it requires making more assumptions that negatively affect its robustness and credibility. The involvement of relevant institutions such as FAO, national statistical offices as well as other relevant stakeholders in methodological and data gathering activities would increase the robustness and credibility of the indicators, and eventually improve acceptance in the policy arena.
As for the work on housing, the alternative indicators selected for the construction phase (number and m2 of new dwellings) might be more relevant than monetary indicators in criteria such as relevance, which improves its acceptability. This comes at the expense of credibility due to the necessary assumptions to split the footprint, value added and final consumption expenditure of the construction sector between the activities related to dwellings and other construction work. For the use phase, the functional indicator selected (households resource use by purpose, e.g. energy use for space heating) is a clear improvement over monetary metrics. Especially in the field of energy, this type of monitoring is relatively well established and is used to inform energy policies. The latter can rely on other data sources than EXIOBASE, i.e. data from the Odyssee-Mure project.
Our overall conclusion is that it seems to be clear that using alternative indicators as reference yields different results than when using monetary metrics. Given that the latter could potentially be misleading when used in the wrong context, this type of work should be further encouraged to eventually develop better metrics of well-being in general, and resource efficiency in particular. For now, due the level of uncertainty involved, the first results should be used with some caution.
3 Indicator development: optimal set of resource efficiency indicators
After completion of the EE-IO time series, the database could be used for systematic analysis. In this chapter key results are described of: 1.) analyses on driving forces behind resource efficiency indicator results (to help identify the main ‘driver’ indicators); 2.) calculation methods and possibilities for MR EE-IO model simplification; and 3.) a systematic analysis of options to minimise the indicator set to a small ‘optimal’ set.
3.1 Driving forces behind Resource Efficiency indicator results
Resource-efficiency indicator results give information on the overall environmental impact, but are not able to capture the driving forces behind resource-efficiency indicator scores. This leaves policy makers with knowledge of how well a socio-economic system is performing, but with little (quantitative) knowledge of why the system is performing as observed. Unravelling the mechanisms that drive performance is considered essential in designing effective policy to attempt to influence future impact.
Drivers of impact can be seen from a number of perspectives – firstly, the temporal aspect where drivers of change in indicators are identified such that impact is analysed over time, and with respect to different socio-economic variables. Secondly, the life-cycle or supply chain perspective, where key actors in the supply chain of a good or service are identified that drive impact. An example here is the understanding of construction requirements used within the provision of services by the services sector. If we shift to a service based society, how dependent do we remain on bricks and mortar? Systematic analysis of indicator results based on the EE-IO model is possible using structural path analysis – where impact pathways are analysed and ranked based on contribution to overall impact. Finally, a combination of the temporal and structural aspects can be integrated into a single analysis by breaking down change into several key parameters of policy interest.
In DESIRE’s WP9 (task 9.2) indicators were broken down into the factors population growth, population affluence, changing consumption patterns, changing industrial production, changing trade relationships and changing resource efficiency were explored. For all details of this structural analysis of drivers we refer to Wood et al., 2016.
Globally, environmental impacts are growing. Resource efficiency indicators, however, show different trajectories depending on the perspective taken. Production based resource efficiency indicators (the impact in a certain region) generally show resource efficiency improvements for developed countries, and from a decoupling perspective, we thus see decoupling occurring for production based impacts. From a consumption perspective (the impact embodied in final demand in a certain region, utilising supply-chain analysis) we see a very different picture. Whilst the main regions (EU, US, China) are still generally decoupling, the rate of decoupling is greatly reduced, and turns negative in some cases.
At the country level, especially for wealthy countries and most evident in the EU, we see a much greater rate of negative decoupling (that consumption based impacts are outpacing economic growth). When we look at what is driving this upward growth in resource use, we consistently see the impact of population (small steady upward driver), of affluence (strong usually upward driver), and the impact of trade (moderate upward driver) – both to intermediates and final consumers. In terms of products and supply chains that are driving this growth, we see the effect of the construction activity as being one of the key sectors, particularly across material and greenhouse gas indicators. Downward drivers include both impact intensity (the ratio of impact per unit output) – where we see that this has had a strong downward effect on indicator scores over time; and final demand mix – which has had a weak negative impact on indicator scores over time.
Looking at the drivers of this increased consumption based impacts relative to GDP, we can conclude the following strong trends: 1) there is little evidence of the increased impact being due to changing types of products consumed; 2) construction is the only main exception in terms of product mix – it is also the most relevant activity in terms of a number of environmental impacts and has seen the greatest change over time; 3) trade has shifted strongly from developed to developing countries; 4) the overall impact embodied in trade has greatly increased, in line with the increased volume of, but also due to increased impact per unit of trade – trade has shifted to less efficient producers.
3.2 Optimal calculation methods and level of detail
A fundamental challenge in indicator development by European institutes like EEA and Eurostat is the consumption-based perspective. In this perspective, in essence all impacts along the (global) value chains related to European consumption should be taken into account. This implies that insight is needed in the ‘pollution and resources embodied in trade’. Through European Commission regulations it can be ensured that high quality statistical data are available for EU-member states. However, for European institutes, it is much more difficult to ensure that data in the same quality is available for global trade partners. At the same time a full and harmonized (e.g. in terms of international trade and the level of detail) Global Multi-regional Environmentally Extended Input-Output tables with harmonised trade data are needed to assess the ‘pollution and resources embodied in trade’.
Yet, from the perspective of official statistics, ‘science-based’ global MR EE-IO databases such as EXIOBASE version 3 are not preferred for reasons of deviations from original official statistical publications. Trade linking and further detailing of national IO-publications is unavoidable for global environmental impact assessments, however. Within DESIRE’s WP9, Tukker et al. (2016) have therefore analysed how national statistical institutes and EU statistical services can be provided with relevant information that may be acceptable to be used. For this purpose the following research activities have been carried out: 1.) a review of methods to calculate pollution embodied in trade; 2.) a review of various global MR EE-IO tables with listing of their pros and cons; and 3.) a review of uncertainties related to calculations based on MR EE-IOs. For the full details of these analyses we refer to the report of Tukker et al. (2016). We only briefly summarise the main findings below.
Calculation methods
Tukker et al. (2016) found that there is a clear need to cover full international value chains to optimally assess footprints, pollution and resource extraction. Global multi-regional environmentally extended input-output databases seem to be the best source to base calculations on as they not only cover full value chains but also are consistent between the production and consumption perspective. The latter meaning that total emissions and resource extraction by all economic sectors in all countries equal the footprint of global final demand. In contrast, the often used Domestic Technology Assumption (DTA), where it is assumed that imports are made in the same way as domestic production, can lead to erroneous results since production technologies and hence associated environmental impacts can actually vary quite significantly across countries. This is especially apparent in imports of small countries, which often have a fundamentally different production structure than the import’s country of origin. The DTA (including the slightly better price adjusted DTA) can therefore best be used as a ‘last resort calculation method’.
Deeper analyses of calculations based on global MR EE-IO databases showed that differences in allocation principles, definitions and data sources for extensions matter for the footprint results and associated level of uncertainty. The former relates to using a residential instead of a territorial approach. The residential approach takes a global consumption perspective and herewith accounts for all activities/emissions and resource uses in individual countries along the global value chain (including e.g. fuel bunkers). The analyses proofed that different allocation mechanism cause fundamental differences in footprint outcomes. The analysis further showed that differences environmental extension datasets (on e.g. national industrial emission totals) is a cause of uncertainty and differences in calculated country footprints. We will elaborate on this below.
Level of detail
Regarding the level of detail of global MR EE-IO tables, Tukker et al. (2016) concluded that for economic analysis purposes an aggregated sector structure (such as the 30 sectors in the OECD/WTO’s Trade in Value Added database) is quite appropriate. For the calculation and analyses of carbon footprints, aggregated EE-IO tables with sector detail of 30-60 sectors (such as the WIOD or GRAM databases), still provide plausible results.
However, for the calculation of water, material and land footprints, a high level of sector/product detail proved to be essential. Aggregating EXIOBASE to the standard 60 products or sectors that Eurostat uses in official SUT and IO publications, led to clear changes in country footprints. When one wants to look at the environmental footprint of specific product groups, detail certainly is essential. The reason behind is that the material intensity, water intensity and land intensity of specific product categories/industry sectors varies much more than e.g. the carbon intensity or the created value added.
Moreover, the level of detail of environmental and material extensions like CO2 emissions, other emissions, resource extractions, water use and land use is perhaps even the single biggest cause of differences in calculated country footprints. Harmonization of extensions across the various global MR EE-IO databases seems to be a relatively easy option to significantly reduce uncertainty of footprint results (e.g. by using the resource extraction database as recently developed by the UN International Resources Panel). Although further work is still needed here on water, land and emission extensions, it is likely that using a harmonized source for extensions will reduce the level of uncertainty in footprint calculations with more than 50%.
Possibilities to simplify EE-IO data?
The overall conclusion of Tukker et al. (2016) is that for land, water and material footprint calculations no simplification of global MR EE-IOs in terms of a low sector resolution is possible. It is likely that detailed databases such as EXIOBASE will provide superior results over less detailed databases. For carbon footprints and particularly value added, this highly detailed sector resolution is less relevant.
3.3 “Optimal” Indicators
As final task in DESIRE’s work package 9 the potential for reducing the number of indicators for environmental assessments is examined through application of a statistical Principal Component Analysis (PCA) in combination with multiple linear regression. This methodology has been applied by Steinmann and Huijbregts (2016) to two fundamentally different datasets, both with the objective to identify a [statistical] optimal (i.e. smallest) indicator set. The rationale behind is that it might be unpractical to base decisions on product Life Cycle Analysis (LCA) or on environmental policy on a large number of indicators simultaneously. The objective therefore was to identify a limited set of indicators that is sufficiently small for efficient decision making and at the same time still covers overall environmental impact, i.e. covering both midpoint impacts of resource use and the full cause-and effect chain in damage-based or endpoint environmental impacts (e.g. impacts on ecosystems and/or human health).
To find an optimal set of environmental indicators to cover the variance in the rankings of a large number of products, the PCA is first based on a selection of 976 products and 135 environmental indicators from the Ecoinvent 3.1 database (Moreno Ruiz et al., 2013) and, secondly, on 93 impact indicators for 7589 product-sector combinations from the EXIOBASE database. The PCA has been combined with multiple regression analysis to arrive at a minimum set of indicators explaining the variance in the product ranking. In addition the extent to which four commonly used resource-based indicators (fossil energy, water, land and materials) are representative of the total variation in product rankings is tested.
Whether based on the life cycle impacts per kg of material or the impacts per million euros of consumption, strong correlations between the different indicators of impact were found. This means that there is a large potential for reducing the number of indicators.
The analysis based on the Ecoinvent database showed that 92% of the variance in product rankings is covered by only 4 out of the 135 initial environmental indicators. A set of six indicators covered slightly more of the statistical variance (i.e. 92,3%). This best set of six indicators relates to climate change, ozone depletion, terrestrial ecotoxicity, the combined ecosystem effects of acidification & eutrophication, marine ecotoxicity and land use.
In addition, the four resource-based indicators together accounted for 82% of the variance in material rankings. The results suggest that it is best to use the fossil energy indicator if just one of the simple resource-based indicators has to be selected. With an explained variance of 72.9% this seems to be a reasonably good indicator of overall impact. The explained variance can be raised to 76.8% by adding material use. Adding land use raised the explained variance to 80.1%, while a set of all four resource-based indicators, including water use, covers 82.0% of the total variance in our dataset. The water footprint appeared to be less important than the other footprints for our dataset; this is due to the fact that water consumption is related to both the energy-intensive process of electricity generation and the land-intensive process of crop production.
For the EXIOBASE dataset the 93 environmental impact indicators could be reduced to seven indicators related to freshwater and marine ecotoxicity, photochemical oxidation, climate change, acidification & eutrophication, photochemical ozone formation and blue water withdrawal. These seven indicators together covered more than 90% of the variation. Similar to the analysis on the Ecoinvent dataset, the performance of the resource based indicators was also tested. The four resource-based footprints together accounted for only 49% of the variance in product-sector rankings. This means that sets of 1 to 4 resource indicators cannot cover the same amount of variance that can be explained by one toxicity indicator. Supplementing, however, the two best resource-based indicators (energy and land) with the best toxicity indicator (freshwater aquatic ecotoxicity potential, infinite time horizon) the explained variance is increased to 74.8%.
While the optimal sets maximize the amount of covered variance, the recommended indicators are not necessarily the most preferable using additional criteria, such as the RACER (Relevant, Accepted, Credible, Easy and Robust) criteria. For both datasets there are several indicators with approximately the same amount of explanatory power. This means that alternative sets of indicators can be defined which are only marginally worse in terms of explained variance compared to the statistically preferred set of four and seven indicators proposed here.
In both cases, only three out of the four resource-based indicators seem to be of real added value. For both datasets these were the indicators of energy, land and material. This is due to the fact that the (agricultural) water consumption is strongly correlated to the land footprint, especially in the EXIOBASE dataset, making one of the two indicators redundant. Using two or three simple resource based indicators would eliminate the need for the complicated mid- and endpoint damage models, but has limited coverage of the impacts associated with toxic emissions, especially for the EXIOBASE dataset.
The overall conclusion based on the statistical analyses to test options to reduce the environmental indicator set is that the large set of indicators can indeed be reduced to a small key set, representing the major part of the variation in environmental life cycle impacts between materials and of the variation in product-sector combination in a Multiregional Input-Output model.
Potential Impact:
Towards indicator implementation
Now we have summarized the key results of DESIRE’s different work packages and converted these into conclusions, we devote this last chapter on a roadmap towards indicator implementation and options for institutionalization of indicators based on global Environmentally Extended Input-Output databases such as EXIOBASE. In doing so we elaborate on a discussion note that functioned as input to stakeholder discussions during the project’s final conference. Recommendations we took home from this last ‘brokerage and dissemination event’ are integrated in the indicator implementation roadmap.
A major accomplishment of the DESIRE project is to add a time series perspective to EXIOBASE. With the EE-MRIO dataset of EXIOBASE version 3, a powerful tool is now available for analysis of various environmental-economic relations in Europe and beyond. Databases like EXIOBASE e.g. help to provide insights about how consumption drives environmental pressures. EXIOBASE thus is a relevant information source in support of evidence based policy related to e.g. Resource Efficiency and the Circular Economy.
So far, however, the development of the database and first analyses based on it, has been primarily a scientific undertaking. The challenge now is to bridge the gap between this scientific research initiative towards uptake and implementation of indicators in policy processes. One thing that we have learned during various policy-science brokerage and dissemination events is that a more formal status of Environmentally Extended Multiregional Input-Output models is for some users an important precondition for indicator uptake. From this perspective, it is desirable when supra-national statistical institutions adopt databases such as EXIOBASE and further develop these along the lines of international harmonized standards.
On the other hand we have learned that even without such formal institutionalization, and sometimes even with indicators that still have room for methodological or data quality improvement, information from databases as EXIOBASE can already be relevant as ‘early signalling’ or ‘agenda setting’ mechanism.
In this chapter we will place the ‘core set’ of the DESIRE indicator framework, as well as the more novel indicator results, in the context of discussions throughout the project on indicator uptake and implementation. With this we aim to develop an indicator implementation roadmap in which we qualify indicators that already have a “statistical stamp” and could be taken-up in policy processes without further needs for improvement, as well as defining next steps to further improve indicators that are relevant but currently lack the full “statistical quality stamp”.
Uptake and institutionalisation of DESIRE indicators
DESIRE’s global environmentally extended input-output time series database, EXIOBASE version 3, primarily offers a rich knowledge base on which a large set of indicators can be calculated, customized to different analytical purposes or policy needs. In a discussion on potential users of EXIOBASE during DESIRE’s final conference on the 21st of January 2016 in Brussels, it was for example stated that DG JRC could definitely use the indicator framework with underlying raw data for policy-support research. Other (potential) users might have other information needs for other purposes, for which the required data and indicator quality also might differ. It is for this reason that we will report in this section how DESIRE’s indicator framework, in its current state of development and quality level, can be used for different purposes.
Indicator functions: different use purposes, different needs
What should the indicators do?
Figure 1: Indicator function and the required level of quality and detail
1. Signalling and agenda setting: plausible information on recent trends.
One of the lessons we took home from the final conference is that indicators may not have to be of perfect quality yet to signal relevant environmental-economic trends. There is a trade-off between precision and the relevance to inform ongoing policy processes. It was stated that feeding policy processes with relevant (new) information sometimes is more important than waiting for the moment when indicators are of perfect methodological and statistical quality. The latter can be taken up further during formal institutionalization processes.
During the conference a general need for timeliness of indicator results (i.e. showing recent developments based on up-to-date data) was expressed, however it is acknowledged that there often is a time lag in official statistics. In this context the now-casted data of EXIOBASE can already play a role, although there are uncertainties given that GDP is used as primary indicator for the nowcasting (for reasons of high correlation) in combination with trends from earlier data points.
We acknowledge that this methodology works better for environmental indicators related to energy and carbon and has a higher level of uncertainty for material indicators. The latter category would clearly benefit from timely available official statistics. The now-casted data are not yet usable for formal monitoring and in cases where accountability is at stake. However, to communicate trends with the idea to trigger debate on the need and possibilities to intervene with policy action, EXIOBASE’s now-casted data can be a useful information source already.
2. Communication
Early signals and trends can be communicated to policy departments, politicians and the general public, with the aim to stress a sense of urgency and to trigger debate. Easy to understand, yet credible messages are relevant for this indicator purpose. A smaller set of indicators probably enhances the easiness to digest the communicated messages. In this context, DESIRE’s indicator optimization results can come in helpful, especially as there appears to exist strong correlation between various environmental impact indicators. For the purpose of triggering debate there is no use to communicate results of different indicators that are strongly correlated. However, it is recommended to carefully select the indicators with (potentially) higher scores on RACER-criteria easiness and acceptance. We underscore ones more that the toxicity indicators that came out of the statistical optimization analysis might than not be the best choice per se. Careful selection of indicators to support the message one would like to communicate will enhance the meaningfulness of information for the targeted audience.
During a discussion at the DESIRE final conference it was stressed that choices for specific indicators to communicate, might potentially “hide” misleading messages. For example, a too strong focus on resource inputs and waste outputs (e.g. in the context of a Circular Economy) might have a risk of losing focus on carbon dioxide emissions (e.g. in the context of Climate Action). Whereas perhaps the largest “waste” of socio-economic and nature interactions are related to energy and associated carbon emissions. In the same vain it was mentioned that outcomes of Raw Material Consumption calculations might be dominated by the relatively high mass volumes of gravel and sand.
In this context it was recognized as an asset of EXIOBASE that many economic-environmental relations can be shown and communicated. There is value in communicating cause and effect chains of economic-environmental interaction to a wider audience. Probably credibility is more important here than precision to the last decimal. EXIOBASE seems to be fit for this purpose already.
3. Monitoring of progress towards policy goals and targets
A third function of resource efficiency and environmental impact indicators stemming from EXIOBASE relates more to the end of the policy cycle, where progress towards formal goals and targets is monitored. This is where formal institutionalization, stricter requirements for data quality and comprehensiveness will come into play. This is the reason why this indicator function is situated in the lower part of figure 1.
How do DESIRE’s ‘core indicators’ and ‘novel indicators’ currently fit to these purposes?
We consider the ‘core data set’ of EXIOBASE, i.e. the historical data for the years 1995-2011, covering resource use, resource efficiency and environmental impact (i.e. the resource inputs and (metabolic) outputs of energy, materials, waste, land and emissions) in relation to society-nature interactions through production and consumption (including international trade), ready for immediate uptake, fitting all three abovementioned use purposes.
We do need to stress, however, that there are deviations from official (national) statistical publications. Due the high level of detail imposed, and the need for balancing international trade, deviations from official statistical sources are unavoidable (we will elaborate on this in section 4.2). Moreover, the time series are created with the idea to arrive at plausible year-to-year changes. In the construction of time series, as many official data sources with highest level of detail as possible, are used to respect structural changes in national economies. However, balancing the system (i.e. meeting restrictive conditions such as assuring that total supply matches total use) unavoidably implied deviations from official (National Accounts) statistics. For these reasons, one can expect to find deviations from other historical statistical sources when comparing actual numbers from EXIOBASE. It is for this reason that EXIOBASE indicators, in general, do not have a “statistical stamp”. The year-to-year trends are considered plausible, however.
As already mentioned in the previous subsection, the now-casted data (i.e. the years 2012-2016), have a higher level of uncertainty because of estimations based on earlier data points and (partially assumed) correlations. For this reason these now-casted years better serve signaling and agenda setting purposes.
Particularly in relation to monitoring progress in the 5 priority areas of the Circular Economy Package, it was asked during the final conference how DESIRE indicators fit. As for the priority areas plastics, food waste, and construction and demolition, EXIOBASE offers relevant as well as plausible information. During the discussion it was said that EXIOBASE can only to a lesser extend provide information on the priority area biomass and biobased products. Related to the priority area of critical materials, DESIRE’s ‘novel indicator’ development efforts only delivered insight in production and consumption flows for a small selection of case studies, covering a single year. The results are informative, yet not sufficient to fully support monitoring purposes in the this priority area field of the Circular Economy Package.
The results of the other ‘novel indicator domains’, biodiversity and novel reference indicators (beyond GDP), are subject to methodological challenges and uncertainties related to, sometimes strong, assumptions made. Albeit these methodological challenges that indicate a need for further development, the first explorative results do provide relevant insights in the different cases.
In the case of biodiversity, the results on bird species loss, mammal extinction risk and forgone carbon sequestration clearly point to environmental impacts related to production, trade and consumption of specific sectors and products. These insights can surely be used for communication purposes. It is then recommended to accompany the indicator results with a clear narrative in order to enhance the meaningfulness for the targeted audience. Narratives are equally important when using novel reference indicators ‘beyond GDP’. We have found that other indicators than GDP or value added can be relevant and meaningful, however that good contextualization to calculating resource efficiency is needed. Due to methodological challenges and assumption made, in addition to the fact that subjective views can easily be at stake, signalling and communication with the aim to trigger debate must, for now, probably be the dominant use purposes. In specific fields, auxiliary data sources to EXIOBASE (e.g. EU-SILC and Odyssee-Mure) offer relevant information for monitoring purposes. In some cases these data can be better used as such, complementary to EXIOBASE rather than integrated.
Indicator institutionalization
From the perspective of Eurostat, which, besides provision of official statistics, has harmonization of statistical methodology as task, institutionalization of DESIRE’s results is an important condition for implementation. Regular updates of EXIOBASE are required for formal institutionalization.
The normal procedure of Eurostat is to formally ask EU member States to submit data in a predefined (hence harmonized) format. Formal data collection with regard to EXIOBASE can thus only be the final step in an institutionalization procedure. This means that attention should first go to harmonization of data and indicator concepts and methods. In this context it was mentioned during final conference discussions that a comparison of methodology, level of detail, etc. with other existing global multiregional EE-IO databases would be highly relevant. Results of such a comparison could then feed into Eurostat’s deliberations on the best statistical methodology and the way forward (e.g. in relation with ongoing processes with regard to the Trade in Value Added database, together with the OECD and WTO).
Within DESIRE’s WP9, a brief comparative analysis with other existing global MR EE-IO databases, all with their own specific strengths and weaknesses, has been carried out. The main characteristics of the currently available global MR IO databases are shown in table 4.1. The main conclusions of the comparison are that:
• the high level of product/sector detail of EXIOBASE is in particular important for agricultural-, industrial-/manufacturing- (e.g. metal) and energy-producing sectors in relation to environmental issues associated with land use, water use, or resource use.
• IDE-JETRO’s AIIOTs, in contrast, offers the longest time series (with a data point back to 1975), with a relatively detailed product classification (76 sectors). A weakness, however, is its small country coverage. On the other hand, the manual handling of data transformation enables a high level of harmonization among constituent national tables.
• EORA and GTAP discern considerably more countries than WIOD, EXIOBASE, IDE’s AIIOT or GRAM. This has important advantages in assessing impacts of final consumption that take place in relatively poor countries with a low GDP not covered in other databases (Lenzen et al., 2012). Moreover, a large separate country coverage (as opposed to a large aggregated RoW regions) is important to attribute impacts to individual countries.
• Overall, with its broad coverage of countries and varying sector detail per country, EORA seems to split up the global economy in most products and sectors and it is the only database that provides uncertainty information for its estimates.
• WIOD, to conclude, has some clear advantages with regard to institutionalization as it is the only database with a consistent annual time series in both current and previous year’s prices, as well as it is fully consistent with the National Accounts statistics which is important when a link is required to other (socio-)economic data (e.g. for productivity analyses).
Related to institutionalization of DESIRE’s results it was stated during the final conference discussions that there recently seemed to be more emphasis on expanding the level of detail in NACE financial- and other service sectors in data collection for official statistics. The comparison between EXIOBASE and other existing global MR EE-IO databases, as well as other analyses within DESIRE’s WP9 on the optimal level of detail, made clear that a high sector- and product detail is crucial for meaningful assessments of environment-economic interactions and resource efficiency. In the same context it was underscored that policy demand drives the development of official statistics and the associated level of detail that indicators can provide. There is thus a need for coordination between European Member State’s and the European Commission’s policy departments and statistical institutes in the formal institutionalization process of MR EE-IO databases as EXIOBASE to ensure a minimum required level of detail in sectors/products with high environmental-economic interactions.
Table 4.1: Review of the main Global Multiregional Input-Output databases
Recommendations on creating footprint data with a “statistical stamp”, actions and timeline
We concluded that global MR EE-IO models are the preferred calculation method for consumption based accounting of footprints, for which a high level of product sector detail is required to obtain relevant results. At the same time it should be acknowledged that the compilation of global multi-regional input-output databases requires a high level of harmonization and consolidation of often conflicting data sources. For this reason, there is no other option than to deviate (sometimes significantly) from official statistics that national statistical institutes provide. One key reason is that all imports, summed to a global total, do not match the global total of exports, whereas in reality international trade obviously is a zero-sum game. The same imbalances between imports and exports occur when bilateral origin-destination trade flows are confronted with each other. Given that national statistical institutes have a national mandate, it is one of the main reasons that the construction of global (EE) MR-IO databases such as EXIOBASE has mostly been efforts of scientific research consortia. An important question for formal institutionalization is thus how a “statistical stamp” can be attributed to databases as EXIOBASE and the calculation of environmental footprints.
The first and fundamental solution to the main problem of imbalances in international trade data, and perhaps the “Royal route”, is that all national statistical institutes (NSIs) in the World collaborate, e.g. within the context of the UN Commission on Economic and Environmental Accounts (UN CEEA), on a data exchange platform that allows NSIs to end-up with supply and use tables as well as input-output tables that are mutually consistent between countries. This, however, is likely to be a long-term endeavor, unlikely to provide results in the coming 5 years. Our recommendation for this longer term solution is that NSIs and supra-national statistical institutions make use of the experiences from global (EE) MR-IO practitioners to identify the most pressing data inconsistencies at international level.
Based on suggestions of Edens et al. (2015) a second approach is described by Tukker et al. (2016): using a “Single-country National Accounts Consistent” footprint approach. In this approach, an existing global MR-IO database will be adjusted for the single country of investigation, by using the IO-data and environmental extensions from official national statistical sources and fixing these (i.e. imposing a restrictive condition that these national totals cannot change) before rebalancing the whole global MR-IO database again. Only after this rebalancing, the footprints for the country of interest can be calculated in a way that is aligned with national accounts and other official statistics.
The main drawback of this method is, however, that plugging-in the national data in the global MR-IO database and rebalancing the model is labor intensive. Moreover, uncertainties may increase for cross-country footprint comparisons, e.g. as different extension data sources are now confronted with each other (see earlier remarks of this being a potentially big cause of uncertainty).
There currently is one global multiregional input-output database that is produced by a supra-national organization, i.e. the OECD/WTO Trade in Value Added database (TiVA). A further, and more preferred, refinement of a footprint calculation approach with a “statistical stamp” can thus be to use this TiVA database as a starting point. This is a trade-balanced database but with a coverage of 30 sectors too aggregated for the calculation of water- material-, land- and emission footprints. A way forward is to use the detailing procedures as developed particularly for EXIOBASE, and the optimization procedures as developed for EORA, to arrive at a level of 100-200 industry sectors that is appropriate to perform proper footprint calculations. Lastly it is recommended to use internationally harmonized data sets for carbon emissions (e.g. based on IPCCC or IEA energy flows plus emission factors), materials (e.g. the recently published UNEP International Resources Panel), land and water, and add these to the more detailed TiVA database.
In this way, a database could be created that at an aggregated level has the “statistical stamp” provided by the OECD, uses extensions that are harmonized/commonly accepted, but also can provide more detailed information (through a procedure backed by a number of credible, scientific institutes). This would, for the first time, give a global multiregional input-output database that probably has a higher level of credibility as the individual scientific databases such as WIOD, EXIOBASE, GTAP or EORA. Such a database, that holds a middle ground between official statistics and scientific work, could be a good compromise for any NSI or practitioner to work with.
Our recommendations for the short term, in addition to moving forward based on the TiVA database as just described, are to agree in the formal statistical gremia on:
• a preferred and harmonized way to calculate footprints (i.e. using a true global value chain approach rather than other allocation mechanisms;
• in doing so, taking the residential perspective as starting point;
• and avoiding neglecting emissions or resource uses related to e.g. international bunkers;
• ensure the use of harmonized extensions databases.
It is likely that such an approach will significantly reduce differences in footprint calculations for countries. Moreover, given the importance for environmental-economic interactions on a global level, in which Europe is an important driver for environmental impacts outside the EU, it is recommended that the EU uses such a harmonized data and analysis tool in a role of “knowledge broker” to non-EU countries on how objectives such as the UN Sustainable Development Goals can be met.
List of Websites:
Name, title and organisation of the scientific representative of the project's coordinator:
Prof. dr. Arnold Tukker,
Netherlands Organisation for Applied Scientific Research TNO
Tel: + 31 88866 8310 (work), + 31 6 51980344 (mobile)
Fax: + 31 88866 0757
E-mail: arnold.tukker@tno.nl
Project website address: http://fp7desire.eu/
DESIRE is a FP7 project that has developed and applied an optimal set of indicators to monitor European progress towards resource-efficiency. The project ran from September 2012 to February 2016. We proposed a combination of time series of environmentally extended input output data (EE IO) and the DPSIR framework to construct the indicator set. This approach has used a single data set that allows for consistent construction of resource efficiency indicators capturing the EU, country, sector and product group level, and the production and consumption perspective including impacts outside the EU. The project has:
• Improved data availability, particularly by creating EE IO time series and now-casted data
• Improved calculation methods for indicators that currently still lack scientific robustness, most notably in the field of biodiversity/ecosystem services and critical materials. We further have developed novel reference indicators for economic success.
• Explicitly addressed the problem of indicator proliferation and limits in available data that have a ‘statistical stamp’. Via scientific analysis we select the smallest set of indicators giving mutually independent information, and show which shortcuts in (statistical) data inventory can be made without significant loss of quality.
The project comprised further Interactive policy analysis, indicator concept development via ‘brokerage’ activities, Management, and Conclusions and implementation including alignment and hand over of data and indicators to the EU’s Group of Four of EEA, Eurostat, DG ENV and DG JRC.
Partners are:
1. The Netherlands Organisation for Applied Scientific Research (TNO), Delft, Netherlands
2. Wuppertal Institute (WI), Wuppertal, Germany
3. Alpen Adria University - Institute of Social Ecology (UNI-KLU), Vienna, Austria
4. Institute of Environmental Sciences, University of Leiden (UL-CML), Leiden, Netherlands
5. SERI - Sustainable Europe Research Institute, Vienna, Austria
6. Norwegian University of Technology (NTNU), Trondheim, Norway
7. Radboud University (RU), Nijmegen, Netherlands
8. Fundação da Faculdade de Ciências da Universidade de Lisboa (FFCUL), Lisbon, Portugal
9. Vienna University of Economics and Business (WU), Vienna, Austria
10. 2.0 LCA Consultants (2.-0 LCA), Aalborg, Denmark
11. Martin-Luther-Universitaet Halle-Wittenberg (MLU), Halle, Germany
For more information contact the co-ordinator at: arnold.tukker@tno.nl
Project Context and Objectives:
In the past fifty years humans have consumed more goods and services than in all previous generations put together. Various sources estimate that in a business as usual scenario global population and wealth growth until 2050 may lead to an annual resource extraction that is 2 to 4 times higher as in 2000. This is a trajectory that could ‘crash the economy against the Earth’.
The Europe 2020 Strategy establishes resource efficiency as one of its fundamental Flagship Initiatives for ensuring the smart, sustainable and inclusive growth of Europe. The Resource Efficiency Flagship should "help the EU to prosper in a low-carbon, resource constrained world while preventing environmental degradation, biodiversity loss and unsustainable use of resources". In this, a broad definitions of resources is used beyond just e.g. material extraction, including for instance human health and ecosystem health that can be impacted by emissions from economic processes. The Resource Efficiency Flagship has close connections to the Raw Materials Initiative and Innovation Union Flagship. Without access to and smart use of critical materials, European industry will face considerable challenges. Smart resource management and eco-innovation are essential to realise the innovation and competitiveness goals of the Innovation Union Flagship. Resource Efficiency has further relations with other EC policy initiatives, such as the EU Sustainable Development Strategy and the Sustainable Consumption and Production Action Plan. To support this effort, the EU FP7 Environment theme has called for an indicator project in the field of resource efficiency.
This challenge of addressing multiple resource constraints implies that the EU policy agenda in the next decade(s) needs an excellent insight in how resources are used in society, and particularly how access to and efficient use can be fostered. Several key aspects need to be considered:
• A consistent framework of resource indicators should be made available. Indicators for resource efficiency of production processes should be consistent with indicators for the resource efficiency of consumption patterns and these again should be consistent with how the development of resource efficiency is tracked on the national, EU and global level. Such a framework allows one to link consumption to production and to evaluate the contribution of innovations to resource efficiency.
• Material resources, which are preserved in use and recoverable, afterwards must be accounted for taking a consistent life cycle approach. Resource flows must be followed. Indicators should be based on modelling of resource extraction, renewable resources, inputs, flows, within and across industrial sectors, and the final outputs.
• The indicator framework needs to link resource use to the ultimate physical and economic impacts of resource scarcity, ecosystem service or human health and should be developed in a way that it can accommodate new knowledge on mechanisms of action. Interlinkages between different pressures on the state of resources and ecosystem services need to be adequately considered, including environmental impacts of emissions and the like.
• Indicators should look deeper than just the national level. The role of global trade in enhancing resource utilization needs to be adequately considered. An adequate sector breakdown should be conducted, distinguishing in particular the public and the private sector through the use of the national accounts classification (non-financial corporations, financial corporations, general government, households, non-profit institutions serving households and the rest of the world) and also allowing further disaggregation to sections (e.g. manufacturing) and subsections (e.g. manufacture of transport equipment).
• Long time series and now-casting should be considered.
• The reference for calculating resource efficiency (now often economic parameters such as GDP, or population) should be refined to better reflect the benefit of resource use.
The DESIRE project (i.e. Development of a System of Indicators or a Resource efficient Europe) addressed these key aspects. The main goal of DESIRE was to develop and apply an optimal set of indicators and models to analyse and monitor the European progress towards resource-efficiency from both a production as well a consumption perspective. Our key objectives were:
a) to build the indicator system in close connection with the most relevant stakeholders, including DG ENV, DG JRC, EEA and Eurostat, building upon their previous work5.
b) to ensure consistency in the indicator framework. As indicated above this implies consistency between production- and consumption perspectives, national, EU and global level, and industry and product group level. It implies however also aligning various resource-related indicators in different existing indicator frameworks (see Annex 1).
c) to improve data availability, particularly by developing methods that can provide time series and now-casting and create time series and now casted data in the project.
d) to improve calculation methods for indicators that currently still lack scientific robustness, most notably in the field of biodiversity, ecosystem services and critical materials.
e) to explicitly address the problem of indicator proliferation, and demands for indicators that need data that even on the medium term realistically cannot be expected to be supplied by statistical sources. Our aim is to define via scientific analysis the most relevant set of indicators giving mutually independent information, and to show which shortcuts in (statistical) data inventory can be made without significant loss of quality
f) by all of the above, to provide EU policy makers an indicator set that has the highest relevance, is fully accepted, and can be handed over to and implemented by Eurostat, EEA and others with an acceptable effort. This will ensure optimal support to the mentioned policy agenda’s and contribute to a truly sustainable use and management of natural resources (in Europe and worldwide)
A review of the most relevant policy dossiers that should be supported by the DESIRE project tells us that already extensive indicator systems have been developed or are being proposed. Three points stand out from this analysis:
1. There seems almost an overload of indicators related to the different policy agenda’s, also in the field of resources. There seems room for improvement of structure and relations;
2. Most indicator systems seem to select which operational indicators to use rather pragmatically on the basis of data availability;
3. Resource efficiency indicators usually are defined at national level, usually comparing the amount of resources used and the benefits in terms of economic value added or gross domestic product.
DESIRE goes beyond this simple and well-known approach and includes the development of indicators for the impact of resource use on the resource base on one hand and for an extension in measuring the benefit of resource use in terms of alternate measures of welfare on the other hand. In addition, a consistent calculation methodology, based in the framework of national economic accounts, is established and applied to take comprehensively relate resource use to aspects of wellbeing. By this, DESIRE will also ensure that the indicator concept is consistently applicable at the macro, meso and micro level. We propose an indicator framework organised along four dimensions:
1. Use an extended description of the economic system based on UN SEEA. At the core is an extended description of economic processes of resource extraction, transformation, residual production and value added. This description is based on the UN System of Economic and Environmental Accounts (SEEA) and its European implementation (NAMEA). Economic processes of production, consumption and investment are described in monetary terms in input-output tables (IOTs), while satellite accounts track physical processes of resource extraction, land use and emissions by sector. We propose to link these national accounts adding trade information to obtain a picture of the global resource implications of national decisions. We furthermore propose to extend the framework to address the life cycle of small mass flows of valuable ‘critical’ material resources in the economy with tracking their stock and flow.
2. Use the DPSIR concept to relate indicators along the cause-effect chain. The Resource Efficiency Flagship adopts a “wide” definition of resources, including environmental health and ecosystems. The relevant resources considered in DESIRE are ecosystems and their contribution to ecosystem services and biomass production, addressing land, soil quality, water and biodiversity as important preconditions for a healthy state of ecosystems; metals and other valuable minerals; a stable climate, and other environmental conditions conducive for human health. DESIRE adopts the Driving Forces-Pressures-State-Impact-Response (DPSIR) concept to relate different indicators along the chain of cause and effect, and that relates pressures caused by resource use to impacts on the natural resource base (which includes impacts on the environment in broad sense)9. Knowledge of impact mechanisms and ecosystem service provision will be utilized to systematically relate different indicators and select among them. The DPSIR chain also provides guidance for approaches how different indicators can be aggregated, ultimately to single score indicators.
3. Explore alternative reference values for success of the economic system. To assess resource-efficiency, it can be considered to use alternatives for economic references such as value added and Gross Domestic Product. Candidates that may provide more meaningful measures of social objectives, such as prosperity and well-being, include the Human Development Index, perceived happiness, or economic references corrected for external costs etc. At sector and final consumption level one could consider e.g. performance based measures (e.g. for education, health care and elderly care).
4. Use concepts like the Policy Cycle, RACER and scientific cluster/structural analysis to select the minimal needed and most coherent indicator set. There is a need to avoid having too many indicators or indicators that need difficultly available data, whereas at the same time policy needs to be well informed. It hence makes sense to use the Policy Cycle as completeness check for the indicator system by posing the question: does the indicator system – as far as reasonably can be expected - provide the relevant information for different steps in the policy cycle? At the same time, concepts such as the RACER criteria (‘relevant, acceptable, credible, easy and robust’) and scientific cluster and structural analysis can help to identify which indicators are less relevant.
Building the aforementioned, ideal indicator structure for resource efficiency faces a set of challenges. The most important ones are:
1. There is a large variety of indicator systems and suggested list of indicators that relate to the resource efficiency agenda. This needs alignment. Further, quite some indicators (e.g. GDP/DMC, emission data, impact assessment indicators such as GWP) are provided for countries as a whole (rather than industry sectors or final consumption categories) or describe a state of or impact on the environment without that always a clear causal link to economic activities/drivers and pressures is made (e.g. biodiversity indicators). Content wise we propose using the 3 structuring elements: a common data base in EE IO format; linking cause and impact via DPSIR; including alternative reference values for success or ‘outcomes’ of economic activity. We saw however also a need for activities with an important process component to achieve this:
a. Policy-science brokerage
b. Policy analysis
c. (Joint) elaboration of an initial, credible and acceptable indicator framework
2. For a variety of indicators, there is a reasonable agreement how to calculate them, but data availability is still a main problem. This is particularly relevant for institutes like EEA and Eurostat. They are expected to provide authoritative information, which implies considerable quality demands with regard to the data they can use and publish. Creating data sets of ‘statistical quality’ often is difficult. We want to address this point as follows in our project:
a. Gathering high quality data in the project itself (e.g. for time series and nowcasting)
b. Where this is not possible, creating a research database (building upon projects such as EXIOPOL and CREEA) that allows for an analysis which shortcuts in data inventory could be considered without significant quality loss.
c. Performing a scientific correlation analysis and structural analysis to single out the minimum set of indicators for impacts, pressures and drivers with most ‘saying power’, which again can reduce the effort in data inventory.
3. For other indicators there is still a need to improve the scientific basis for their calculation, particularly if we want impact indicators to reflect a causal link with economic activities and drivers at sector or product level. This is particularly true for indicators that assess biodiversity impacts and impacts on ecosystem services, and to a lesser extent indicators related to the analysis of criticality of small resource flows.
4. Finally, uncertainty is a point to consider. Whereas this project has as primary mission to build an indicator system for resource efficiency and not uncertainty analysis per se, some level of attention needs to be given to this issue.
DESIRE created progress beyond the state of the art in the following main areas:
1. Time series and now casting of data, particularly related to a global Multi regional Energy/Physical/Monetary input-output model
2. Indicators for security of supply of critical materials.
3. Indicators for impacts on biodiversity and ecosystem services
4. Novel reference indicators (‘Beyond GDP and value added’)
5. Insight in the optimal ‘resource-efficiency indicator dashboard’ to define the best mix of policy options, and to analyse the progress towards resource-efficiency over time.
Project Results:
1 DESIRE indicator framework
1.1 DESIRE’s conceptual framework
The DESIRE project aims to develop an optimal indicator set to monitor progress towards Resource Efficiency in Europe, yet taking into account the relevant global perspective. Resource efficiency is about using natural resources efficiently, either in a technical sense (i.e. less physical input per physical output) or economic/welfare sense (i.e. economic or societal value generated per unit of resource). The latter implies a need to discern all possible interactions between society and the natural system in a coherent and integrated way. The concept of societal or industrial metabolism is helpful in this regard.
This concept refers to the notion that socio-economic systems require resources (materials, energy, water or land) as input in order to produce goods and services or to maintain socio-economic structures. In addition, production and consumption processes, as well as transportation, put a burden on the environment through their (metabolic) outputs such as wastes and emissions to air, water and soil. “Resources” thus address different categories and issues, all with different impacts to “environment”, i.e. climate, biodiversity, ecosystems, health, etc. Resource [use] efficiency indicators therefore need to address complex interactions between society and the environment in order to empower political action; to set meaningful targets; and to adequately monitor the [global] use of resources.
Figure 1.1: Society-nature interactions as metabolism
DESIRE captures these metabolic relations between the natural environment and society, in a framework of Multi-Regional Environmentally Extended [economic] Input-Output relations with production and consumption flows. This [MR EE-IO] framework is referred to as “EXIOBASE”.
With this framework it is then possible to assess how production and consumption impacts the natural system. The DESIRE indicator framework builds on the causal Driver-Pressure-State-Impact-Response (DPSIR) frame, adopted by the European Environmental Agency (EEA), to understand society-nature interactions. This framework helps to structure and organise indicators along a cause and effect chain. That is: drivers for resource use; that put pressure on the environment; resulting in environmental impacts; and causes changes in the state of the natural system. These insights can eventually be a trigger for responses from relevant actors. Responses that feedback on drivers and thus closes the cycle (see figure 1.2). In the DESIRE project, the DPSIR-framework is integrated with the metabolic perspective of society-nature interactions (see figure 1.3)
Figure 1.2: The DPSIR framework
Source: EEA
Drivers
The structure and characteristics of the socio-economic system, its economic processing, and household consumption patterns are considered driving forces (drivers), which are strongly shaped by the cultural, political, and economic context they are embedded in.
Pressures
Resource use and management activities put pressure on and potentially change the natural system, its ecosystems and ecosystem services and thus the underlying natural State.
Impact
Effects of pressures on the natural system are considered environmental impacts. These impacts can, for example, form a threat on human health, human well-being (i.e. a broad concept of welfare) or economic wealth. Environmental impacts could be interpreted or weighted against a certain amount or quality of natural capital stocks, e.g. planetary boundaries.
Responses
Responses are the decisions and choices made within the socio-economic system by individuals or by policy makers as a response to changes in the societal as well as natural systems with the aim to adapt to these. Examples are: tax regulations, legislation or other (thematic) policy response packages.
Figure 1.3: Conceptual framework for DESIRE’s indicator set on resource use
Source: Eisenmenger et al., 2014
The conceptual interactions between socio-economic activities and the natural environment as depicted in figure 1.3 can be integrated in an environmentally extended input-output framework (see figure 1.4).
Input-output (IO) tables originate from economic accounting. Together with its two main building blocks – supply- and use tables (SUTs) – IO-tables form the backbone of the system of National Accounts. Typically, an input-output table includes inter-industry flows in monetary units as well as flows between industries and the final demand categories (e.g. households, government spending, capital investment, exports). By that, IO tables allow for tracing goods from the extraction process through manufacturing and down to final demand and provide information on the inputs needed by an industry to provide the respective industry output. One could say, it supplies the “ingredients” for one unit of output of the manufacturing industries, either derived directly or also indirectly from other industries. From an IO table, the main economic output indicator, Gross Domestic Product (GDP) can be derived. National statistical offices from all around the world provide SUTs and IO-tables, typically at an interval of a few years.
Figure 1.4: Integration of DESIRE’s conceptual frame in a MR EE-IO framework
Source: Eisenmenger et al., 2014
The standard economic input- and output flows can be complemented with environmental extensions such as material extraction, land use and emissions to air, soil or water (i.e. environmental pressure indicators). The environmental or natural resource inputs enter the production process of a certain sector and are than further distributed via inter-sectoral deliveries until they end up in one of the final demand categories. Thus, the environmental extensions represent the resource use indicators and data, i.e. pressure indicators in absolute values. Figure 1.5 shows these relations in a multi-regional set-up.
Figure 1.5: A Multi-Regional Environmentally Extended Input-Output model
Source: Eisenmenger et al., 2014
1.2 Complementing and improving EXIOBASE
The indicator framework of DESIRE builds on previous versions of EXIOBASE that were developed with support from the EU’s Sixth and Seventh Framework Programmes, i.e. the FP6-project EXIOPOL (that delivered EXIOBASE version 1) and FP7-project CREEA (that delivered EXIOBASE version 2).
EXIOBASE version 1 comprised an extensive economic-environmental database that followed the principles of the System of National Accounts [SNA 1993] and System of Environmental-Economic Accounting (SEEA 2003), based on the International Standard Industrial Classification of All Economic Activities (ISIC and its European equivalent NACE ) and Statistical Classification of Products by Activity in the European Economic Community (CPA). EXIOBASE version 1 provides data for base year 2000.
Accordingly, in the CREEA project, the database and indicator framework was refined according to the new accounting approaches of SNA 2008 and those that were proposed for inclusion in SEEA 2012 regarding four priority areas: water, waste and materials, forestry, and climate change issues.
Version 2 of EXIOBASE provided data at an at an unprecedented level of detail in terms of sectors, products, emissions and resources with coverage of 43 countries; 27 EU countries and the largest non-EU economies (summing up to 95% of the global GDP) with over 150 smaller countries combined in 5 ‘Rest of the World‘ groups by continent. EXIOBASE version 2 provides data for base year 2007.
Moving-on from here, DESIRE complemented and improved the database with new statistical information and expanded the indicator framework by adding the 28th EU Member State, Croatia, and by compiling time series, including now casting of recent years for which not all data is available yet from official statistical publications. With these improvements EXIOBASE version 3 can now be used to analyse consumption and production based accounts as well as stressors embodied in imports and exports, including developments over time.
EXIOBASE version 3 has the following characteristics:
EXIOBASE 3
Base-years 1995 – 2011/16 *)
Products 200
Industries 163
Countries 44 (28 EU member plus 16 major economies)
Rest of the world regions 5 (Europe, Asia, Africa, America, Middle East)
Water accounts 194 (Water blue and green per source, including final demand)
Material accounts 189 (Energy products, including final demand)
222 (Used extractions)
222 (Unused extractions)
Land accounts 14 (Including build up land for final demand)
Social accounts 14 (Employment per skill level and gender; vulnerable employment)
Emissions 28 (from combustion including final demand)
410 (non-combustion)
3 (HFC, PFC, SF6)
*) Historic time series for up to 2011, the rest of the years has been now-casted.
Source: Stadler et al., 2016
1.3 DESIRE’S Resource Efficiency indicator framework
DESIRE’s indicator framework differentiates three main categories of indicators: resource use, resource efficiency and environmental impacts (the columns in table 1.1 below). In addition the framework comprises six flow types of resource inputs or (metabolic) outputs (as indicated by the rows in table 1.1). In accordance, figure 1.7 shows how these Resource Efficiency indicators are positioned in the multi-regional EE-IO model.
Table 1.1: Environmental indicators in the DESIRE indicator framework
Figure 1.7: Positioning of Resource Efficiency indicators in the EE-IO model
Source: Eisenmenger et al., 2014
Pressure indicators, related to direct society-nature interactions, i.e. resource use indicators in absolute physical terms, are a good starting point to measuring resource efficiency. These are input flows of materials, energy, water and land to intermediate use and final use in the IO-framework, as well as output flows thereof of waste and emissions.
Accordingly, resource use has to be complemented by indicators that capture the effects on the natural system (impact indicators) as well as effects on the socio-economic system. The latter are commonly referred to as resource efficiency indicators. In DESIRE’s indicator framework two types of resource efficiency indicators are distinguished: resource efficiency indicators in relation to the economy and resource efficiency indicators in relation to services to society.
Resource efficiency indicators in relation to the economy (production and consumption based) can be measured against GDP or value added, and hence directly be linked to economic transactions in the IO-framework. Efficiency indicators in relation to services to society, on the other hand, often need (detailed) auxiliary data from sources outside the IO-framework. Services to society concern functional outputs rather than economic outputs, such as adequate housing, space heating, nutrition, etc. By their nature these outputs have a ‘beyond GDP’ connotation and can sometimes best be expressed on a rather micro level. An example of such a resource efficiency indicator is the required energy consumption for space heating per m2 of dwelling floor space. In the DESIRE project a work package was devoted to test opportunities for using novel reference indicators ‘beyond GDP and value added’.
The third main indicator category in the DESIRE framework, environmental impacts of resource use, put socio-economic pressures in relation to the natural state, and inform about both quantitative and qualitative aspects of natural capital stocks. They can inform about the state of stocks and changes thereof over time, e.g. depletion, degradation, climate change, biodiversity loss. For these type of indicators a satellite account of natural capital stocks need to be connected to the IO-framework. This is the lower block of data indicated in figure 1.7.
Table 1.2 summarises the rationale behind the indicator types and their position in the DPSIR-framework. It covers the pressure indicators in absolute values (columns in yellow) as well as the “resource efficiency” indicators (columns in yellow as well) which result from relating resource use to macro-economic added value (e.g. GDP) or macro-economic well-being indicators. The columns on the left cover the socio-economic system. Additionally to the resource efficiency indicators that link to socio-economic macro indicators, resource efficiency can be analysed as the relation between resource use and specific societal services provided (column in red). This covers all the activities that directly deal with biophysical flows, however no longer structured along the macro-economic IO matrix but along societal services. The socio-political responses (column in pink) cover the social, political, economic, or cultural responses.
Table 1.2: DESIRE indicator types and their position in the DPSIR-frame
Source: Eisenmenger et al., 2014
The green columns on the right side cover the efficiency of resource use in relation to the environmental impacts on the natural system in two dimensions, quantitatively and qualitatively. These environmental impacts are structured along the commonly used environmental threats (boxes on the very right). Thus, the environmental impacts do not follow the IO structure, just as the socio-economic activities at macro level.
The general idea behind DESIRE’s indicator framework is to apply a 2-level system: a limited set of headline indicators that covers resource use, resource efficiency and environmental impacts on the macro (i.e. country) level, and an accompanying second level of indicators addressing specific questions within each category.
1.4 A set of readily available Resource Efficiency indicators
EXIOBASE contains the physical layers energy, water, materials and land, which can be tracked as resource inputs to the economic production process. In addition there are various material extensions that provide information on metabolic outputs of production and consumption processes, such as emissions and waste. EXIOBASE covers:
• Greenhouse gas emissions, in kilograms of CO2, CH4, N2O;
• Polluting emissions: SOx, NOx, NH3, CO, Benzenes, Indeno (1,2,3-cd) pyrene, PAHs, PCBs, PCDD_F, HCB, VOCs, PM10, PM2.5 TSP, As, Cd, Cr, Cu, Hg, Ni, Pb, Se, Zn, SF6, HFCs, PFCs);
• Nitrogen and phosphorous emissions to water;
• Domestic material extraction of various types of crops, wood, metal ores, industrial and construction minerals & fossil fuels (differentiated in used and unused extracted materials);
• Withdrawal of blue water, differentiated by the manufacturing, electricity production and domestic use sector;
• Green and blue water consumption, differentiated by use category, for various types of agriculture, livestock, manufacturing, electricity production and domestic consumption;
• and land use (by different types of arable land, pastures and forests).
Of the full set of possible indicators that are directly calculable from EXIOBASE, some might score differently on criteria that are relevant for their uptake and implementation in policy making or for monitoring purposes, i.e. on so-called RACER-criteria: Relevance, Acceptability, Credibility, Easiness and Robustness. The proposed indicators in the 2-level indicator system are therefore tested along the lines of these RACER-criteria. The results of this exercise are shown in Annex 1, indicating the relevance of indicators and possible needs for further development. Table 1.3 (below) provides a sample of indicators that can directly be calculated.
From IO-tables coefficients of industry requirements in monetary terms can be calculated (i.e. direct requirement coefficients and Leontief multipliers). Similarly, these factors can be calculated for environmental impacts at two levels of detail:
Scope 1: Direct environmental interventions
Direct environmental interventions (for example, air emissions or material extraction) are available for each industry without further calculations from the table of environmental extensions (i.e. the dark-green boxes as shown earlier in figure 1.5 and the “resource use part” in figure 1.7). One can sum the extensions for all industries in the region, plus the interventions reported under final use (= use phase interventions), to derive the country total. Scope 1 is a territorial perspective, expressing environmental consequences, which originate within a country’s or region’s territory.
Table 1.3: Examples of environmental indicators calculable with EXIOBASE
Scope 2: Total (direct plus indirect) environmental interventions
The main advantage of IO-models applied to environmental issues is that they allow calculating the total direct plus indirect effects for all products and all sectors, also those with very complex supply chains, as the whole economic system is included in the calculation system. IO-analysis thus avoids so-called “truncation errors” often occurring in coefficient-based approaches, i.e. errors resulting from the fact that the whole complexity of production chains cannot be fully analysed based on Life Cycle Assessment (LCA) approaches, where as a consequence certain up-stream chains have to be “cut off”. IO-analysis thus avoids imprecise definition of system boundaries, which is one key advantage over other approaches. IO-models also avoid double counting, as different supply chains are clearly distinguished from each other in the monetary input-output tables. Thus, a specific resource input can only be allocated once to final consumption, as the supply and use chains are completely represented.
However, IO-analysis also contains some disadvantages. Whereas LCA-type approaches are able to cover both upstream and downstream environmental effects, IO accounts only for upstream inputs to the production processes and ultimately to final consumption. Environmental consequences from the use-phase are only given in a single table entry, at the intersection of the final use column and the environmental extension row (see the blue square in the lower right corner of figure 1.5). For CO2 emission for example, this single number includes many types of direct emissions like those from private car use or the emissions related to heating our homes and drinking our soft drinks. Hence, typical use-phase oriented indicators, such as the “per capita CO2 emissions from the housing and infrastructure sector” (i.e. part of the EEA core set of indicators) are difficult to derive from IO tables directly. In order to calculate these indicators, the vector of private consumption would need to be split up by consumption categories, for example following the COICOP classification, which disaggregates consumption by purpose (e.g. food, housing, transport, communication, etc.).
It shall be emphasised that EXIOBASE is different from other MRIO databases (such as GTAP or EORA) because it contains physical layers at the industry level. Therefore, EXIOBASE contains data on direct physical imports and exports, which allow calculating material flow-based indicators, such as DMC, which would not be possible to calculate with other MRIO systems without physical layers. It also contains detailed waste data, which allows calculating specific indicators, such as recycling rates.
1.4.1 Results of resource efficiency indicators over time
For a full set of resource efficiency indicator results over time and details on underlying data and documentation of all readily available indicators we refer to the report of Stadler et al, 2016 (Desire deliverable D9.1).
We would like to stress that working with DESIRE’s MR EE-IO framework and indicators involves a large amount of data, such that some programming skills are required to analyse the detailed datasets. For the purpose of easing these analyses an open source tool, Pymrio, has been developed. This tool has been used to calculate production and consumption based accounts for each stressor per product and per country.
These results can for example be used to assess decoupling of economic growth from environmental pressures. Doing such an analysis for the EU 27 countries shows that, taking a production based account, the European Union appears to have achieved decoupling (see figue 1.8 top). Taking, however, the consumption based perspective (figure 1.8 botom), material usage and fuel combustion grow in the same rate as the economy. In addition, CO2 emissions, water use and land use are only recently decoupled from economic growth.
Figure 1.8: Time series of production- (top) and consumption based accounts (bottom) of the European Union.
Indicators show the resource efficiency as impact per capita.
Source: Stadler et al., 2016 (calculations based on EXIOBASE v3.2.3)
2 Indicator development: novel indicators
DESIRE aimed to push the state of the art for EE MRIO compilation and calculations. For that purpose, besides developing approaches for now-casting EE MRIOs (see Stadler et al., 2015), DESIRE’s research efforts also focussed on more experimental indicator development. This included efforts to develop or improve resource efficiency indicators in the domains: Critical Materials, Biodiversity and ecosystem services, and Novel reference indicators ‘beyond GDP’, as well as efforts to linking them to the MR EE-IO framework. In general, linking pressures to state and impacts in these fields proved to be complex sometimes, as it involved manifold or non-linear relations. This made determining causal relations between society-nature interactions and final environmental impacts to the least ‘challenging’. In this chapter we report the key results per each of the three, what we call ‘novel indicator’ domains, and explain the extent to which these results could be linked to the EE-IO framework.
2.1 Critical material indicators
The objectives of DESIRE’s work package on critical material indicators (WP6) were to specify and define a number of relevant critical material indicators and to link these to the IO-framework. Regarding critical material indicators, economy wide aggregated indicators as proposed in DESIRE’s ‘core’ indicator framework, are not suitable as most of the energy, water, land, carbon and other emissions indicators do not refer to the criticality problem of materials. In addition, the indicators on waste and materials in DESIRE’s indicator framework are on a still too aggregated level to signal criticality problems of material substances as fraction of products.
After the first stage of DESIRE’s research activities on critical materials a list of indicators were considered relevant.
It was suggested to pursue an approach based on material flow assessment, linked with input output (IO) tables. Furthermore, it was proposed to develop a link between IO- tables and elaborate waste statistics, which are highly relevant for quantifying recycling options. It was concluded that a hybrid mixed unit (i.e. combined physical- and monetery information) model is in principle suitable for modelling the flows of critical materials in a MR-IO framework, and to determining flow-based resource efficiency indicators. However, at the same time it was acknowledged that availability of data could be problematic to accurately build such a hybrid IO-model. Research efforts in DESIRE therefore also focussed on possibilities to overcome data challenges.
There are numerous challenges involved with connecting material flows and critcality assessments to economic accounts. Some of the most practical challenges are:
• The detailed industrial production, Europroms statistics (PRODCOM), do not distinguish between primary and secondary commodities (or the items are difficult to classify as such). Moreover, the Europroms statistics seem to be poor in monitoring waste and scrap flows.
• Incompatibility between IO-tables from the System of National Accounts and data on material flows. This problem is most apparent when one tries to connect detailed Substance Flow Analysis (SFA) to the product and sector classifications of input-output tables (the latter are often much more aggregated). Other incompatibility issues arise around: unit of measurement (kg of substance vs. monetary units), coverage of stocks (which are absent in IO-tables) and continuity over time of the model (the IO table being strongly bound to an annual set-up).
• A specific challenge with critical materials is to specify quantities of materials in specific products and accordingly to bridge these to the product classification of supply- and use- or input-output tables. In fact, large uncertainties are introduced by the need to have more detailed information on the critical material content in (most often non-ferrous metal) products. Not only is this information rarely available, also several conversions between monetary and physical layers and assumptions of homogenous products have to be made.
As proof of concept for the ‘dynamic Technology-Hybridized Environmental-Economic Model’, several case studies have been conducted within DESIRE’s work package on critical material indicators. The scope of the substance flow analysis case studies was for practical reasons limited to the regions “EU27” and the year “2007”. Case studies are performed for Tantalum (Ta), Indium (In), Neodymium (Nd) and several steel alloying elements Chromium (Cr), Manganese (Mn), Molybdenum (Mo) and Vanadium (V). All these case studies had the same sequence of methodological steps.
In order to derive critical material flows through Europe the EUROPROMS statistics (EUROSTAT Prodcom statistics) are used to first determine the apparent consumption of critical material-containing ores, materials and products. Secondly, a review of critical material concentrations in those products is performed. Combining these data and performing some additional data processing steps yields the apparent consumption of critical material in various European ores, materials and products, which were then categorized into production stages to result in a highly detailed flow-diagram of critical material in raw materials, semi-finished products as well as in products for final consumption (i.e. “Sankey diagrams”). The scope of the case studies within DESIRE’s WP6 was limited to the compilation of Sankey diagrams.
It is acknowledged, however, that as next steps the apparent consumption of end uses of critical materials in products can be used as input for a waste assessment model. With this model it would be possible to assess the expected future critical material recycling potential from consumer wastes by assuming Weilbull life-time distributions.
The results of the case studies (i.e. the Sankey Diagrams) for Indium, Tantalum and Neodymium are described in detail in the deliverables.
2.2 Biodiversity indicators
Ultimately, the aim with DESIRE’s indicator framework is to show the impacts that environmental stressors cause. To do so, the stressor results need to be characterized into various impact categories. One separated set of characterization factors was setup for assessing the impacts on biodiversity and ecosystem service functions. The research efforts in DESIRE focussed on the impacts of land use on biodiversity. In Marques et al. (2015), based on a literature review, it is explained that land use change is currently one of the main drivers of biodiversity loss in terrestrial ecosystems. Habitat loss and habitat degradation affect more than 80% of globally threatened mammals, birds, amphibians and plants.
Within DESIRE’s work package on biodiversity, three indicators have therefore been developed which can be coupled to the EE MR-IO framework:
(1) Bird species lost;
(2) Cumulative extinction risk of carnivorous mammals; and
(3) Carbon sequestration foregone.
The development of these indicators was dependent on spatially explicit information of land use by sectors included in EXIOBASE. That is because impacts on biodiversity and ecosystems can be very location specific (for example clearing a km2 of Amazon forest will represent very different impacts on biodiversity than clearing a km2 of an agricultural field). For the characterization of pressures the production and consumption accounting scheme from the MR EE-IO tables is used.
For each of three indicator types we summarize the conceptualisation, methodology and key results are described in detail in the deliverables.
2.3 Novel reference indicators beyond GDP
DESIRE’s work on novel reference indicators started from the acknowledgement that GDP and value added are not bad indicators for economic output as such, yet not always sufficiently connected to the input of natural resources and outputs in relation to services to society, i.e. a broader perspective on welfare. Resource efficiency ‘beyond GDP’ requires a link between natural resources (as input) to human well-being (as output). A challenge is than that ‘quality of life’ or ‘human well-being’ aspects have no one-to-one relation with production and consumption. There are much more, and more complex, relations to take into account.
Consumption of goods and services is only one of many mechanisms that contribute to human well-being. It is therefore crucial to acknowledge that the backbone of DESIRE’s indicator framework, the MR EE-IO model, focusses on environmental aspects of economic production and consumption transactions only. As for the part consumption is relevant to consider in a broader welfare context, there are no straightforward ways to linking consumption expenditures to human need satisfaction. Moreover, because consumption of a single product or service can satisfy several human needs at the same time. Another limitation of assessments through an IO-framework is that it represents flows in a single year, while past investments or changes in quality and quantity of stocks of different types of capital are important determinants for some aspects of human well-being, such as health or nature’s recreational values.
As alternative, Freyling et al. (2014) therefore proposed a preliminary framework where human needs and quality of life are at the core, and it is assessed how the output of economic activities that use natural resources satisfy human needs. This framework was further developed by Usubiaga et al. (2015), of which we provide a brief summary shortly below. The overall objective of this stream of work within DESIRE was to explore whether using alternative reference indicators (i.e. alternatives to GDP or value added) in the calculation of resource productivity or resource efficiency better reflects how the use of natural resources contributes to services to society, i.e. to human needs satisfaction. And accordingly, to conclude whether or not the use of novel reference indicators leads to different conclusions when compared to resource efficiency measures with relation to the economy.
A framework for ‘beyond GDP’ resource efficiency
The conceptual framework used in DESIRE for the development of ‘novel reference indicators’ depicts the various mechanisms through which natural capital, social capital, human capital, financial capital and manufactured capital contribute to people’s quality of life (or human well-being). This framework integrates elements from Max Neef’s human scale development and human needs (Max Neef et al., 1991; Max Neef, 1992), Ekin’s four capital model (Ekins, 1992), Sen’s capability approach (Sen, 1981, 1985, 1999), as well as the Conference of European Statisticians’ recommendations on measuring sustainable development (UNECE, 2014) that builds on previous work of Smits and Hoekstra (2011). In doing so the logic behind the System of National Accounts (SNA) and the System of Environmental-Economic Accounting (SEEA) is followed. Figure 2.1 gives a schematic overview of the conceptual framework.
Figure 2.1: DESIRE’s conceptual frame for ‘beyond GDP’ resource efficiency
Source: Usubiaga et al., 2015
Max-Neef considers that human needs are finite, few, classifiable, and non-hierarchal (Max-Neef, 1992). Thus, needs are not dependent on time or cultural factors, they remain invariable. In contrast to needs, their satisfiers are infinite and changeable. Satisfiers go beyond the goods and services provided by the economy. They can take many forms, including that of economic goods. Satisfiers can be seen as the forms of being, having, doing and interacting that represent everything that helps us in a specific time and place to meet our needs. Furthermore, needs remain always constant and inalterable, while the ways individuals choose to satisfy them varies across cultures and time.
Max-Neef grouped human needs in two categories: existential (fulfilling a variety of inherent states of activities) and axiological (value-based), which intercept and form the matrix shown in table 2.1. While the existential category refers to activities such as being, having, doing and interacting, the axiological category covers needs such as subsistence, protection, affection, understanding, participation, idleness, creation, identity and freedom.
Table 2.1: Fundamental Human Needs
Need Being (qualities) Having (things) Doing (actions) Interacting (settings)
Subsistence physical and mental health food, shelter, work feed, clothe, rest, work living environment, social setting
Protection care, adaptability, autonomy social security, health systems, work co-operate, plan, take care of, help social environment, dwelling
Affection respect, sense of humour, generosity, sensuality friendships, family, relationships with nature share, take care of, make love, express emotions privacy, intimate spaces of togetherness
Understanding critical capacity, curiosity, intuition literature, teachers, policies, educational analyse, study, meditate, investigate, schools, families, universities, communities,
Participation receptiveness, dedication, sense of humour responsibilities, duties, work, rights cooperate, dissent, express opinions associations, parties, churches, neighbourhoods
Leisure imagination, tranquillity, spontaneity games, parties, peace of mind day-dream, remember, relax, have fun landscapes, intimate spaces, places to be alone
Creation imagination, boldness, inventiveness, curiosity abilities, skills, work, techniques invent, build, design, work, compose, interpret spaces for expression, workshops, audiences
Identity sense of belonging, self-esteem, consistency language, religions, work, customs, values, norms get to know oneself, grow, commit oneself places one belongs to, everyday settings
Freedom autonomy, passion, self-esteem, open-mindedness equal rights,
means of communication dissent, choose, run develop awareness anywhere
Source: (Max-Neef et al., 1991)
For sector-specific ‘alternative’ outcome oriented indicators EXIOBASE’s MR EE-IO model is not sufficient. In order to connect novel reference indicators to human needs, next to economic indicators on production, income and consumption, functional output and outcome-oriented indicators are needed (most often at the meso or even micro level), as well as subjective indicators. Moreover, an ideal (panel) dataset would than provide explicit information on intentions behind actual consumer expenditures, i.e. to learn which needs are satisfied with certain purchases.
With such a dataset it would be possible to calculate the footprint of the households for every need category according to their expenditure and also calculate the footprint of government consumption in those same categories. In that sense we could assess (1) resource efficiency in terms of objective novel output and outcome indicators and also (2) resource efficiency in terms of subjective well-being.
Cross comparison of such indicators would reveal the (1) efficiency of different regions to provide outcomes from resources, (2) the efficiency of different households to achieve high levels of well-being through resource use, and (3) the different narratives that emerge when comparing indicators based on outcome-oriented and subjective well-being indicators.
Satisfaction of human needs in a resource efficiency context: 3 case studies
We have tried to operationalise this framework by means of three case studies that differ in their scope and level of detail: 1.) linking products in EXIOBASE to human needs (as satisfier), which is, for pragmatic reasons, based on expert judgement rather than the abovementioned ‘ideal panel dataset’; 2.) food and nutrition systems (as satisfier for the need subsistence); 3.) housing in a resource efficiency context (as satisfier for the need protection). The first case study proved that the needs of subsistence and protection are among the most environmentally intensive. This finding formed the rationale to explore their most important satisfiers, food and housing, in more detail in the second an third case study.
2.4 Conclusions on the extent to which indicators of the three “novel reference indicators” case studies could be linked to EXIOBASE
None of the three case studies has managed to completely couple all the proposed alternative reference indicators to the EXIOBASE classification. In specific cases within the food and housing case studies, a one-to-one correspondence between EXIOBASE product groups and the indicators was possible. Nonetheless, besides these exceptions, the indicators proposed should be seen as a separate set of metrics that can only be soft-linked to the input-output model.
The first case study has proposed novel reference indicators at the level of human needs, since there is not simple correspondence between the product groups represented in EXIOBASE and Max-Neef’s human needs. Due to the complexity and interlinkages between needs and products, it was not possible to use indicators at product group level that can describe the specific contribution of each product to the need(s) it intends to meet.
In the food case study, indicators at product (food available for human consumption, food intake) and aggregated satisfier level (i.e. food as a whole: food available for human consumption, food intake, net healthy food intake, quality-corrected net healthy food intake) have been proposed.
Strictly speaking the metrics given at product level could be linked to EXIOBASE from the consumption side, but there are two main factors to consider: The first one refers to the correspondence between the primary data source (Food Balance Sheets [FBS]) used to calculate the indicators and EXIOBASE. This correspondence is problematic, since FBS provide information on food made available for human consumption, yet the classification used represents primary products. Before reaching the consumer, most food products are usually processed one way or another, which requires assumptions to link primary products to the manufactured food categories in EXIOBASE.
The robustness of the results depends on these assumptions. Second, food is not only consumed as part of final consumption activities, but also in ‘food-related services’ such as health services, education, etc. This intermediate consumption is ultimately embodied in the final consumption of other goods and services. In the case study we have not assigned the related food consumption to the end product or service in which it is embodied, but to the food product itself. Hence, the product-level indicators need careful interpretation if they are to be related to individual product groups in EXIOBASE.
With regard to the metrics given for food as an aggregated satisfier, the consumption (intermediate and final) of all food products is combined to depict the per-capita level of nutritional well-being achieved by a country’s population. The set of indicators provided at the level of human need are provided separately and could only be soft-linked to the input-output model.
The housing case study has made use of two types of alternative reference indicators: functional output and deprivation. The functional output indicators used as reference for the construction phase (new dwellings built in terms of number and m2 floor space) can be attributed entirely to the construction sector. However, the activities of the construction sector go beyond building new dwellings (e.g. maintenance work and construction of non-residential infrastructure such as industrial facilities or roads). For this reason, there is not a one-to-one correspondence between the alternative indicator and the product group in EXIOBASE’s MRIO model.
As for the use phase, final consumption of energy by households is considered a good reference. In the case study we have used climate-corrected final energy use by purpose (e.g. heating, electrical appliances, etc.) from an external source. In itself indicators on physical amounts of energy use by use purpose are already effective monitoring instruments to assess developments in resource efficiency. There is future research potential to combine these kind of data with the existing EXIOBASE data classification by energy product. In that way footprints of type of energy used can be connected to the use purposes, which would increase the relevance of the functional unit of measurement even more.
The alternative indicators proposed following the deprivation approach relate the functional output indicators to the amount of people who have structural problems or are not able to keep their homes warm. The functional output indicators cover the material footprint or energy consumption of the whole population, while the deprivation indicator refers only to a fraction of the population. Therefore, the latter cannot be linked to the EXIOBASE product groups as an add-on item, but has to be considered separately and interpreted carefully.
2.5 Conclusion on DESIRE’s novel indicators: ready for uptake?
In DESIRE’s more experimental indicator development, just as in the ‘core framework’, we tried to causally link pressures stemming from socio-economic and nature interactions to the state of the environment and environmental impacts. In this subsection we generalize the lessons and conclusions from our development of resource efficiency indicators in the three ‘novel indicator’ domains: critical materials, biodiversity, and novel reference indicators ‘beyond GDP’.
Indicator concepts and methodology
A commonality of DESIRE’s research efforts in the novel indicator domains is that they required highly detailed information and data, beyond the scope of environmental extensions and the economic IO-framework as included in EXIOBASE. All three novel indicator domains therefore had to follow a case study approach, or in case of biodiversity impacts, had to set a very clear scoping boundary (i.e. bird species loss, mammals extinction risk, carbon sequestration forgone related to a maximum of 16 out of the 163 EXIOBASE production sectors.
Biodiversity is the only novel indicator domain where an appropriate link with DESIRE’s EE-IO framework could be made, in particular for ‘land use intensive sectors’, i.e. agricultural production and forestry. Environmental pressures of production and consumption are, through the link with land use and biomes/species habitats in particular geographical locations, characterized with the EE-IO framework. In addition, only for crop growing and harvesting in the year 2000, explorative calculations were executed on carbon sequestration forgone.
Regarding critical materials the required high level of detail for appropriate substance flow analysis made it a labor intensive as well as methodological challenging task to make sound linkages to EXIOBASE’s full EE-IO framework. Rather than full time-series analysis, pragmatically operationalized case studies on EU-level (EU27), for a single year (2007) were performed.
Regarding the research on novel reference indicator ‘beyond GDP’, just as in the critical materials domain, dedicated case studies needed to be executed to address specificities of resource use and environmental impacts from a human well-being and human needs perspective. None of three novel reference indicator case studies was successful in making a full coupling to the EE-IO framework of EXIOBASE. The primary reason being that there is no one-to-one relation between production and consumption and a broader perspective on welfare and human well-being. EXIOBASE as such is therefore not sufficient for resource efficiency assessments based on other references than GDP or value added.
In line with the findings for critical materials and impact on biodiversity, the general conclusion is that DESIRE’s indicator framework could not be complemented with a full set of alternative reference indicators for resource efficiency assessments for all products and sectors included in EXIOBASE. In the next subsection we will elaborate on the results, the involved uncertainty and what this means for indicator uptake.
Results, uncertainties and considerations for indicator uptake
The critical material case studies for Indium, Tantalum and Neodymium resulted in highly detailed flow diagrams of semi-finished products as well as final consumables. The results are especially informative as ‘standalone’ case study output as each critical material has its own specific use applications and hence impacts of supply risks etc. It is more difficult to draw generalized conclusions on overall resource efficiency. Particularly the Tantalum case study delivered a novel insight (i.e. its importance in hard disks). The Neodymium case, on the other hand, seems to highlight the level of uncertainty involved in the analyses’ methodology (a potentially large overestimation of European consumption when compared to absolute numbers from literature sources).
The work on biodiversity impact indicators proofed that the level of detail in which ‘land use intensive’ economic sectors are covered is crucial to appropriately link economic processes to regional specificities such as species rich biomes and species’ habitat preferences. EXIOBASE offers a good framework to track and trace the impacts of production and consumption on biodiversity, however the method depends on availability of good quality spatially explicit auxiliary data on land use changes, biomes and habitats, as well as stocks of vegetation for carbon sequestration. Uncertainties primarily relate to availability and quality of these auxiliary data, in addition to assumptions that had to be made on e.g. the grow back potential of vegetation for carbon sequestration. Especially for the latter, the carbon sequestration results should be considered explorative. For the indicators on bird species loss and extinction risk of mammals, uncertainties arise through averaging impacts on local grid level to country level before ‘footprints’ can be calculated with EXIOBASE.
Albeit the methodological uncertainties, the first results do provide plausible insights in the causal relation between, for example, agricultural production in, and harvesting of wood from, particular species rich biomes and how this affects other country’s ‘footprint’ of species loss or extinction risk through international trade. Such insights are valuable and can be used immediately in narratives. On the other hand, it might be more difficult for a receiving audience to attach a meaningful interpretation to the indicator results. For example, the average European citizen might not easily understand in what way they can counter any negative impact on e.g. bird species or mammal extinction risks abroad. This probably underscores the need for a good accompanying narrative in relation to the envisaged purpose with the indicator (a topic that we will further address in chapter 4).
Although contextualization of novel reference indicators to calculating resource efficiency is needed, the results of the three different case studies have in common that alternatives to monetary references such as GDP or value added were found to provide relevant, sometimes even more meaningful, insights.
In the food case study we have proposed four indicators (food available for human consumption, food intake, net healthy food intake, quality-corrected net healthy food intake) that describe the role of the different components of the food and nutrition system in meeting the need of subsistence. Compared to monetary indicators each of these metrics has advantages and disadvantages from a RACER perspective. Among the four proposed, only food available for human consumption is provided by a recognised source (FAO). Hence, food available for human consumption has more credibility than the rest, since its methodology is well established. Conversely, this metric does not sufficiently capture the contribution of food to nutritional well-being, since it leaves out key factors such as food waste at consumer level, overconsumption or dietary quality.
For instance, food waste is considered in food intake indicators, overconsumption in net healthy food intake and dietary quality in quality corrected net healthy food intake. Therefore, the relevance of the latter is higher than that of the others (and probably more acceptable in this context), yet it requires making more assumptions that negatively affect its robustness and credibility. The involvement of relevant institutions such as FAO, national statistical offices as well as other relevant stakeholders in methodological and data gathering activities would increase the robustness and credibility of the indicators, and eventually improve acceptance in the policy arena.
As for the work on housing, the alternative indicators selected for the construction phase (number and m2 of new dwellings) might be more relevant than monetary indicators in criteria such as relevance, which improves its acceptability. This comes at the expense of credibility due to the necessary assumptions to split the footprint, value added and final consumption expenditure of the construction sector between the activities related to dwellings and other construction work. For the use phase, the functional indicator selected (households resource use by purpose, e.g. energy use for space heating) is a clear improvement over monetary metrics. Especially in the field of energy, this type of monitoring is relatively well established and is used to inform energy policies. The latter can rely on other data sources than EXIOBASE, i.e. data from the Odyssee-Mure project.
Our overall conclusion is that it seems to be clear that using alternative indicators as reference yields different results than when using monetary metrics. Given that the latter could potentially be misleading when used in the wrong context, this type of work should be further encouraged to eventually develop better metrics of well-being in general, and resource efficiency in particular. For now, due the level of uncertainty involved, the first results should be used with some caution.
3 Indicator development: optimal set of resource efficiency indicators
After completion of the EE-IO time series, the database could be used for systematic analysis. In this chapter key results are described of: 1.) analyses on driving forces behind resource efficiency indicator results (to help identify the main ‘driver’ indicators); 2.) calculation methods and possibilities for MR EE-IO model simplification; and 3.) a systematic analysis of options to minimise the indicator set to a small ‘optimal’ set.
3.1 Driving forces behind Resource Efficiency indicator results
Resource-efficiency indicator results give information on the overall environmental impact, but are not able to capture the driving forces behind resource-efficiency indicator scores. This leaves policy makers with knowledge of how well a socio-economic system is performing, but with little (quantitative) knowledge of why the system is performing as observed. Unravelling the mechanisms that drive performance is considered essential in designing effective policy to attempt to influence future impact.
Drivers of impact can be seen from a number of perspectives – firstly, the temporal aspect where drivers of change in indicators are identified such that impact is analysed over time, and with respect to different socio-economic variables. Secondly, the life-cycle or supply chain perspective, where key actors in the supply chain of a good or service are identified that drive impact. An example here is the understanding of construction requirements used within the provision of services by the services sector. If we shift to a service based society, how dependent do we remain on bricks and mortar? Systematic analysis of indicator results based on the EE-IO model is possible using structural path analysis – where impact pathways are analysed and ranked based on contribution to overall impact. Finally, a combination of the temporal and structural aspects can be integrated into a single analysis by breaking down change into several key parameters of policy interest.
In DESIRE’s WP9 (task 9.2) indicators were broken down into the factors population growth, population affluence, changing consumption patterns, changing industrial production, changing trade relationships and changing resource efficiency were explored. For all details of this structural analysis of drivers we refer to Wood et al., 2016.
Globally, environmental impacts are growing. Resource efficiency indicators, however, show different trajectories depending on the perspective taken. Production based resource efficiency indicators (the impact in a certain region) generally show resource efficiency improvements for developed countries, and from a decoupling perspective, we thus see decoupling occurring for production based impacts. From a consumption perspective (the impact embodied in final demand in a certain region, utilising supply-chain analysis) we see a very different picture. Whilst the main regions (EU, US, China) are still generally decoupling, the rate of decoupling is greatly reduced, and turns negative in some cases.
At the country level, especially for wealthy countries and most evident in the EU, we see a much greater rate of negative decoupling (that consumption based impacts are outpacing economic growth). When we look at what is driving this upward growth in resource use, we consistently see the impact of population (small steady upward driver), of affluence (strong usually upward driver), and the impact of trade (moderate upward driver) – both to intermediates and final consumers. In terms of products and supply chains that are driving this growth, we see the effect of the construction activity as being one of the key sectors, particularly across material and greenhouse gas indicators. Downward drivers include both impact intensity (the ratio of impact per unit output) – where we see that this has had a strong downward effect on indicator scores over time; and final demand mix – which has had a weak negative impact on indicator scores over time.
Looking at the drivers of this increased consumption based impacts relative to GDP, we can conclude the following strong trends: 1) there is little evidence of the increased impact being due to changing types of products consumed; 2) construction is the only main exception in terms of product mix – it is also the most relevant activity in terms of a number of environmental impacts and has seen the greatest change over time; 3) trade has shifted strongly from developed to developing countries; 4) the overall impact embodied in trade has greatly increased, in line with the increased volume of, but also due to increased impact per unit of trade – trade has shifted to less efficient producers.
3.2 Optimal calculation methods and level of detail
A fundamental challenge in indicator development by European institutes like EEA and Eurostat is the consumption-based perspective. In this perspective, in essence all impacts along the (global) value chains related to European consumption should be taken into account. This implies that insight is needed in the ‘pollution and resources embodied in trade’. Through European Commission regulations it can be ensured that high quality statistical data are available for EU-member states. However, for European institutes, it is much more difficult to ensure that data in the same quality is available for global trade partners. At the same time a full and harmonized (e.g. in terms of international trade and the level of detail) Global Multi-regional Environmentally Extended Input-Output tables with harmonised trade data are needed to assess the ‘pollution and resources embodied in trade’.
Yet, from the perspective of official statistics, ‘science-based’ global MR EE-IO databases such as EXIOBASE version 3 are not preferred for reasons of deviations from original official statistical publications. Trade linking and further detailing of national IO-publications is unavoidable for global environmental impact assessments, however. Within DESIRE’s WP9, Tukker et al. (2016) have therefore analysed how national statistical institutes and EU statistical services can be provided with relevant information that may be acceptable to be used. For this purpose the following research activities have been carried out: 1.) a review of methods to calculate pollution embodied in trade; 2.) a review of various global MR EE-IO tables with listing of their pros and cons; and 3.) a review of uncertainties related to calculations based on MR EE-IOs. For the full details of these analyses we refer to the report of Tukker et al. (2016). We only briefly summarise the main findings below.
Calculation methods
Tukker et al. (2016) found that there is a clear need to cover full international value chains to optimally assess footprints, pollution and resource extraction. Global multi-regional environmentally extended input-output databases seem to be the best source to base calculations on as they not only cover full value chains but also are consistent between the production and consumption perspective. The latter meaning that total emissions and resource extraction by all economic sectors in all countries equal the footprint of global final demand. In contrast, the often used Domestic Technology Assumption (DTA), where it is assumed that imports are made in the same way as domestic production, can lead to erroneous results since production technologies and hence associated environmental impacts can actually vary quite significantly across countries. This is especially apparent in imports of small countries, which often have a fundamentally different production structure than the import’s country of origin. The DTA (including the slightly better price adjusted DTA) can therefore best be used as a ‘last resort calculation method’.
Deeper analyses of calculations based on global MR EE-IO databases showed that differences in allocation principles, definitions and data sources for extensions matter for the footprint results and associated level of uncertainty. The former relates to using a residential instead of a territorial approach. The residential approach takes a global consumption perspective and herewith accounts for all activities/emissions and resource uses in individual countries along the global value chain (including e.g. fuel bunkers). The analyses proofed that different allocation mechanism cause fundamental differences in footprint outcomes. The analysis further showed that differences environmental extension datasets (on e.g. national industrial emission totals) is a cause of uncertainty and differences in calculated country footprints. We will elaborate on this below.
Level of detail
Regarding the level of detail of global MR EE-IO tables, Tukker et al. (2016) concluded that for economic analysis purposes an aggregated sector structure (such as the 30 sectors in the OECD/WTO’s Trade in Value Added database) is quite appropriate. For the calculation and analyses of carbon footprints, aggregated EE-IO tables with sector detail of 30-60 sectors (such as the WIOD or GRAM databases), still provide plausible results.
However, for the calculation of water, material and land footprints, a high level of sector/product detail proved to be essential. Aggregating EXIOBASE to the standard 60 products or sectors that Eurostat uses in official SUT and IO publications, led to clear changes in country footprints. When one wants to look at the environmental footprint of specific product groups, detail certainly is essential. The reason behind is that the material intensity, water intensity and land intensity of specific product categories/industry sectors varies much more than e.g. the carbon intensity or the created value added.
Moreover, the level of detail of environmental and material extensions like CO2 emissions, other emissions, resource extractions, water use and land use is perhaps even the single biggest cause of differences in calculated country footprints. Harmonization of extensions across the various global MR EE-IO databases seems to be a relatively easy option to significantly reduce uncertainty of footprint results (e.g. by using the resource extraction database as recently developed by the UN International Resources Panel). Although further work is still needed here on water, land and emission extensions, it is likely that using a harmonized source for extensions will reduce the level of uncertainty in footprint calculations with more than 50%.
Possibilities to simplify EE-IO data?
The overall conclusion of Tukker et al. (2016) is that for land, water and material footprint calculations no simplification of global MR EE-IOs in terms of a low sector resolution is possible. It is likely that detailed databases such as EXIOBASE will provide superior results over less detailed databases. For carbon footprints and particularly value added, this highly detailed sector resolution is less relevant.
3.3 “Optimal” Indicators
As final task in DESIRE’s work package 9 the potential for reducing the number of indicators for environmental assessments is examined through application of a statistical Principal Component Analysis (PCA) in combination with multiple linear regression. This methodology has been applied by Steinmann and Huijbregts (2016) to two fundamentally different datasets, both with the objective to identify a [statistical] optimal (i.e. smallest) indicator set. The rationale behind is that it might be unpractical to base decisions on product Life Cycle Analysis (LCA) or on environmental policy on a large number of indicators simultaneously. The objective therefore was to identify a limited set of indicators that is sufficiently small for efficient decision making and at the same time still covers overall environmental impact, i.e. covering both midpoint impacts of resource use and the full cause-and effect chain in damage-based or endpoint environmental impacts (e.g. impacts on ecosystems and/or human health).
To find an optimal set of environmental indicators to cover the variance in the rankings of a large number of products, the PCA is first based on a selection of 976 products and 135 environmental indicators from the Ecoinvent 3.1 database (Moreno Ruiz et al., 2013) and, secondly, on 93 impact indicators for 7589 product-sector combinations from the EXIOBASE database. The PCA has been combined with multiple regression analysis to arrive at a minimum set of indicators explaining the variance in the product ranking. In addition the extent to which four commonly used resource-based indicators (fossil energy, water, land and materials) are representative of the total variation in product rankings is tested.
Whether based on the life cycle impacts per kg of material or the impacts per million euros of consumption, strong correlations between the different indicators of impact were found. This means that there is a large potential for reducing the number of indicators.
The analysis based on the Ecoinvent database showed that 92% of the variance in product rankings is covered by only 4 out of the 135 initial environmental indicators. A set of six indicators covered slightly more of the statistical variance (i.e. 92,3%). This best set of six indicators relates to climate change, ozone depletion, terrestrial ecotoxicity, the combined ecosystem effects of acidification & eutrophication, marine ecotoxicity and land use.
In addition, the four resource-based indicators together accounted for 82% of the variance in material rankings. The results suggest that it is best to use the fossil energy indicator if just one of the simple resource-based indicators has to be selected. With an explained variance of 72.9% this seems to be a reasonably good indicator of overall impact. The explained variance can be raised to 76.8% by adding material use. Adding land use raised the explained variance to 80.1%, while a set of all four resource-based indicators, including water use, covers 82.0% of the total variance in our dataset. The water footprint appeared to be less important than the other footprints for our dataset; this is due to the fact that water consumption is related to both the energy-intensive process of electricity generation and the land-intensive process of crop production.
For the EXIOBASE dataset the 93 environmental impact indicators could be reduced to seven indicators related to freshwater and marine ecotoxicity, photochemical oxidation, climate change, acidification & eutrophication, photochemical ozone formation and blue water withdrawal. These seven indicators together covered more than 90% of the variation. Similar to the analysis on the Ecoinvent dataset, the performance of the resource based indicators was also tested. The four resource-based footprints together accounted for only 49% of the variance in product-sector rankings. This means that sets of 1 to 4 resource indicators cannot cover the same amount of variance that can be explained by one toxicity indicator. Supplementing, however, the two best resource-based indicators (energy and land) with the best toxicity indicator (freshwater aquatic ecotoxicity potential, infinite time horizon) the explained variance is increased to 74.8%.
While the optimal sets maximize the amount of covered variance, the recommended indicators are not necessarily the most preferable using additional criteria, such as the RACER (Relevant, Accepted, Credible, Easy and Robust) criteria. For both datasets there are several indicators with approximately the same amount of explanatory power. This means that alternative sets of indicators can be defined which are only marginally worse in terms of explained variance compared to the statistically preferred set of four and seven indicators proposed here.
In both cases, only three out of the four resource-based indicators seem to be of real added value. For both datasets these were the indicators of energy, land and material. This is due to the fact that the (agricultural) water consumption is strongly correlated to the land footprint, especially in the EXIOBASE dataset, making one of the two indicators redundant. Using two or three simple resource based indicators would eliminate the need for the complicated mid- and endpoint damage models, but has limited coverage of the impacts associated with toxic emissions, especially for the EXIOBASE dataset.
The overall conclusion based on the statistical analyses to test options to reduce the environmental indicator set is that the large set of indicators can indeed be reduced to a small key set, representing the major part of the variation in environmental life cycle impacts between materials and of the variation in product-sector combination in a Multiregional Input-Output model.
Potential Impact:
Towards indicator implementation
Now we have summarized the key results of DESIRE’s different work packages and converted these into conclusions, we devote this last chapter on a roadmap towards indicator implementation and options for institutionalization of indicators based on global Environmentally Extended Input-Output databases such as EXIOBASE. In doing so we elaborate on a discussion note that functioned as input to stakeholder discussions during the project’s final conference. Recommendations we took home from this last ‘brokerage and dissemination event’ are integrated in the indicator implementation roadmap.
A major accomplishment of the DESIRE project is to add a time series perspective to EXIOBASE. With the EE-MRIO dataset of EXIOBASE version 3, a powerful tool is now available for analysis of various environmental-economic relations in Europe and beyond. Databases like EXIOBASE e.g. help to provide insights about how consumption drives environmental pressures. EXIOBASE thus is a relevant information source in support of evidence based policy related to e.g. Resource Efficiency and the Circular Economy.
So far, however, the development of the database and first analyses based on it, has been primarily a scientific undertaking. The challenge now is to bridge the gap between this scientific research initiative towards uptake and implementation of indicators in policy processes. One thing that we have learned during various policy-science brokerage and dissemination events is that a more formal status of Environmentally Extended Multiregional Input-Output models is for some users an important precondition for indicator uptake. From this perspective, it is desirable when supra-national statistical institutions adopt databases such as EXIOBASE and further develop these along the lines of international harmonized standards.
On the other hand we have learned that even without such formal institutionalization, and sometimes even with indicators that still have room for methodological or data quality improvement, information from databases as EXIOBASE can already be relevant as ‘early signalling’ or ‘agenda setting’ mechanism.
In this chapter we will place the ‘core set’ of the DESIRE indicator framework, as well as the more novel indicator results, in the context of discussions throughout the project on indicator uptake and implementation. With this we aim to develop an indicator implementation roadmap in which we qualify indicators that already have a “statistical stamp” and could be taken-up in policy processes without further needs for improvement, as well as defining next steps to further improve indicators that are relevant but currently lack the full “statistical quality stamp”.
Uptake and institutionalisation of DESIRE indicators
DESIRE’s global environmentally extended input-output time series database, EXIOBASE version 3, primarily offers a rich knowledge base on which a large set of indicators can be calculated, customized to different analytical purposes or policy needs. In a discussion on potential users of EXIOBASE during DESIRE’s final conference on the 21st of January 2016 in Brussels, it was for example stated that DG JRC could definitely use the indicator framework with underlying raw data for policy-support research. Other (potential) users might have other information needs for other purposes, for which the required data and indicator quality also might differ. It is for this reason that we will report in this section how DESIRE’s indicator framework, in its current state of development and quality level, can be used for different purposes.
Indicator functions: different use purposes, different needs
What should the indicators do?
Figure 1: Indicator function and the required level of quality and detail
1. Signalling and agenda setting: plausible information on recent trends.
One of the lessons we took home from the final conference is that indicators may not have to be of perfect quality yet to signal relevant environmental-economic trends. There is a trade-off between precision and the relevance to inform ongoing policy processes. It was stated that feeding policy processes with relevant (new) information sometimes is more important than waiting for the moment when indicators are of perfect methodological and statistical quality. The latter can be taken up further during formal institutionalization processes.
During the conference a general need for timeliness of indicator results (i.e. showing recent developments based on up-to-date data) was expressed, however it is acknowledged that there often is a time lag in official statistics. In this context the now-casted data of EXIOBASE can already play a role, although there are uncertainties given that GDP is used as primary indicator for the nowcasting (for reasons of high correlation) in combination with trends from earlier data points.
We acknowledge that this methodology works better for environmental indicators related to energy and carbon and has a higher level of uncertainty for material indicators. The latter category would clearly benefit from timely available official statistics. The now-casted data are not yet usable for formal monitoring and in cases where accountability is at stake. However, to communicate trends with the idea to trigger debate on the need and possibilities to intervene with policy action, EXIOBASE’s now-casted data can be a useful information source already.
2. Communication
Early signals and trends can be communicated to policy departments, politicians and the general public, with the aim to stress a sense of urgency and to trigger debate. Easy to understand, yet credible messages are relevant for this indicator purpose. A smaller set of indicators probably enhances the easiness to digest the communicated messages. In this context, DESIRE’s indicator optimization results can come in helpful, especially as there appears to exist strong correlation between various environmental impact indicators. For the purpose of triggering debate there is no use to communicate results of different indicators that are strongly correlated. However, it is recommended to carefully select the indicators with (potentially) higher scores on RACER-criteria easiness and acceptance. We underscore ones more that the toxicity indicators that came out of the statistical optimization analysis might than not be the best choice per se. Careful selection of indicators to support the message one would like to communicate will enhance the meaningfulness of information for the targeted audience.
During a discussion at the DESIRE final conference it was stressed that choices for specific indicators to communicate, might potentially “hide” misleading messages. For example, a too strong focus on resource inputs and waste outputs (e.g. in the context of a Circular Economy) might have a risk of losing focus on carbon dioxide emissions (e.g. in the context of Climate Action). Whereas perhaps the largest “waste” of socio-economic and nature interactions are related to energy and associated carbon emissions. In the same vain it was mentioned that outcomes of Raw Material Consumption calculations might be dominated by the relatively high mass volumes of gravel and sand.
In this context it was recognized as an asset of EXIOBASE that many economic-environmental relations can be shown and communicated. There is value in communicating cause and effect chains of economic-environmental interaction to a wider audience. Probably credibility is more important here than precision to the last decimal. EXIOBASE seems to be fit for this purpose already.
3. Monitoring of progress towards policy goals and targets
A third function of resource efficiency and environmental impact indicators stemming from EXIOBASE relates more to the end of the policy cycle, where progress towards formal goals and targets is monitored. This is where formal institutionalization, stricter requirements for data quality and comprehensiveness will come into play. This is the reason why this indicator function is situated in the lower part of figure 1.
How do DESIRE’s ‘core indicators’ and ‘novel indicators’ currently fit to these purposes?
We consider the ‘core data set’ of EXIOBASE, i.e. the historical data for the years 1995-2011, covering resource use, resource efficiency and environmental impact (i.e. the resource inputs and (metabolic) outputs of energy, materials, waste, land and emissions) in relation to society-nature interactions through production and consumption (including international trade), ready for immediate uptake, fitting all three abovementioned use purposes.
We do need to stress, however, that there are deviations from official (national) statistical publications. Due the high level of detail imposed, and the need for balancing international trade, deviations from official statistical sources are unavoidable (we will elaborate on this in section 4.2). Moreover, the time series are created with the idea to arrive at plausible year-to-year changes. In the construction of time series, as many official data sources with highest level of detail as possible, are used to respect structural changes in national economies. However, balancing the system (i.e. meeting restrictive conditions such as assuring that total supply matches total use) unavoidably implied deviations from official (National Accounts) statistics. For these reasons, one can expect to find deviations from other historical statistical sources when comparing actual numbers from EXIOBASE. It is for this reason that EXIOBASE indicators, in general, do not have a “statistical stamp”. The year-to-year trends are considered plausible, however.
As already mentioned in the previous subsection, the now-casted data (i.e. the years 2012-2016), have a higher level of uncertainty because of estimations based on earlier data points and (partially assumed) correlations. For this reason these now-casted years better serve signaling and agenda setting purposes.
Particularly in relation to monitoring progress in the 5 priority areas of the Circular Economy Package, it was asked during the final conference how DESIRE indicators fit. As for the priority areas plastics, food waste, and construction and demolition, EXIOBASE offers relevant as well as plausible information. During the discussion it was said that EXIOBASE can only to a lesser extend provide information on the priority area biomass and biobased products. Related to the priority area of critical materials, DESIRE’s ‘novel indicator’ development efforts only delivered insight in production and consumption flows for a small selection of case studies, covering a single year. The results are informative, yet not sufficient to fully support monitoring purposes in the this priority area field of the Circular Economy Package.
The results of the other ‘novel indicator domains’, biodiversity and novel reference indicators (beyond GDP), are subject to methodological challenges and uncertainties related to, sometimes strong, assumptions made. Albeit these methodological challenges that indicate a need for further development, the first explorative results do provide relevant insights in the different cases.
In the case of biodiversity, the results on bird species loss, mammal extinction risk and forgone carbon sequestration clearly point to environmental impacts related to production, trade and consumption of specific sectors and products. These insights can surely be used for communication purposes. It is then recommended to accompany the indicator results with a clear narrative in order to enhance the meaningfulness for the targeted audience. Narratives are equally important when using novel reference indicators ‘beyond GDP’. We have found that other indicators than GDP or value added can be relevant and meaningful, however that good contextualization to calculating resource efficiency is needed. Due to methodological challenges and assumption made, in addition to the fact that subjective views can easily be at stake, signalling and communication with the aim to trigger debate must, for now, probably be the dominant use purposes. In specific fields, auxiliary data sources to EXIOBASE (e.g. EU-SILC and Odyssee-Mure) offer relevant information for monitoring purposes. In some cases these data can be better used as such, complementary to EXIOBASE rather than integrated.
Indicator institutionalization
From the perspective of Eurostat, which, besides provision of official statistics, has harmonization of statistical methodology as task, institutionalization of DESIRE’s results is an important condition for implementation. Regular updates of EXIOBASE are required for formal institutionalization.
The normal procedure of Eurostat is to formally ask EU member States to submit data in a predefined (hence harmonized) format. Formal data collection with regard to EXIOBASE can thus only be the final step in an institutionalization procedure. This means that attention should first go to harmonization of data and indicator concepts and methods. In this context it was mentioned during final conference discussions that a comparison of methodology, level of detail, etc. with other existing global multiregional EE-IO databases would be highly relevant. Results of such a comparison could then feed into Eurostat’s deliberations on the best statistical methodology and the way forward (e.g. in relation with ongoing processes with regard to the Trade in Value Added database, together with the OECD and WTO).
Within DESIRE’s WP9, a brief comparative analysis with other existing global MR EE-IO databases, all with their own specific strengths and weaknesses, has been carried out. The main characteristics of the currently available global MR IO databases are shown in table 4.1. The main conclusions of the comparison are that:
• the high level of product/sector detail of EXIOBASE is in particular important for agricultural-, industrial-/manufacturing- (e.g. metal) and energy-producing sectors in relation to environmental issues associated with land use, water use, or resource use.
• IDE-JETRO’s AIIOTs, in contrast, offers the longest time series (with a data point back to 1975), with a relatively detailed product classification (76 sectors). A weakness, however, is its small country coverage. On the other hand, the manual handling of data transformation enables a high level of harmonization among constituent national tables.
• EORA and GTAP discern considerably more countries than WIOD, EXIOBASE, IDE’s AIIOT or GRAM. This has important advantages in assessing impacts of final consumption that take place in relatively poor countries with a low GDP not covered in other databases (Lenzen et al., 2012). Moreover, a large separate country coverage (as opposed to a large aggregated RoW regions) is important to attribute impacts to individual countries.
• Overall, with its broad coverage of countries and varying sector detail per country, EORA seems to split up the global economy in most products and sectors and it is the only database that provides uncertainty information for its estimates.
• WIOD, to conclude, has some clear advantages with regard to institutionalization as it is the only database with a consistent annual time series in both current and previous year’s prices, as well as it is fully consistent with the National Accounts statistics which is important when a link is required to other (socio-)economic data (e.g. for productivity analyses).
Related to institutionalization of DESIRE’s results it was stated during the final conference discussions that there recently seemed to be more emphasis on expanding the level of detail in NACE financial- and other service sectors in data collection for official statistics. The comparison between EXIOBASE and other existing global MR EE-IO databases, as well as other analyses within DESIRE’s WP9 on the optimal level of detail, made clear that a high sector- and product detail is crucial for meaningful assessments of environment-economic interactions and resource efficiency. In the same context it was underscored that policy demand drives the development of official statistics and the associated level of detail that indicators can provide. There is thus a need for coordination between European Member State’s and the European Commission’s policy departments and statistical institutes in the formal institutionalization process of MR EE-IO databases as EXIOBASE to ensure a minimum required level of detail in sectors/products with high environmental-economic interactions.
Table 4.1: Review of the main Global Multiregional Input-Output databases
Recommendations on creating footprint data with a “statistical stamp”, actions and timeline
We concluded that global MR EE-IO models are the preferred calculation method for consumption based accounting of footprints, for which a high level of product sector detail is required to obtain relevant results. At the same time it should be acknowledged that the compilation of global multi-regional input-output databases requires a high level of harmonization and consolidation of often conflicting data sources. For this reason, there is no other option than to deviate (sometimes significantly) from official statistics that national statistical institutes provide. One key reason is that all imports, summed to a global total, do not match the global total of exports, whereas in reality international trade obviously is a zero-sum game. The same imbalances between imports and exports occur when bilateral origin-destination trade flows are confronted with each other. Given that national statistical institutes have a national mandate, it is one of the main reasons that the construction of global (EE) MR-IO databases such as EXIOBASE has mostly been efforts of scientific research consortia. An important question for formal institutionalization is thus how a “statistical stamp” can be attributed to databases as EXIOBASE and the calculation of environmental footprints.
The first and fundamental solution to the main problem of imbalances in international trade data, and perhaps the “Royal route”, is that all national statistical institutes (NSIs) in the World collaborate, e.g. within the context of the UN Commission on Economic and Environmental Accounts (UN CEEA), on a data exchange platform that allows NSIs to end-up with supply and use tables as well as input-output tables that are mutually consistent between countries. This, however, is likely to be a long-term endeavor, unlikely to provide results in the coming 5 years. Our recommendation for this longer term solution is that NSIs and supra-national statistical institutions make use of the experiences from global (EE) MR-IO practitioners to identify the most pressing data inconsistencies at international level.
Based on suggestions of Edens et al. (2015) a second approach is described by Tukker et al. (2016): using a “Single-country National Accounts Consistent” footprint approach. In this approach, an existing global MR-IO database will be adjusted for the single country of investigation, by using the IO-data and environmental extensions from official national statistical sources and fixing these (i.e. imposing a restrictive condition that these national totals cannot change) before rebalancing the whole global MR-IO database again. Only after this rebalancing, the footprints for the country of interest can be calculated in a way that is aligned with national accounts and other official statistics.
The main drawback of this method is, however, that plugging-in the national data in the global MR-IO database and rebalancing the model is labor intensive. Moreover, uncertainties may increase for cross-country footprint comparisons, e.g. as different extension data sources are now confronted with each other (see earlier remarks of this being a potentially big cause of uncertainty).
There currently is one global multiregional input-output database that is produced by a supra-national organization, i.e. the OECD/WTO Trade in Value Added database (TiVA). A further, and more preferred, refinement of a footprint calculation approach with a “statistical stamp” can thus be to use this TiVA database as a starting point. This is a trade-balanced database but with a coverage of 30 sectors too aggregated for the calculation of water- material-, land- and emission footprints. A way forward is to use the detailing procedures as developed particularly for EXIOBASE, and the optimization procedures as developed for EORA, to arrive at a level of 100-200 industry sectors that is appropriate to perform proper footprint calculations. Lastly it is recommended to use internationally harmonized data sets for carbon emissions (e.g. based on IPCCC or IEA energy flows plus emission factors), materials (e.g. the recently published UNEP International Resources Panel), land and water, and add these to the more detailed TiVA database.
In this way, a database could be created that at an aggregated level has the “statistical stamp” provided by the OECD, uses extensions that are harmonized/commonly accepted, but also can provide more detailed information (through a procedure backed by a number of credible, scientific institutes). This would, for the first time, give a global multiregional input-output database that probably has a higher level of credibility as the individual scientific databases such as WIOD, EXIOBASE, GTAP or EORA. Such a database, that holds a middle ground between official statistics and scientific work, could be a good compromise for any NSI or practitioner to work with.
Our recommendations for the short term, in addition to moving forward based on the TiVA database as just described, are to agree in the formal statistical gremia on:
• a preferred and harmonized way to calculate footprints (i.e. using a true global value chain approach rather than other allocation mechanisms;
• in doing so, taking the residential perspective as starting point;
• and avoiding neglecting emissions or resource uses related to e.g. international bunkers;
• ensure the use of harmonized extensions databases.
It is likely that such an approach will significantly reduce differences in footprint calculations for countries. Moreover, given the importance for environmental-economic interactions on a global level, in which Europe is an important driver for environmental impacts outside the EU, it is recommended that the EU uses such a harmonized data and analysis tool in a role of “knowledge broker” to non-EU countries on how objectives such as the UN Sustainable Development Goals can be met.
List of Websites:
Name, title and organisation of the scientific representative of the project's coordinator:
Prof. dr. Arnold Tukker,
Netherlands Organisation for Applied Scientific Research TNO
Tel: + 31 88866 8310 (work), + 31 6 51980344 (mobile)
Fax: + 31 88866 0757
E-mail: arnold.tukker@tno.nl
Project website address: http://fp7desire.eu/
