Sustainable Consumption and Production - Development of an Evidence Base. Project Ref.: SCP001 Resource Flows

Sustainable Consumption and Production

-

Development of an Evidence Base

Appendix to Final Project Report

Appendix SCP001 1/196

Projec Ref.: SCP001 Resource Flows

May 2006

Appendix SCP001 2/196 Contents I Review of Resource Flow Studies (Section 4) 6 I.1 Specific assessment requirements 6 I.1.1 General Methodological Procedure 6 I.1.2 Assessment Criteria 7 I.1.3 The Scorecard System 12 I.2 Characteristics of General MFA Methodologies 15 I.3 Detailed Assessment Results 19 I.3.1 Economy-wide Material Flow Analysis 19 I.3.2 Bulk Material or Material System Analysis – The Biffaward Series 31 I.3.3 Analysis of Material Flows by Sector: NAMEAs, Generalised Input-Output Models, and Physical Input-Output Analysis 35 I.3.4 Life Cycle Inventories 47 I.3.5 Substance Flow Analysis 54 I.3.6 Hybrid Methodologies 61 I.4 Environmental Impact Assessments 66 I.5 Policy Analysis 73 I.6 Discussion of Study by Van der Voet et al. (2005) 77 II Review of Biffaward Studies (Section 5) 79 II.1 Tables: Detailed assessment criteria, policy agendas, list of Biffaward studies 79 II.2 Assessment notes made for each study 87 III Development of an Indicator for Emissions and Impacts associated with the Consumption of Imported Goods and Services (Section 6) 152 III.1 Specific assessment requirements 152 III.2 Review of existing approaches described in the literature 155 III.2.1 Studies not involving input-output calculations 155 III.2.2 Studies involving input-output calculations 159 III.3 Detailed Assessment Results 176 III.4 Technical Specification 179 III.4.1 General framework (ideal model) 179 III.4.2 How to deal with data problems 181 III.4.3 Details of extending the model 188

Appendix SCP001 3/196 List of Figures Figure I.1 - General Assessment Procedure...... 7 Figure I.2 - Scorecard system for assessing the policy relevance of the different MFA methods...... 7 Figure I.3 - General Regional Material Flow Accounting Scheme ...... 19 Figure I.4 - Official Japanese Material Flow Policy Targets ...... 25 Figure I.5 - Measuring Decoupling via EMFA indicators: DMC vs GDP ...... 26 Figure I.6 - Relationship between the volume of physical flows and their impact per unit ...28 Figure I.7 - Japanese MFA policy embedded in the vision of a material-cycle economy...... 29 Figure I.8 - The four discrete stages of a life cycle assessment...... 47 Figure I.9 - Incomplete Coverage of Supply Chain by process based LCA methodologies ...50 Figure I.10 - A schematic concept of SFAs ...... 54 Figure I.11 - Good practice: working procedure SFA ...... 56 Figure I.12 - Quantitative policy analysis in a static perspective ...... 73

List of Tables Table I.1 - Scorecard ratings...... Error! Bookmark not defined. Table I.2 - Different flow types as distinguished by Eurostat ...... 16 Table I.3 - Overview general MFA methodologies...... 18 Table I.4 - EMFA Indicators ...... 21 Table I.5 - Environmental Indicators included in the UK's NAMEA publication ...... 42 Table I.6 - LCA/LCI applications for government and companies ...... 52 Table I.7 - Applications for LCI/A data ...... 53 Table I.8 - Criteria for choosing LCI methods ...... 64 Table I.9 - Impact categories of LCA approaches ...... 69 Table II.1 - Assessment criteria ...... 82 Table II.2 - SCP policy agendas – Used in the assessment of Criteria P1...... 83 Table II.3 - Other environmental policy agendas - Used in the assessment of Criteria P2 .....84 Table II.4 - The Biffaward studies...... 86 Table III.1 - Criteria used in the assessment of studies on embedded emissions ...... 154 Table III.2 - Overview of a monetary supply and use table framework...... 187 Table III.3 - Consumption of water resources by industrial sector in the UK, 1997/98 ...... 190 Table III.4 - Total waste arisings in the United Kingdom, 2002/03 ...... 192

List of Boxes Box I.1 - Initiatives and programmes relevant to the UK's SCP agenda ...... 13 Box I.2 - Policy Questions Economy-wide Material Flow Analysis applied to UK ...... 20 Box I.3 - An introduction to monetary and physical input-output...... 36 Box I.4 - Policy questions EIOA ...... 37 Box I.5 - Good Practice Generalised Input-Output Analysis ...... 44 Box I.6 - Policy questions lifecycle assessment ...... 48 Box I.7 - Overview ISO LCI/A standards ...... 49 Box I.8 - Policy Questions - Substance Flow Analysis ...... 55 Box I.9 - Good Practice: Standardised SFAs in Denmark...... 59

Appendix SCP001 4/196 Appendix SCP001 5/196 Review of Resource Flow Studies (Section 4)

I.1 Specific assessment requirements

I.1.1 General Methodological Procedure The general assessment procedure for the review of the MFA methodologies is described in detail here. The assessment criteria are defined in detail in Section I.1.2. Drawing on the project tender specifications these criteria cover three broad categories:

1. Methodological robustness and soundness

2. Data availability and robustness 3. Policy Relevance Each of these categories is assessed separately and judgements are expressed in an ordinal rating system distinguishing “good”, “satisfactory” and “unsatisfactory” following the colour scheme previously outlined. While the assessment of “methodological robustness and soundness” is kept general, “data availability and robustness” as well as “policy relevance” are assessed in a UK-specific context to maximise the value to DEFRA and the UK’s research and policy community of this report. In this part of the report no distinction is made between useful and required characteristics. Moreover, no scoring point system is maintained and no ordinal final verdict is provided. The reason for this is that this part of the study compares very different methodologies rather than the quality of individual studies. This cannot easily be achieved through a rigid framework.

The goal of this assessment is to understand for each unique methodology its different strengths and weaknesses and its relevance and applicability to the UK’s SCP agenda. The assessment matrix is therefore used to identify potential methodological limitations and to demonstrate in which policy context different approaches could be best applied. The general assessment procedure is summarised in Figure 0.1.

Appendix SCP001 6/196 Figure 0.1 - General Assessment Procedure Further adjustments have been undertaken in the assessment of the policy relevance of the various resource flow methodologies. As policy generates demand for statistics the primary question is how the different methodologies fit into the UK’s SCP, SD, and general environmental agenda. Therefore, three sets of policy questions from the SCP agenda are used here: one referring to general SCP objectives derived from “Changing Patterns” (DEFRA, 2003), a second to specific SCP objectives derived from “One Planet Economy” (DEFRA, 2005b) and the third to the specific questions raised by DEFRA in the course of this tender. The questions were written down on scorecards (see Section I.1.3) and for each methodology it was assessed whether or not a MFA methodology can actually contribute to answering them. The general way of assessing the policy relevance is summarised in Figure 0.2.

Figure 0.2 - Scorecard system for assessing the policy relevance of the different MFA methods

I.1.2 Assessment Criteria Below, the different assessment criteria are defined. For each criterion a specification is provided what qualifies a certain judgement in the ordinal rating system (good, satisfactory, unsatisfactory).

General Assessment Criteria Criterion Description Specification B1 Good Clearly defined system Unambiguous identification of material system boundaries in human Unambiguous identification of the economic/human System System and material sphere

Boundaries Boundaries system

Appendix SCP001 7/196 Satisfactory System boundaries only Physical flows or relevant part of human system not partially defined, but unambiguously identified method comprises HOWEVER systematic way of Methodology clearly recommends which flows should identifying most be included, and/or, relevant entities/ Methodology clearly recommends which activities in physical flows in the the human system should be included, and/or, human and material Methodology is provided to identify the most relevant sphere flows Unsatisfactory Arbitrary and not Physical flows are not unambiguously defined and no satisfactorily defined clear recommendations are provided where to draw the system boundaries C1 Good Yes A standard is established and well documented Satisfactory Influential paradigmatic There is a paradigmatic approach available, which approach available serves as a template for existing studies An approach is termed paradigmatic if more than 50% of the reviewed studies directly or indirectly (a study refers to the methodological procedure of a study that directly refers to the methodology of the paradigmatic

Method standardized approach) refer to its methodological procedure. Unsatisfactory no There is no methodological standard identifiable C2 Good Yes The approach directly relates to other internationally agreed standards (e.g. SEEA, SEA, ISO etc.) Satisfactory Approach closely The approach relates closely to the basic concepts of relates to concepts of internationally agreed standards internationally agreed standards Unsatisfactory No The approach does not relate to an international

Consistency with other with other Consistency standard classification even though there exists a

nationally/ internationally suitable one C3 Good Employs an Methods prescribes the use of international standard internationally classification systems (or their national standardized implementations) classification Satisfactory Provides alternative The method proposes an alternative classification classification system, system, which can be fully re-converted into a which relates to standard system standard classification No internationally agreed classification standard system exists – method proposes new classification (based on standardised data sources) Unsatisfactory Does not relate to The method does not classify according to a standard standard classification classification system even though there exists a codes suitable one

C4 Good Adheres to material The method requires the establishment of a material balance principle balance in the data collection process Satisfactory Does not apply material The method does not apply material balancing, but balance principle, but provides good or satisfactorily defined system comprises careful boundaries in the material sphere (see specification Consistency with Consistency

Material Balance Material Balance procedures to include all B1) relevant physical flows

Appendix SCP001 8/196 Unsatisfactory Does not apply material The method does not require material balancing and balance principle and provides unsatisfactorily defined system boundaries includes physical flows in arbitrary way M1 Good Approach is robust No perceivable flaws in the methodology Satisfactory Minor flaws in the Any shortcomings in methods are credibly applied, approach are transparently discussed and are taken account of. However, no better practices readily available Unsatisfactory Serious flaws in the Serious flaws in approach (Lack of transparency or

Method Robustness Robustness Method approach credibility in the use of approaches and methods) M2 Good Integral part of Sensitivity Testing is a standard part of the methodology methodology Satisfactory Has been undertaken in Representative sensitivity testing has been undertaken peer reviewed paper in a peer reviewed paper or in an official report Unsatisfactory Uncertain No reliable information regarding the sensitivity of estimates

Sensitivity Analysis Sensitivity Assessment is available, but cannot be seen as representative for the methodology M3 Good Methodology is Applied in at least 15 peer reviewed papers frequently applied in peer reviewed papers Satisfactory Methodology has been Applied in 0

understanding understanding Unsatisfactory Methodology has not Has never been applied in peer reviewed papers been applied in peer

Peer reviewed and width of and Peer reviewed reviewed papers R1 Good Yes System boundaries must be fully defined, method must be standardized and robust and adopt a standard classification Satisfactory partially - some degrees See above, but at least one satisfactory assessment of freedom in result – no unsatisfactory methodological set-up makes detailed

comparable results results comparable comparison difficult Unsatisfactory no At least 1 unsatisfactory evaluation of above Method designed to provide toMethod designed criterions

UK specific assessment - Data Criterion Description Specification D1 Good 2001+ Data refers to the years 2001 and later Satisfactory 1998-2000 Data refers to reporting periods 1998-2000 Unsatisfactory 1997 and earlier Data is from reporting periods 1997 and older Data Availability n

bli D2 Pu cat io

Appendix SCP001 9/196 Good Yearly or more Data published at least once a year often Satisfactory On a regular basis Data published in regular intervals not bigger than 5 years Unsatisfactory Once, irregularly or Data published once, irregularly or in publication in very large intervals bigger than 5 years intervals D3 Good >25 A time series of n=> 25 available Satisfactory >10 A time series of 10

Times series Unsatisfactory <10 A time series of n<10 available D4 Good Uses robust real and Only uses official government statistics and applies modeled data (fit to and robust real data estimation procedures (fit for all all purpose) purpose) Satisfactory Uses real data of Uses official government statistics and unofficial data good quality and of good quality from trustworthy sources and robust modeled data (fit to real data procedures (fit for general purpose) general purpose) Data Robustness Data Robustness Unsatisfactory Robust and modeled Concern about data source and method of data not robust construction and unstable data estimation procedures D5 Good All data publicly All data can be easily obtained from public domain available Satisfactory Some data only Some data only available on request -> this includes available on request data which can only be accessed in certain locations; qualifies as long as every researcher has the

Data Access Access Data opportunity to access the data Unsatisfactory Restricted access Data is not accessible for any research institutions/ researcher in the UK/ special permission D6 Good Data well The data is well documented and understandable to documented and any reader understandable to any reader Satisfactory Documented Data is well documented and understandable to any reader, but important documents are not readily available Data is documented and documentation easily accessible, but some information is lacking Unsatisfactory Not properly Severe shortcomings in documentation of data and/or

Data Documentation Data Documentation documented problems in accessing this documentation

D7 L Good UK approach(es) Data availability exhausts methodological potential – fully exhausts all policy questions can be answered methodological potential Satisfactory UK approach(es) Data availability exploits most of the methodological exhausts potential - important policy questions can be tackled methodological

Methodological Potential partially

Appendix SCP001 10/196 Unsatisfactory UK approach Data availability does not exploit the methodological exhausts potential - crucial policy questions cannot be methodological answered potential insufficiently

UK specific assessment - Policy Criterion Description Specification P1 Good Relevant to a crucial Methodology contributes significantly to at least 1 policy agenda crucial policy agenda (see assessment file) Satisfactory Relevant to a general Methodology must contribute to at least 1 general policy agenda policy agenda (see assessment file), or contrbute generally to at least one crucial policy agenda (see assessment file)

Policy Agendas Unsatisfactory Only of hypothetical Otherwise relevance to policy agendas P2 Good Provides crucial Methodology contributes significantly to answering 1 information for SCP of the 30 SCP key policy questions (see assessment strategy file) Satisfactory Provides general Methodology contributes generally to answering 1 of information to SCP the 30 SCP key policy questions (see assessment file) strategy Unsatisfactory Not helpful for Methodology does not contribute to answering any

SCP Policy Objectives government’s SCP policy questions (see assessment file) strategy P3 Good Provides crucial Methodology contributes significantly to answering at information for least one DEFRA policy question (see assessment answering DEFRA's file) key SCP questions Satisfactory Provides general Methodology contributes generally to answering at information for least one DEFRA policy question (see assessment answering DEFRA's file) key SCP questions Unsatisfactory Not helpful to answer Otherwise DEFRA's key SCP

DEFRA priorities priorities DEFRA questions

P4 Good The approach is officially being endorsed / adopted by a decision-making Endorsement Endorsement authority

Appendix SCP001 11/196 Satisfactory The approach is officially endorsed by another country or another supranational body/ international organisation Unsatisfactory No official acknowledgement of the approach

I.1.3 The Scorecard System The scorecard questions form the heart of assessment of the policy relevance. They are divided up into three assessment categories: 1. general scorecard questions – they relate to four general SCP objectives – decoupling, impacts, productivity and information - derived from the Government’s SCP framework (see DEFRA, 2003); 2. specific scorecard questions – they refer to seven specific SCP objectives – products, production, consumption, procurement, innovation, sustainable business, waste - derived from the Government’s SCP strategy (see DEFRA, 2005a); 3. DEFRA SCP tender questions – these are the questions that have been raised by DEFRA within the tender process.

General scorecard Questions

The scorecard questions associated with the general SCP objectives are shown below: 1. Decoupling environmental growth and environmental degradation • Does the approach assess whether decoupling is taking place relatively and absolutely in one of the relevant areas defined by the government? • Does the approach take ‘burden shifting’ into these considerations? 2. Focussing on the most important environmental impacts associated with the use of particular resources • Does the approach allow monitoring resource flows with relevant environmental impacts and help to understand how they can be addressed? 3. Increasing resource productivity of material and energy use, as part of the broader government commitment to increase the productivity of the nation • Does the approach facilitate an assessment of the productivity of relevant material flows? 4. Encouraging and enabling active informed individual and corporate consumers who practice more sustainable consumption • Does the approach assess the resource flow implications of (economic) consumption? • Does the approach provide relevant information that can inform private, corporate and public consumers in their decision?

Appendix SCP001 12/196

Specific Scorecard questions

The specific scorecard questions were derived from a variety of programmes and initiatives, which are of relevance for the Government’s SCP strategy.1 They are summarised below in Box 0.1. Market Transformation Program Pollution Prevention and Control Business Resource Efficiency and Waste Programme (BREW) Environmental Action Fund Roundtable for Sustainable Consumption Consumer, Products and the Environment Program Environmental Technologies Action Plan (ETAP) DEFRA Environmental Technology Initiative Sustainable Procurement Action Plan DTI Sustainable Technology Programme DEFRA Environmental Industry Sector Unit (EISU) initiatives DTI Business Opportunity Programme Waste and Resource Action Pan (WRAP) Waste Implementation Programme

Box 0.1 - Initiatives and programmes relevant to the UK's SCP agenda

In the end 25 questions were assigned to the seven specific SCP objectives on the scorecard. The scorecard questions are outlined below. Products Source(s): Market Transformation Programme Q1: Does the approach provide information about the impacts of different products across the lifecycle? Q2: Does the approach help to identify where the environmental impact of everyday products can be best reduced across the lifecycle? Q3: Does the approach help to identify relevant policy measures to close the loop in the way we use resources and promote more radical design solutions? Q4: Does the approach help to set environmental product standards?

Production Sources: Pollution Prevention and Control; BREW (Business Resource Efficiency and Waste Programme) family; Market Transformation Programme Q5: Does the approach provide information relevant for promoting resource efficiency? Q6: Does the approach allow a detailed assessment of supply chains for key physical flows (e.g. fossil fuels, toxic materials etc.)? Q7: Does the approach help to identify environmental key sectors, where pollution prevention should take place (benchmarking, reporting, indicators)?

Consumption Sources: Environmental Action Fund Programme; Roundtable for Sustainable Consumption; Consumer, Products and the Environment Programme Q8: Does the approach allow identifying which sectors/ items drive the resource use associated with household consumption?

1 In fact, all these initiatives were directly linked to Government’s SCP website, where they were categorised into the seven specific SCP objectives: http://www.defra.gov.uk/environment/business/scp/

Appendix SCP001 13/196 Q9: Does it allow taking health considerations into account? Q10: Does the approach help to identify how household consumption patterns can be changed in an environmentally desirable way? Q11: Provide information of environmental performance of different products to private and public consumers to promote more sustainable choices?

Procurement Sources: Environmental Technologies Action Plan (ETAP), DEFRA Environmental Technology Unit, Sustainable Procurement Action Plan Q12: Does the approach help to assess the resource use environmental impact of government consumption arising on its own estate and in the supply chain? Q13: Does the approach help to identify where sustainable procurement can deliver the most significant environmental and competitiveness outcomes? Q14: Does the approach help to analyse how can we make more efficient use of government consumption, e.g. through reduced energy consumption and reduced packaging? Q15: Can the approach help in the development of further public procurement targets and a corresponding indicator set for government procurement?

Innovation Sources: DTI Sustainable Technology Initiative, DEFRA Environmental Industries Sector Unit (EISU), DEFRA Environmental Technology Unit, ETAP Q16: Can the approach help to benchmark environmental technologies? Q17: Does the approach allow identifying where technologies can be best applied to minimize the environmental pressure generated throughout a profucts life cycle?

Sustainable Business Sources: DTI Business Opportunity Programme Q18: Can the approach contribute to minimise environmental impact in a structured way through a formal environmental management system? Q19: Can the approach help addressing SCP on a sectoral basis? Q20: Can the approach help to identify/set environmental product standards?

Waste Sources: BREW, Waste and Resource Action Programme (WRAP), Waste Implementation Programme Q21: Does the approach help to identify where waste is arising across supply and demand entities? Q22: Does the approach help to understand where and how waste can be most effectively minimised? Q23: Does the approach provide relevant information about recycling? Q24: Does the approach allow distinguishing different types of wastes? Q25: Does the approach allow identifying where the various types of waste end-up?

On these scorecards it was then recorded whether a methodology could make a significant, a general, or no contribution for answering a particular policy question. The visual representation of the assessment is summarised by the colour scheme shown in Error! Reference source not found.. If at least one question within one SCP objective such as “waste” could be directly addressed, the contribution of the method received a good assessment. If one objective could be addressed well with a method, we established that the overall contribution for answering specific SCP objectives was good as well. This means that one good scorecard assessment was sufficient for a good overall rating of the policy relevance in the course of answering general, specific or DEFRA SCP questions. This approach can be debated, but it appeared to be the least arbitrary assessment for the purposes of this

Appendix SCP001 14/196 study. 2 Satisfactory overall ratings were given when no good rating was achieved on a scorecard, but at least one satisfactory rating was achieved. Ratings good Question directly addressed – significant contribution satisfactory Question generally addressed – generally addressed - Question not addressed

Table I.1 - Scorecard ratings

I.2 Characteristics of General MFA Methodologies

The different material flow methodologies included in this report have already been distinguished according to their aggregation level in the economic and material sphere. This is very helpful to develop a broad classification, but for a deeper understanding it is important to establish how the distinctive features of the methodologies differ. Therefore some general methodological characteristics have been developed. Table I.2 provides a general overview of these characteristics and how they relate to the various general MFA methodologies. They are grouped in three broad categories: • Material Sphere • Human Sphere • Accounting and Others In the material sphere, the first characteristic is flow coverage. Eurostat (2001) proposes to distinguish three dimensions of resource flows: • The territorial dimension distinguishes material flows according to some political or institutional boundaries, i.e. whether the materials have been extracted from the domestic economy or from the rest of the world (ROW). • The lifecycle dimension distinguishes whether to consider only direct materials flows or also indirect material flows, which occur higher up in the supply-chain. • The product dimension distinguishes materials according to whether they enter the economic process or not, i.e. whether materials are used within the economic system or not (unused). According to these three dimensions, five different flow types can be distinguished. They are shown in Table I.1 with the appropriate terminology. Note that hidden flows are defined as the sum of unused domestic and foreign extraction, as highlighted by the grey shaded areas.

2 In the course of this project we considered requiring a higher number of significant contributions for answering SCP policy questions to achieve a good assessment. Here, the arbitrariness of the choice of the threshold number, and the weighting of the actual choice and phrasing of the SCP policy questions led us to the conclusion that the currently applied less demanding approach is more appropriate. Note that the number of significant and general contributions in answering SCP policy questions will be indicated in the assessment matrix. DEFRA can then decide whether or not to value a certain approach higher, because it answers more policy questions or whether they put more weight on approaches, which answer particular questions out of the set.

Appendix SCP001 15/196

Lifecycle Used or Unused Domestic or ROW Terminology Direct Used domestic (used) domestic extraction not applied Unused domestic unused domestic extraction Direct Used ROW imports indirect (upstream) Used ROW indirect (input) flows indirect (upstream) Unused ROW associated with imports

Table I.1 - Different flow types as distinguished by Eurostat

The second characteristic in the material sphere is the flow type distinction introduced earlier between aggregate flows (adding up the full weight of all materials within a system), materials (including products) and substances. In the human sphere the different approaches might refer to different geographical, political, or institutional scales – global, national, regional, local – and to different institutional entities within the system – economy, sector, company/enterprise, product/service. The remaining characteristics are shown in Table I.3.

It needs to be emphasised that is based on personal judgement. It might for example be debatable whether or not domestic and foreign unused flows as well as indirect used flows should be seen as a characteristic feature of EMFA. For example, Eurostat (2001: 24) highlights that empirical work often leaves out indirect used flows. However, as these flows are included in some of the most frequently discussed indicators proposed in the Eurostat guide such as total material requirements, they are seen as characteristic here.

Each methodology will be discussed later in more detail, even though some general observations arising hould be highlighted. First, the Biffaward studies cover almost every characteristic. This is due to the fact that they are what might be seen as a methodological hodgepodge – very different studies carried out within the Biffaward Programme taking place on different levels in the human as well as the material sphere, orchestrated around the material balance principle as a key feature. Second, only lifecycle assessment shows unique features, while all other methods share their characteristics with at least one other study. As long as these characteristics are in demand on the policy level and these general methodologies are robust, they have a unique selling proposition.

Appendix SCP001 16/196

Appendix SCP001 17/196 Table I.2 - Overview general MFA methodologies

Appendix SCP001 18/196 I.3 Detailed Assessment Results

I.3.1 Economy-wide Material Flow Analysis

Description Economy wide material flow analysis (EMFA) can be seen as an attempt to systematically measure and understand the overall physical size of a country’s societal metabolism, in tonnes, for a given reporting period (usually a particular year) within clearly defined geographical boundaries. This includes all physical flows out of and into the natural environment as well as the physical size of imports and exports. While the method focussed entirely on the physical inflows at the beginning (e.g. Adriaanse et al., 1997), the establishment of a material balance has become a typical methodological feature since (Eurostat, 2001). In this course physical inflows and outflows are balanced by net accumulation in the economic sphere.

As implied by its name the focus of EFMA is solely on flows and therefore only includes changes in stocks, but does not try to quantify the total physical stock size. Measurement is carried out at the system boundaries between nature and a defined region (see Figure 0.3). EMFA methods have most prominently featured on the national level (e.g. Matthews et al., 2000; Eurostat, 2002; Van der Voet et al., 2005) even though studies have also been carried out at the sub-national (e.g. SEI, et al., 2006) and supra-national level (e.g. Moll et al., 2003; Eurostat, 2002). The physical (product) flows within the economy are not described, i.e. the economy is usually not disaggregated into (some of) its components and dealt with as a black box (see Femia and Moll, 2005; Eurostat, 2001).

Figure 0.3 - General Regional Material Flow Accounting Scheme (Eurostat, 2001)

The ultimate aim of material flow accounting is the provision of a set of indicators at the highest aggregation level that describe the total material flows of a region at the input and output side of the economy as shown in Table I.3. Based on this set of indicators policy makers and analysts can address

Appendix SCP001 19/196 a variety of general policy questions as shown in Box 0.2 below. Some of them can be answered just by looking at the indicators (over time), others require more sophisticated statistical methods for analysis. However, EMFA accounts are also increasingly used for more detailed analysis of particular groups of materials (e.g. Moll et al., 2003; Van der Voet et al., 2005).

Questions describing the overall state of the Physical Flow System What is the size of the physical system supporting the UK economy? What is the composition of physical inflows and outflows – what materials drive the total weight? How much of a particular resource input is used by the economy and how much of a particular waste/pollutant is generated? What share of the total material inputs into the UK economy stem from reuse and recycling?

Questions describing the system dynamics (change of the system over time) Is the physical size of the UK economy increasing or decreasing? Is a decoupling of aggregate resource use and economic growth observable in the UK? Is this decoupling of relative or absolute nature? Could the final disposal amount of municipal waste in the UK be reduced over the last years? Is there an increase in reused and recyled material input in the UK?

Questions describing the physical trade relationships How much does the UK rely on the resources provided by its own territory? Is the physical trade balance of the UK positive or negative?

Box 0.2 - Policy Questions Economy-wide Material Flow Analysis applied to UK

In contrast to many other MFA methodologies, EMFA attempts to fully cover what has been frequently called “hidden flows”. While direct and indirect used flows at home and abroad are frequently (fully) addressed, only very few other methodologies try to estimate domestic and foreign unused extraction.

Appendix SCP001 20/196

Abbrevi Indicator Name Indicator Name ation DMI Domestic DMI measures the direct input of materials for use into the economy, Material Input i.e. all materials which are of economic value and are used in production and consumption activities; DMI equals domestic (used) extraction plus imports. DMI is not additive across countries. TMI Total Material TMI includes, in addition to DMI, also unused domestic extraction, Input i.e. materials that are moved by economic activities but that do not serve as input for production or consumption activities (mining overburden, etc.). Unused domestic extraction is sometimes termed ‘domestic hidden flows’. TMI is not additive across countries. TMR Total Material TMR includes, in addition to TMI, the (indirect) material flows that Requirement are associated to imports but that take place in other countries. It

Input Indicator measures the total ‘material base’ of an economy. Adding indirect flows converts imports into their ‘primary resource extraction equivalent’. TMR is not additive across countries. DTMR Domestic Total DTMR includes domestic used and unused extraction, i.e. the total Material of material flows originating from the national territory. Domestic Requirement TMR equals TMI less imports. Domestic TMR is additive across countries. DMC Domestic DMC measures the total amount of material directly used in an Material economy (i.e. excluding indirect flows). DMC is defined in the same Consumption way as other key physical indictors such as gross inland energy consumption. DMC equals DMI minus exports. TMC Total Material TMC measures the total material use associated with domestic Consumption production and consumption activities, including indirect flows imported (see TMR) but less exports and associated indirect flows of exports. TMC equals TMR minus exports and their indirect flows. NAS Net Additions to NAS measures the ‘physical growth of the economy’, i.e. the Stock quantity (weight) of new construction materials used in buildings and other infrastructure, and materials incorporated into new durable goods such as cars, industrial machinery, and household appliances. Materials are added to the economy’s stock each year (gross additions), and old materials are removed from stock as buildings are Consumption Indicators demolished, and durable goods disposed of (removals). These decommissioned materials, if not recycled, are accounted for in DPO (see below). PTB Physical Trade PTB measures the physical trade surplus or deficit of an economy. Balance PTB equals imports minus exports. Physical trade balances may also be defined for indirect flows associated to Imports and Exports. DPO Domestic DPO measures the total weight of materials, extracted from the Processed domestic environment or imported, which have been used in the Output domestic economy, before flowing to the environment. These flows occur at the processing, manufacturing, use, and final disposal stages of the production-consumption chain. Included in DPO are emissions to air, industrial and household wastes deposited in landfills, material loads in wastewater and materials dispersed into the environment as a result of product use (dissipative flows). Recycled material flows in the economy (e.g. of metals, paper, glass) are not included in DPO. An uncertain fraction of some dissipative flows (manure, fertiliser) is ‘recycled’ by plant growth, but no attempt is made to estimate this fraction and subtract it from DPO. Output Indicator

(unavailable for UK) (unavailable for DMO Direct Material DMO is the sum of DPO plus exports. This indicator represents the Output total quantity of material leaving the economy after use either towards the environment or towards the rest of the world. DMO is not additive across countries. TMO Total Material TMO measures the total of material that leaves the economy. TMO Output equals TDO plus exports. TMO is not additive across countries.

Table I.3 - EMFA Indicators as proposed by Eurostat (2001)

Appendix SCP001 21/196

Assessment - Methodology EMFA is a methodology that has been standardised just relatively recently by Eurostat (2001). This document serves as the basis for the methodological assessment. It brings together the insights from a task force of European EMFA experts and might be seen as a synthesis of their experiences gained during the 1990s (e.g. Adriaanse et al., 1997; Matthews et al., 2000). The basic accounting principles and classifications have made their way into the System of Integrated Environmental and Economic Accounting (SEEA) (see UN, 2003). This has not only further spread the outreach into the international community, but also secured integration potential with other environmental-economic accounting frameworks through shared basic concepts. The integration of accounting frameworks in various units is a pre-condition for any comprehensive, quantitative SCP or SD inquiry. For example, in policy analysis physical flows are commonly modelled dependent on human behaviour in the socio-economic system, which is commonly derived from monetary data. Consistency in the physical accounting process is ensured through the application of the material balance principle, which positively influences the method’s robustness.

The system boundaries are well defined both in the human as well as in the material sphere. EMFA accounts for all used and unused, direct and indirect physical flows occurring at home and abroad, which are associated with production and consumption patterns of an economy. However, EMFAs do not measure the total weight of an economy. Materials like water and air are excluded as their physical weight would overwhelm the remaining flows, which are seen as more relevant.3 For water and air the establishment of separate accounts is recommended. Good care is taken of borderline cases such the treatment of landfills or the plant stock in agriculture with clear recommendations. These recommendations seem to contribute positively to the policy relevance of the method as comparability and reliability of the various estimates is improved.

The method robustness received a satisfactory rating, because concerns arose in the context of the methodology for establishing accounts for hidden (domestic and foreign unused extraction) flows, which lack the clarity and precision of information in the rest of the guide. Hidden flows are very difficult to estimate as they cannot be easily derived from standard data sources (see, Gazley and Francis, 2005). It is surprising that no standard procedures are established in the guide to determine how the imputation coefficients for the various material types can be obtained on a comparable basis.

3 This, however, in a way already questions the significance of the whole idea of deriving aggregate indicators for the “weight of a nation”.

Appendix SCP001 22/196 However, due to the modular build-up of the EMFA framework, hidden flows can be easily excluded, if desired. Moreover, most EMFA indicators are not built on hidden flow estimates in a first instance as shown in Table I.3. In fact, the EMFA indicators adopted by the UK government do not rely on hidden flow estimates even though TMR and TMI are estimated by ONS.

Sensitivity analysis is only occasionally performed in the course of EMFAs. Further confirmation of soundness and robustness of the method stem from the large amount of papers published in peer- reviewed academic papers. Taken together the methodological features jointly contribute to a high level of comparability in results across studies as long as the practitioner follows the advice provided in the Eurostat guide. This high degree of comparability is confined for especially for hidden flows. In these area greater methodological standardisation would be desirable. However, as the Wuppertal Institute for Climate, Environment and Energy – in some capacity - has been involved in many studies, that have taken place throughout Europe, even in these less standardised areas the degree of comparability across existing evidence might be higher than indicated by the “satisfactory” rating.

Assessment - Data Originally compiled by the Wuppertal Institute for Climate, Environment and Energy (DEFRA, 2002; DEFRA, 2002) maintenance, continuation and development of the UK’s economy-wide MFA account (as a representative for EMFA accounts in general) has been subsequently taken over by the Office for National Statistics. The accounts are published bi-annualy as part of the environmental accounts. The data is available from 1970 to 2004, i.e. the most recent data point in the account time series is always T-1 with T being the publication year. Such a long time series is an excellent basis for the analysis of economy-wide material flows as it opens up a wide field for the application of statistical methods such as regression analysis, decomposition analysis, cluster analysis and many more. The wide applicability of the data in different analytical frameworks should also open up a lot of opportunities to provide a tailor-made analysis fitting the demand of various policy agendas like SCP. However, to our knowledge only very limited use has been made from this potential (an exception e.g. is Bringezu et al., 2004) and studies have largely been restricted to descriptive uses. The EMFA data is publicly available on an aggregated level. However, a detailed analysis of the flows of individual groups of bulk materials is not enabled with this standard data. This reduces the potential policy relevance, because it does not allow for a more detailed understanding of the relationship between the change in the overall level of flows and the flow composition. This is particularly important when the analyst wants to develop an understanding on how the change in the overall level relates to potential environmental pressures caused. However, it was confirmed by ONS in the course of this project that researchers can obtain the detailed EMFA data on request even though in specific cases disclosure issues might occur, which then need to be resolved on a case-by- case basis.

Appendix SCP001 23/196

The documentation of the data is remarkably good. In particular, the first publication (DEFRA, 2001; DEFRA, 2002) has provided publicly available material providing a detailed outline of most practical problems encountered during the compilation process. The assessment of the data robustness is assumed to be high due to the high institutional quality standards at ONS. In general, sensitivity analysis carried out by the Wuppertal Institute showed significant variations in estimates when lower or higher bound estimations for imputed values were used. However, these stayed usually well below the ±10% mark even though some outliers can be found for certain indicators (see DEFRA, 2001). Despite the good rating concerns about the robustness of hidden flow estimates are reiterated, while the UK specific adjustment of the estimates is recognised and the improvement in reliability appreciated (see, Lawson et al., 2003). In terms of the UK data available the biggest limitation is posed by the absence of data for landfilled wastes as well as emissions to water. This limits the fulfilment of the methodological potential in two respects: firstly, no complete material balance can be established and therefore important consistency checks of the data cannot be fully carried out. Secondly, important output indicators cannot be established. With an increasing attention directed towards sinks in the environmental debate, in general, and waste on the UK’s SCP agenda in particular, this has a direct, negative feedback for the policy relevance of the approach. Moreover, the application of EMFA data in the context of waste policies seems to be one of the most interesting applications.

Assessment - UK Policy Relevance Economy-wide Material Flow Analysis is usually motivated with reference to the large amount of natural resources used and wastes generated by mature economies, i.e. many of the most pressing environmental problems such as global warming, biodiversity loss and deforestation at home or abroad might be (mainly) driven by the sheer amount of resource use in the developed world. Once accepted that a general reduction in resource use is necessary to solve environmental problems on the source and sink side of the economy, EMFA provides insights to a variety of policy questions (see Box 0.1). Aggregate indicators such as direct material input (DMI) or domestic material consumption (DMC), for example, try to provide a picture of where an economy is moving in terms of resource use similar to aggregate economic indicators like GDP.4 In this context it has been prominently proposed that the size of resource use patterns in industrialised countries and the need for development in low

4 Interestingly, in the course of the publication of the autumn edition of the UK’s environmental accounts the ONS website showed as one of the main news: DMI increased the second year in a row.

Appendix SCP001 24/196 and middle income countries requires a reduction in resource use by a factor four (von Weizsaecker et al., 1997) or even ten (Schmidt-Bleeck et al., 1997). Some countries like Germany, Japan, Italy, the Netherlands, Belgium, Czech Republic, Austria, Latvia and others have followed the rationale of EMFA approaches and already formally established quantitative reduction or resource productivity targets in their sustainable development strategies. Germany, for example, has committed to reduce energy use by a factor 2 by 2020 compared with 1993 levels, and raw materials by a factor of 2.5 (Federal Government of Germany, 2002 and 2004). In Sweden even a factor 10 improvement within the next 20-50 years has been suggested by the Eco- cycle Commission. Japan has a commitment to increase resource productivity by 40% between 2000 and 2010 to 390000 Yen GDP per tonne DMI (see

Figure 0.4). Overall, more than 20 countries have established MFA accounts in the OECD and the methodology is officially promoted by EU institutions (Eurostat, EEA) as well as the OECD (see Femia and Moll, 2005).

Figure 0.4 - Official Japanese Material Flow Policy Targets (Japanese Ministry for the Environment, 2003)

The UK Government has been much more careful in integrating EMFA targets into their SD and SCP strategy due to concerns about the suitability of the information to effectively inform the policy process. In “Changing Patterns” (DEFRA, 2003) a focus on environmental impact rather than total level of resource use was still stressed. Since then some change in attitude has taken place. Even though no specific targets have been set out by the Government domestic material consumption (DMC) has entered the Government’s shortlist set of 20 SD framework indicators (see ONS and DEFRA, 2005a) as well as the set of SCP decoupling indicators (see, ONS and DEFRA, 2005b).

Appendix SCP001 25/196 So what contribution can EMFA provide to the UK’s environmental agenda, in general, and the SCP agenda in particular? How useful is it to identify, set and monitor targets for SCP policy? As stated earlier an answer to these questions can only start from the policy level, i.e. policy should generate demand for statistics in a first instance. So to which objectives outlined by the Government can EMFA contribute? First, the overarching SCP goal of decoupling is seen by the Government as the only way of maintaining economic progress in the long run without reaching the limits of the Earth’s ecosystems ability to absorb pollution and to provide natural resources (DEFRA, 2003: 11). In this course the government has endorsed DMC as one official SCP decoupling indicator and therefore created demand for EMFA statistics itself. With a sufficiently long time series of data available (1970-2004) it can help to answer a variety of relevant policy questions such as: • Is decoupling of resource use and environmental pressure from economic growth taking place? • Is this a relative or an absolute decoupling? • What are the drivers behind, and how has the composition of material inflows changed over time? • Are the reductions taking place due to an increased burden shifting from the UK to other countries in the world through increases or changes in the composition of physical trade relationships? The required “international perspective” of research and policy efforts can be accommodated by EMFA but it should be noted that the DMC indicator does not take into account the indirect used flows associated with imports to the UK. It therefore still underestimates the degree to which environmental pressures are exported. In this light it should be considered how to take these flows into account in a new, tailor-made MFA indicator. The use of TMC is not recommended by the

authors’ as the relevance and robustness of hidden flows remains unclear.

Figure 0.5 - Measuring Decoupling via EMFA indicators: DMC vs GDP (ONS, 2005)

Second, another key concern of the Government on the SCP agenda is to improve resource efficiency. In this context EMFA provides an overall picture of the economy’s performance by setting

Appendix SCP001 26/196 derived indicators like DMI or TMR in a relationship with economic indicators such as GDP or GNP (see Figure 0.5). Typical policy questions that can be raised from there are: • Does the economy use resources more efficiently as a whole? • Does this translate into gross reductions or are these relative improvements eaten up by additional economic activities? • How efficient are certain bulk material inputs of interest such as fossil fuels or biomass used? Both sets of questions directly feed into the “general SCP agenda” as set out in the assessment matrix receiving a good assessment in the decoupling area and a satisfactory one resource productivity one. Decoupling performs better due to its official endorsement by the Government. The question of resource productivity is re-iterated on the “specific SCP agenda”, which is the only one that can be addressed by the standard EMFA methodology given the UK data availability.

In order to provide final recommendations for the usefulness of EMFA to inform SCP policy in the specific UK context, two issues should be further discussed: • Usefulness of aggregate EMFA indicators; • Comprehensive information provided by the underlying EMFA accounts. The usefulness of aggregate EMFA indicators for informing government policy might be best framed in a discussion of whether or not the government should set general reduction targets for overall levels of material flows like the Japanese government (see

Figure 0.4). Such targets are only of interest for any government’s agenda, if there are reasons to believe that a reduction in aggregate flows is environmentally beneficial, i.e. there needs to be a positive relationship between the state of the environment and the size of material flows.

Critics of EMFA have pointed out that aggregate EMFA indicators add-up very different materials and substances with very different potentials to contribute to environmental problems. Therefore, reducing domestic material flows by 20 percent does not necessarily mean that a single environmental problem has been tackled. In support of this argument critics of EMFA often refer to Figure 0.6, which shows the relationship between the volume of resource flows and their impact per unit. In this context it is highlighted that high impact flows tend to be small in size and low impacts flows big. These high impact low volume flows might get completely lost within the aggregate measure, whilst being of great importance in terms of the environmental pressure generated.

This is certainly a valid argument but there are several possible responses. One response is that where there is high consumption of low impact materials the absolute impact could still be high. Therefore, a reduction in the physical flow size is most likely to relate to a reduction in environmental pressure. Even in a country like Britain there are restrictions imposed on the use of water to protect the local environment during the summer. The reason is that society’s water consumption is so gigantic that it cannot be provided for during slightly drier periods. By-passing certain critical groundwater levels could then pose serious harm to this local ecosystem.5

5 In the course of this discussion Klaus Hubacek made an interesting point in his review of an preliminary version of this report. With regard to Figure 2.6 it could be said that once critical levels are reached low impact flows can become high impact flows. Once certain concentration levels of pollutants in the

Appendix SCP001 27/196

Figure 0.6 - Relationship between the volume of physical flows and their impact per unit (UN, 2003)

In addition, it could be said that high volume, low impact flows might be a bigger problem, because they tend to be under-regulated. Even if the absolute flow size becomes of environmental concern, regulation might proof difficult due to the high dependency of societies on these materials. Change can therefore only be induced slowly and certainly needs to overcome difficult institutional barriers. Climate change might be a good example. An early identification therefore is of great importance and EMFA could proof useful in this context. This conflict in arguments cannot be fully resolved. The only robust way of interpreting the indicators is to refer to what they measure - the total physical size of the system. For attaching any further meaning great care is required. At most an interpretation in terms of potential environmental pressure might be justifiable in the sense that a bigger system size has more potential to create environmental stress – without saying whether or not this is actually the case. This, of course, is very general and might not be seen as good basis for decision-making. For target setting in the specific UK SCP policy context the indicators therefore seem inappropriate. The UK policy agenda stresses the need to prioritise action and to tackle the environmentally most significant flows first. This can neither be achieved via general reduction targets on such an aggregate level nor be usefully informed.

While doubts are justified to whether any robust environmental interpretation can be given to changes in aggregate EMFA indicators in terms of (potential) environmental impacts, the number of countries, which already use EMFA indicators, might suggest that they have proven successful at least in a complementary way: the indicators are rooted in a comprehensive set of accounts recording (individual and groups of) materials on lower aggregation levels. Significant changes in aggregate

atmosphere are reached, for example, every additional unit might cause substantial harm, i.e. in times of high environmental stress their might be high volume high impact flows. This should be considered in the course of the climate change debate as well.

Appendix SCP001 28/196 indicators can therefore be traced back to changes in the aggregate flow composition. The sources of changes can then be evaluated on this lower aggregation level. If DMC goes up, for example, due to an increased consumption of sand, policy makers might be less concerned than if this increase is due to an increased consumption of fossil fuels. The evaluation of changes in flow composition might then lead to further analysis of identified “problem flows” using other MFA methodologies and the identification of how and where these might be best tackled. Using EMFA accounts in such a way for more specific analysis has become increasingly popular during the last years (e.g. Moll et al., 2003; Van der Voet et al., 2005) and might be seen as the most interesting and relevant application of EMFA in the UK specific SCP context. Hence, it is not necessarily the information that DMC has increased over the last two years, which is of relevance for the SCP policy agenda, but the information what caused this increase and how the flow composition has changed hidden in the EMFA accounts.

Figure 0.7 - Japanese MFA policy embedded in the vision of a material-cycle economy This is where the unique contribution of EMFA for informing SCP policies and the build-up of the physical evidence base lies. EMFA is the only methodology, which provides a “complete” set of physical accounts and therefore to comprehensively accounts for the substitution of materials induced by SCP policies. This avoids missing out unintended physical side effects of policies, which are easily

Appendix SCP001 29/196 missed out with less comprehensive data, and is therefore crucial for consistent use of physical flow evidence. However, some of the policy relevance rests on the assumption that it is easier to link effects on the environment to materials than to aggregate flows. This is certainly the case for some materials like fossil fuels and their relationship to climate change. However, these links are not so established for many other physical flows entering and leaving the economy. Even though these links have been strengthened recently in novel approaches (De Bruyn et al., 2004; Van der Voet et al., 2005) further work needs to be done in this key area. The more established the links, the more useful the contribution of the EMFA accounts for a prioritised SCP policy action will be.

Hence, overall it seems important to tailor aggregate material flow policies and indicators to the UK’s needs. Japan is a good role model. Instead of just devising aggregate resource flow targets Japan firstly devised a material flow policy (Figure 0.7). Based on the vision of a cyclic Japanese economy, priority areas for manufacturing and waste management were established. Indicators where then designed to monitor progress on a structural level and targets were defined. Hashimoto and Moriguchi (2004) then reviewed these indicators and devised six alternative ones, which avoided problems with double-counting and were closer to the actual policy process.

Appendix SCP001 30/196 I.3.2 Bulk Material or Material System Analysis – The Biffaward Series

Description Funded by the landfill tax credit scheme, Biffaward has commissioned a series of more than 50 Mass Balance studies between 1997 and 2005. These studies were intended to fill an important gap providing information on the use of materials in the UK. The studies were established to demonstrate the need for, and value of approaches to track the use of resources and the arising of wastes in the UK economy. In policy terms, the relevance of the mass balance studies were generally motivated by the need for a quantitative evidence base for resource use and movement, which would not guarantee, but make it more likely to work towards factor 4 or 10 improvements (in resource use) for sustainability. Not only due to the funding source, but also due to the general shift in attention in the debate towards environmental sinks, waste reduction received particular attention. To move the UK towards a more sustainable development path the need for action in two major areas was highlighted:

• Improvements in resource productivity were required , “such that far more value is added to each unit of resource that resides in the economy”.

• Improvements in resource recovery (recycling, reuse) were required , “such that material resources stay in use in the economy far longer than at present (Lindstead and Ekins, 2001: 9).

The general idea behind the Biffaward studies to promote reductions in pollution and waste through prevention, recovery and productivity improvements is also at the heart of the Government’s SCP strategy and has been on the SD agenda already for a while (DEFRA, 1999; DTI, 2000). In the course of the build-up of this evidence base the methodological framework was intentionally kept (much more) flexible to allow the addressing of policy relevant questions on the most appropriate level in the human as well as in the material sphere. In general, four dimensions for the tracking of physical flows were identified to be of interest for Mass Balance studies:

• “The quantities of basic materials themselves;

• The quantities of the major components and products the materials are incorporated;

• The flow of materials through the economic sectors that own the materials for consumption or production purposes, and which produce zero- or negative value-materials of waste;

• The region within which the materials are physically located.” (Linstead and Ekins, 2001: 10)

Depending on the purpose of the project, these four dimensions were combined in various ways. Studies exist that focus on the overall material flows of a region, the flows of a particular bulk material, product or substance within the whole economy or a particular sector of the economies and also Ecological Footprint studies based on material flow estimates. In contrast, to the methodologies we have reviewed previously, the Biffaward studies, therefore, do not all follow the same methodological procedure. In contrast, methodological flexibility to study resource use and address policy questions is the most characteristic feature of the Biffaward Programme and from a policy viewpoint probably also the most powerful idea. Therefore, the Biffaward studies in a way present a cross-section of the methodological procedures introduced earlier. In terms of the classification of MFA approaches, the studies might best fit into what has been called in other MFA reviews Bulk Material Flow Analysis (Daniels and Moore, 2002a; Daniels and Moore, 2002b, Daniels, 2002) or

Appendix SCP001 31/196 Material Systems Analysis. This is a group of MFA methodologies, which is the most heterogenous among the ones reviewed here and has not been standardised to any extent. “It is rather a composition of certain MFA applications and case studies that are not characterised by a unique methodology, but by a common meso-level of aggregation” (Femia and Moll, 2005: 34).

On the project level methodological flexibility was crucial, because policy relevance was reinforced within the Biffaward Programme through a particular tailoring of the funding requirements. Money was only granted, if a co-funding body was brought into the project to provide another 10 percent of the amount given by the Biffaward Programme. The idea was that this would guarantee policy relevance and partner involvement, as co-funding would only be provided, if projects were of direct value to these institutions/bodies. The information provided by the project would be directly used to design policies enabling the project partners to directly address the most resource inefficient and wasteful practises within their sphere of influence.

However, to allow for compatible research outputs and the establishment of a comprehensive cross- sectoral and geographical picture of resource flows in the UK, it was seen as a basic requirement to establish a systematic, regulated approach for the data collection process. Therefore, a guide to data coordination was commissioned by Forum for the Future (see Linstead and Ekins, 2001). This can be seen as a general research framework, which establishes some guiding principles for tracking physical flows. At their heart was the mass balance principle, but many other useful propositions to allow for some degree of methodological standardisation in data collection were put forward. Forum for the Future also took responsibility for co-ordinating, presenting and interpreting the data. In this course the projects’ outcomes were intended to feed into a well-designed database, maintained by Forum of the Future, which acted as a portal for all the material flow data collected in all of the mass balance studies.6 While Section 4 will focus on the evaluation of the Biffaward Programme in terms of its intention to provide policy relevant research, which was directly streamlined into policy action, and a robust material flow evidence base, which is of immediate use for policy makers until today, we will discuss here to what extent the data co-ordination guide provided sufficient instructions to the practitioners to foster comparable high quality MFA research. This is important for the final discussion in Section 4 as we will observe large fluctuations in quality of studies. Therefore, this analysis is key to learn lessons from the Biffaward Programme to carry the good practises into new projects.

Assessment The data co-ordination guide by Linstead and Ekins (2001) and its follow-up (Lindstead et al., 2003) intended to ensure consistency and compatibility of the data collection efforts in different projects of the Biffaward programme. Even though studies with potentially very different foci, the system boundaries were well defined in the human as well as in the material sphere. This was mainly facilitated by the provision of detailed instructions on how to classify the data temporarily, geographically, economically as well as from a material perspective.

6 Here it needs to be said that no material flow data really existed at that time in the UK and not a lot of interest was shown on the political level to change this.

Appendix SCP001 32/196 Consistency with a variety of established national and international classification systems was key for this achievement. For an unambiguous definition of the studies geographically, the standard administrative boundaries in the UK from the national to the local authority level was suggested.7 The guide further advised to classify industrial sectors/industries according to the standard industrial classification (SIC) and products according to the Combined Nomenclature (CN). CN was preferred to the Standard Industrial Trade Classification (SITC) as well as the Harmonised Commodity Description and Coding System (HS) due to the higher level of detail. This is seen as a good choice, because all three classification systems are fully compatible and it is therefore wise to pick the most detailed one to allow maximum flexibility for potential future conversion needs. The CN classification was further complemented by the Environmental Agency’s UK Waste Classification scheme to allow inclusion of product categories that are not included or are more comprehensively defined than in CN.

In the material sphere “no systems for material classification were found to be widely accepted by either industry or the agencies involved in waste management” (Linstead and Ekins, 2001: 17). Therefore, the authors devised a new classification scheme for the mass balance data framework, “which differentiates materials from products and which provides a hierarchical structure for the classification of materials.” It is our view that it would have probably been preferable to follow the material classification developed in the Eurostat guide for Economy-Wide Material Flow Accounting (Eurostat, 2001), which has been established as a European standard and entered the international arena as part of the System of Integrated Environmental and Economic Accounting (UN, 2003). However, as long as this classification system is usable, sufficiently comprehensive and adopted in the various projects of the Biffaward programme, no problems at all arise for the compatibility of the data collected by the various projects of the Biffaward programme. The only problem with this classification scheme for the mass balance scheme was that is was not fully defined in the guide, because the distinction between materials and products required knowledge of the process of manufacture. This made a complete definition impossible before the framework was implemented and a definition and classification of materials is carried out on a project basis in consultation with the project researcher. It is unclear whether or not such a consultation has taken place in all projects. However, a not fully defined material classification does not contribute to comparability and integratability of results. Despite the good practise in all other areas only a satisfactory assessment has been given for the consistency with standard classification systems.

Consistency with the basic SEEA concepts as the relevant international accounting standard was not sought. However, a satisfactory assessment could still be provided due to the fundamental importance of the mass balance principle within the Biffaward programme and the accounting instructions that helped avoid double counting (see Linstead and Ekins, 2001: 3.3.1 – 3.3.3).

The robustness of the methodology also received a satisfactory assessment. A good assessment could not be awarded as too many decision were still left to the practitioner. As mentioned in the instruction this was partially necessary due to the wide scope of the programme. However, there are many practical problems material flow practitioners are commonly confronted, which were not addressed within this general framework (see Eurostat, 2001). Even though the boundaries were well-defined on

7 For even further disaggregation the guide suggests the use of the official postcode areas for an unambiguous definition of the spatial study boundaries.

Appendix SCP001 33/196 a general level, it is, for example, important to decide whether or not (changes in) physical flows associated with certain material stocks like landfills or agricultural stocks of plants were included within the economic boundaries or not. No indication was made how different classification schemes like CN and the UK Waste Classification could be integrated and how flows in the integrated categories could be consistently coded. Most importantly no advice was given with regard to the availability of physical data in the UK, i.e. which suitable data sources exist from which the various projects could start from. Therefore, it was also not identified were potentially data shortages might exist and how they could be overcome. Finally, the framework did not ask for any sort of sensitivity analysis to test the assumptions made in the course of the data compilation in order to provide an idea about the reliability of the results. This could have been expected from a project that tries to integrate a variety of different data collection efforts. The general soundness of the method once good practises are established is reflected in some peer- reviewed publications, which have emerged from the Biffaward programme. Due to the framework’s dependency on good practice a satisfactory assessment was assigned for the design of the method to obtain compatible8 results. In summary, it could be said that the guide for data co-ordination provided a sound general framework without considering the practical problems, which might arise within the various studies. However, these practical considerations are of importance as well, if a consistent and compatible research database is to achieved.

8 This change in language from comparable to compatible is intended, because the different studies are too different in scope to provide comparable results almost by definition.

Appendix SCP001 34/196 I.3.3 Analysis of Material Flows by Sector: NAMEAs, Generalised Input- Output Models, and Physical Input-Output Analysis

Description Input-output tables could be described as a disaggregated national production account on the meso level. They try to depict the interconnectedness of economic activity within a defined reporting period in a matrix format. There are two types of input-output tables: (1) Monetary input-output tables (MIOT); (2) Physical Input-Output Tables (PIOT). Both share the same general structure, which is shown in Box 0.3, and both can be used to perform environmental input-output analysis (EIOA). The general structure of input-output tables is shown.

For the analysis of material flows input-output tables can serve as the basis for model estimations. Combined with NAMEA type sectoral environmental accounts data (see UN, 2003) they can be used to model the pathways of physical flows through the economic network from entry to release. Some of the key characteristics of input-output are: • high degree of sectoral detail - allows the detailed study of sectoral supply-chains and resource productivity; • production and consumption entities unified in a systemic approach - goods and services delivered to final demand entities can be traced back the production-consumption chain in a lifecycle perspective; • based on different sets assumptions full resource use required in the rest of the world to provide for domestic imports can be calculated. This is addressed in detail in Section 6 of this report. EIOA models can then answer questions such as how much resources are used by a certain sector, a certain final demand category, a certain item in the household or government consumption basket directly and from a lifecycle perspective? Where are the most resource intensive nodes in the supply chain? How does sectoral resource use relate to employment, GDP generation, or taxes and subsidies raised/provided by the government? How much environmental burden is shifted to other countries through the trade relationships? Does it seem to be true that pollution intensive goods and services tend to be imported?

In the context of these questions the strength and main characteristic of input-output models is that they can take into account the direct resource use of economic activities and all the indirect resource use occurring higher up in the supply chain in a lifecycle perspective. Moreover, based on various sets of assumptions, which are outlined in detail in Section 6 of this report, direct and indirect resource use associated with imports can be estimated as well. These questions directly feed into the Government’s SCP framework and strategy (DEFRA, 2003; DEFRA, 2005a), where lifecycle views and international considerations in terms of burden shifting are highlighted.

Appendix SCP001 35/196 MIOTs are restricted to the flows of goods and services within the economy. The n different industries use a selection of the k different primary inputs such as labour and capital and combine them in their production processes with intermediate goods and services (as well as imports from ROW). This is recorded in the first n columns of the input-output matrix adding up to total inputs. The goods and services produced by the industries are then sold either to other producing sectors as intermediates or to the l different final demand categories, where they are either consumed by households or governments, added to the fixed capital stock or exported to the rest of the world. This is recorded in the first n rows of the input-output matrix. By definition for each sector the total input must equal the total outputs.

A very similar story line can be developed for physical input-output tables. The only difference is that the coverage of the tables does not only comprise the circular flows of goods within the economy, but also all the physical exchange with the natural environment – all measured in tons (instead of money). In this sense, it could be said that PIOTs depict the industrial or societal metabolism in a matrix format. From a physical perspective the n different industries use primary material inputs such as raw material inputs from the domestic environment, inputs from the domestic capital stock inputs as well as imports from the rest of the world in combination with intermediate goods purchased from other industrial sectors. The first n columns of the PIOT matrix therefore give the total material inputs taken up by each industry. They sell these goods as intermediates to the other sectors or as final products to the different final demand entities such as households or the government. Others add to the capital stock or get exported to the rest of the world. However, during all these production processes not only goods, but also pollution and wastes are created. They are ultimately released to the environment even though some of them might be temporarily stored within the economy – for example in landfills.9 Both releases to controlled landfill and the natural environment are separately recorded as final demand categories. The columns of a PIOT therefore record the total material output of sectors. This output comprises goods and bads.10 The material balance principle then requires that total sectoral material inputs equal total sectoral material outputs. PIOTs are the only physical accounting tool, which provides complete sectoral material balances.

Box 0.3 - An introduction to monetary and physical input-output

9 The reader might ask herself, whether or not recycling is covered by PIOTs. The answer is “yes” and will be discussed later in the text. 10 There has been an active debate during the last three years how to treat wastes in modelling efforts. For an overview of this debate see Dietzenbacher (2005). However, the debate will also briefly covered here later on.

Appendix SCP001 36/196 Although there are some methodological differences, MIOT and PIOT based models address very similar policy questions.11 The question of choice between MIOT and PIOT based models very much depends on two interrelated issues: the methodological robustness and associated differences in answers to policy questions they provide. While leaving this discussion for the methodological assessment some terminology needs to be established. When the term “environmental input-output analysis” (EIOA) is used, both MIOT and PIOT based approaches are referred to. When models based on PIOTs are referred to, the term physical input-output analysis (PIOA) is used. Finally approaches integrating monetary and physical data are termed ‘generalised input-output analysis’ (GIOA) (see Miller and Blair, 1985).

• What are the absolute and relative direct contribution of the individual industries to environmental pressures? • What are the technological, and behavioural driving forces behind changes in the use of resources and the generation of wastes? • How resource efficient are the different industries? • Which resources, and in what quantities, is consumed by an industry? How much emissions and waste are released by different institutions? • Where do the most significant resource flows occur in the supply in the supply chain of a certain product (group)? • Which product groups drive waste flows associated with household and government consumption and where might be the best intervention points in the supply chain? • What are the total CO2 emissions associated with services? • How does inequality in a society contribute to resource use patterns? • Is the UK economy rather a through-flow or a cyclic material system? Is the flow behavior of materials in the British economy increasingly cyclic?

Box 0.4 - Policy questions EIOA

Assessment - Methodology EIOA is largely dependent on the data provided by National Statistical Offices due to the large amount of resources that are needed for the compilation of input-output and sectoral NAMEA type environmental account data. Geographically the system boundaries are therefore restricted to the national level in the UK, but could be applied to any other well defined system as well.12 The same holds for the material sphere. For the rare case of self-compilation efforts clear advice on system boundary choice can be found in the SNA93 (UN, 1993). For PIOTs there is still an on-going debate over which processes should be included within the boundaries of the material system. The German tables, for example, include the whole biological metabolism of plants or landfill wastes within the economic boundaries while these flows are usually not included in the tables for other countries (e.g. Pedersen, 1999).

11 There is some unexplored analytical potential rooted in the fact that PIOTs are the only analytical tool which allows fully tracing the material transformation processes from raw material to goods to wastes and residuals. Moreover, PIOTs cover recycling processes more comprehensively even though MIOTs can be adjusted to do so as well. 12 See www.inforum.org

Appendix SCP001 37/196 The compilation of MIOT and NAMEA type accounts are standardised internationally through the basic concepts and classifications defined in the SNA92 - nationally through the blue and pink book. For the compilation of PIOTs such standards do not exist yet even though the available prototypes usually closely relate to SEEA concepts (see Stahmer et al., 1998). In the literature the German and the Danish approach seem to serve as role models. The correspondence with standard classification systems provided by the international accounting community is a generic feature of all these economic and environmental accounting efforts.

It is a unique feature of PIOTs that they establish material balances at the sectoral level. The required procedures have the positive side-effect that data gaps in the environmental accounts are revealed. The compilation of PIOTs therefore helps to increase data robustness of the environmental accounts data in general (see Stahmer et al., 1998). This is one major reason why PIOTs are collected in Germany and serve as the foundation of their environmental accounting framework called MEFIS.

While the data robustness does not seem to be of great concern due to the national and international accounting standards, the method robustness need to be discussed in some more depth. Most relevant in this context is the actual imputation of the physical structure of the economy by input-output models. In an ideal world PIOTs are fully derived in a bottom-up approach from physical data and do not require any such imputations. In practice this is not the case. For most sectors the use structure is derived from monetary data even though an extremely low level of aggregation is used (see Stahmer et al., 1997; Pedersen, 1998) – in the case of Germany more than 5000 product groups are mapped into more than 100 industries. On such a low level of aggregation the relationship between the value of goods and their weight is very close. Moreover, for price differentiation was controlled for, where possible. Even though this is not fully the case in the real world (see Stahmer et al., 1997; Pedersen, 1998), it might be still be assumed here that PIOTs are completely compiled from physical bottom-up data for the sake of the discussion.

GIOAs in contrast need to impute this structure based on the simple assumption that for each sector the same amount of resource use/waste occurs per unit of product/output delivered to intermediate and final demand entities. This assumption is only justifiable, if the output per sector is sufficiently homogenous. Statistical offices usually try to ensure a maximum homogeneity through their accounting procedures. However, most important is the aggregation level of the available tables. The simple rule is: the more detailed the table the smaller the error introduced.

Hence, as they are based on real rather than modelled data PIOAs are more robust than GIOAs in a first instance. However, it has already been mentioned that PIOTs are usually compiled in aggregate units – for the production structure these are total tonnage of intermediate products circulating among the sectors. The SCP agenda is mainly interested in flows of specific substances, resources, wastes or pollutants. Even though PIOTs can also be established on lower aggregation levels (see Pedersen, 1999), this is an even more work intensive task and not always very likely to happen. In these cases also for the PIOT certain resource and waste flows need to be imputed via a similar estimation process than for GIOAs using NAMEA type environmental account data. In this light the question of method robustness is not so clear cut and depends on the aim of the analysis. This comes down to whether the flow of resources and wastes can be better imputed using the physical or the monetary information on the production structure. No author in the debate seems to

Appendix SCP001 38/196 think that monetary information is more suitable as a basis for estimation. This is very intuitive as the amount of money spent on a unit of output might vary for different buyers at different times during a reporting period. This creates a need to take into account price differentiation in the supply chain as well as price fluctuations during a reporting period; both could lead to significant differences and physical data is not prone to these complications In addition industries may sell very different things to very different sectors. This raises the issue of heterogeneity again. In such cases physical data may provide the more reliable basis for imputations. As a rule of thumb this may be truer the smaller the ratio between the resource flow under consideration and the sectoral output (measured in tons). For example, flows of CO2 might be imputed more robustly using physical tables than low volume flows like mercury.

Some people do think that there is no significant difference between the production structure in a monetary and a physical table and that one can be converted into the other by using an appropriate price vector. Dietzenbacher (2005) has shown why this argument does not apply at the typical aggregation levels at which input-output tables are usually presented. It could be concluded therefore that PIOAs are more robust than GIOAs.

Despite these issues the vast amount of GIOA literature indicates a sufficient level of confidence in the research community into the reliability and robustness of the method for such ex-post analysis. In the light of the unavailability of PIOTs in the UK and the relatively high costs associated with their compilation, the question of choice between GIOA and PIOA models for informing policy comes down to whether or not they lead to significantly different policy implications. This question has not been comprehensively answered in the literature. The debate surrounding PIOTs is still very young and has so far mainly focussed on methodological issues surrounding the treatment of wastes in physical input-output models. For the reasons discussed here we have preliminarily rated the method robustness of PIOTs with good and then one of MIOTs with satisfactory but more research in this area needs to be done. Germany and Denmark could provide the required database information which would be very helpful in deciding whether or not a PIOT for the UK should be established. If PIOTs do provide significantly better estimates they could greatly contribute to the policy relevance of environmental input-output methodologies. In fact, as soon as this is established ONS would be generally willing to start a development programme (Francis and Gazley, 2005).

In the light of the general lack of resources for the timely compilation of the MIOTs (see below) it is therefore recommended to wait with the compilation of tables until their importance for informing policy has been established.

Appendix SCP001 39/196

Assessment - Data Before discussing the data quality and availability in the UK, two things need to be noted first. The PIOT needs to be excluded from this category, as there is none available for the UK. A discussion of the required and available data sources for compiling a PIOT would go beyond the scope of this report. The available data for carrying out GIOA will be assessed for material flow data and input- output data separately. However, we take the view that a method can only be as good as its weakest element. The two data components required for GIOAs are the MIOT as well as the NAMEA environmental account data (Vaze, 1998). The data situation for MIOT and NAMEA could not be more different. While NAMEA data is released twice a year together with the environmental accounts always for the reporting period T-1 with T being the current year, the last MIOT was published in 1995. However, the publication of the MIOT 2000 is overdue since 2005 even though the UK is committed to Eurostat to deliver such statistics timely. Usually MIOTs get published every five years in the UK with the most recent table available for T-5 with T being the current year. Building analysis on tables, which are always between five and ten years old – currently even older – might not be seen as an ideal situation and certainly does not improve the robustness of the results.

It is one strength of GIOAs that they can estimate where substances, materials or aggregate resource inputs are used and where waste and pollution arises within the supply chain. This is information with direct relevance for the Government’s SCP agenda. However the assumption that resource flows and waste streams can be approximated by using a production structure that is 11 years old is a brave one. The flows among sectors and institutions will be more and more biased with increasing age of the input-output data. For increasingly outdated data this also means that the more detailed information one wants to obtain from input-output estimations, the less confidence one can have into the estimates.

The publication frequency just fulfils the criterion for a satisfactory rating but this is particularly disadvantageous, as most of the old tables available are not consistent with the concepts and classifications of SNA9313 (see UN, 1993). In fact only one table published for 1993 is consistent with the 1995 one. The older publications can hardly be used at all for longitudinal analysis without

13 As explained earlier, SNA93 is the international accounting standard applied in current ONS publications

Appendix SCP001 40/196 increasing the aggregation level substantially. An available time-series of tables could be used for example to study thoroughly the effect of certain policies on resource use like the invention of a new or the revision of an old tax or subsidy scheme.

There are supply and use tables (SUTs) available for the UK. They are published bi-annually together with the national accounts. SUTs can be seen as a (more flexible) pre-form of analytical input-output tables (see, ONS, 2002) and can be applied for input-output estimations. However, the UK version shows three problems: First, the supply table is only available on a much higher level of aggregation. Second, in the supply tables some data is suppressed. Third, supply and use tables are recorded in different price systems. Therefore, the SUT publication is of no use for GIOAs as long as only publicly available information is used. 14 In the light of the data situation it is clear that the methodological potential of MIOTs in the context of MFA is not exhausted. In the UK only Stockholm Environment Institute and Cambridge Econometrics (see Wiedmann et al., 2006) have managed to update input-output data up to 2003 based on the SUT publication even though the resulting data will not be of the same quality than the one produced by ONS. It is therefore clear that currently the methodological potential of GIOAs for informing the government’s SCP agenda in a robust way is certainly not fully exploited. However, it is of greatest importance for the government to improve this situation. Different solutions with different amount of financial resources required are thinkable. • Encourage the provision of sufficient data material to easily update input-output tables for every year since 1992; • Encourage the over-due provision of the input-output table for the year 2000 and automatisation of table compilation; • Encourage the establishment of a time series of input-output tables as common in many other Eurpean countries as done for SUT publication; • If further resources are available or the policy need for physical tables is motivated, promote the construction of PIOTs; The NAMEA data only received a satisfactory rating due to their coverage of pollutants missing out some key ones for a good physical data coverage of SCP policies. This is particularly true for waste data, but also for water, land-use, toxic chemicals and transport (see Destatis, 2005). The environmental indicators included are shown in Table I.4.

14 Stockholm Environment Institute managed with help of Camebridge Econometrics to derive an input-output table for 2000, which is updatable to 2004. However, to our knowledge these are the only two institutions in the UK using IO data of more recent years in their resource flow models (see Wiedmann et al., 2006).

Appendix SCP001 41/196 Air Pollution Energy Green House Gases Natural Gas Acid Rain Precursors Coal Carbon dioxide Petrol Methane1 Derivatives Nitrous oxide2 Fuel oil Sulphur dioxide Gas oil Nitrogen oxides3 Aviation fuel Ammonia4 Other carbon based fuels Particulate matter Net electricity Carbon monoxide Volatile Organic Compounds Benzene 1-3 Butadiene Lead Cadmium Mercury Arsenic Chromium Copper Nickel Selenium Zinc Vanadium

Table I.4 - Environmental Indicators included in the UK's NAMEA publication

Assessment – Policy Relevance Input-output based MFA models interact with the specific and general SCP objectives very well and can provide significant contributions to a wide range of policy questions raised within the score card system. Due to this high level of responsiveness with the Government’s SCP agenda, the various

Appendix SCP001 42/196 contributions cannot all be discussed in detail. Instead the discussion will be organised according to three main features of input-output models15: 1. Detailed coverage of supply chain: Input-output tables like probably no other data source provide a detailed depiction of the flows of goods and services between sectors. This information can be used to model the associated flows of resources and wastes. As a characteristic feature of input-output models a distinction between on-site and off-site emissions is facilitated. Different modelling procedures such as structural path analysis (e.g. Treloar, 1997; Defouny and Thorbecke, 1989), key sector analysis (e..g. Lenzen, 2003), environmental flow analysis (e.g. Suh, 2004c) or field of influence analysis (e.g. Sonis and Hewings, 1999, Lenzen, 2003) can be used to identify the most important resource and waste flows in a supply chain, to measure the cyclic behaviour of flows, to identify key sectors in terms of resource use and the arising of waste or to perform comprehensive sectoral benchmarking (e.g. Lenzen, 2005). By doing so a whole array of policy questions can be answered such as is the British economy a throughflow or a cyclic economy? Can we observe an increasing closing of the loop in terms of resource use over time in the whole economy or in certain sectors? Where do the main wastes arise in the supply chain and where are most of the resources used up? Which are the sectors pollution prevention efforts might be most effective?

Manfred Lenzen (2005) and his colleagues from the Univeristy of Sydney carried out a benchmarking for all 135 sectors of the Australian economy following a triple bottom accounting approach. The difference to conventional triple bottom line assessments is their use of a variety of socio-economic and environmental performance indicators estimated from an input-output methodology. The positive side effect are not only an increased comparability due to the use of such a standard framework, but also that it allows them to assess the total impact in the various dimensions taking into account all upstream and down stream impacts.

The study is not only a good lesson in terms of how to enrich sector reporting through an input-output framework, but also in how the large amount of information created by the model can be effectively condensed for effectively inform policy makers. The spider diagram in Error! Reference source not found. shows the normalised figures of the headline indicator set. It shows, for example, that the aluminium industry uses 7 times more primary energy than the average Australian sector, but only causes a tenth of the average land disturbance. For experts than a lot more details were provided like the identification of the most important supply chain

15 Even though this distinction might not be fully clear cut, it helps to easily grasp the feedback of questions, which can be raised with GIOAs and PIOAs.

Appendix SCP001 43/196 nodes, which provided the basis for in depth consultation where resource flows could be tackled in the supply chain. Such an in depth, comparable sector benchmark for the UK economy would fit very well into the Government’s agenda (e.g. BREW programme) and contribute to strengthen the SCP evidence base.

Box 0.5 - Good Practice Generalised Input-Output Analysis 2. A Systemic view of the production-consumption system: It is another main appeal of (generalised) input-output models that they can treat the various institutions of production and consumption in a systemic approach. Their ability not only to take into account the resource flows triggered directly by the various final demand categories, but also the indirect resource flows required in the various supply chain layers to produce the goods and services, inherently links the various production and consumption entities. In a sense it could be said that for a certain reporting period GIOAs allow us to take a lifecycle view at the meso level. Due to the emphasise of the lifecycle principle on the government’s SCP agenda, there is a strong policy relevance in several areas. In particular this applies to demand side issues such as “consumption” and “procurement” and parts of the “waste” agenda. GIOAs allow us to ask questions like:

• How much resources are used up by local or national government consumption over the lifecycle of the purchased goods and services? • Which items in the government consumption basket contribute most in terms of resource use and wastes generation? • Where in the supply chain can preventive policy effectively start from? • What triggers the resource use patterns of households over the lifecycle? • Which consumption patterns across socio-economic groups are more and which less resource intensive and why? • Which product groups are the most polluting ones? In the same way more general questions can be asked such as what are the economy-wide and sectoral drivers behind de-coupling? By taking a consumption perspective of the economic process GIOA methods are also able to take into account the direct and indirect resource flows associated with the domestic consumption of imports based on a variety of assumptions (for a good overview (see Lenzen et al., 2004; Munksgaard et al., 2005). This also allows the international dimension of resource use and the associated burden shifting emphasised by the Government to be addressed. Questions that could be answered in this context include how much burden is shifted to the rest of the world or which of the traded product groups foster these patterns? 3. A Flexible tool, which allows for integration with various other methodologies and data: The biggest advantage of input-output models in terms of their policy relevance may be their flexibility in incorporating new data sources and integrating other methodologies. First of all the integrated monetary-physical data structure allows us to deal with a variety of socio-economic variables such as employment, GDP, gross revenue and others at the same time. This means that socio-economic variables are already linked with environmental ones, which make the models particularly appealing structurally for sustainable development inquiries. Without raising policy questions in this context the reader is referred to sector benchmarking example in Box 0.5 to get an idea of the potential triple bottom line nature of GIOAs.

Appendix SCP001 44/196 Further information can be added by entering new data sources. Social Accounting Matrices, for example, can be seen as monetary input-output tables, which are further disaggregated for categories of final demand and value added. By doing so the (re-)distributional flows of money can be taken into account. In a study for Wales a socially disaggregated environmental input- output model identified that certain socio-economic low-income cluster groups showed very high energy consumption. As they happened to live in council houses the council had opportunities to target these high energy consumption practices specifically (SEI, 2005).

GIOAs can directly feed into standard economic modelling approaches such as applied general equilibrium models (Kojima, S., 2005) as well as large scale econometric models (e.g. Meyer, 2005), which use monetary input-output tables as a direct data input. This is particularly interesting for policy analysis and allows questions like ‘which is the most effective policy instrument to implement a certain policy?’ and ‘how will the economy adapt as a whole?’ to be answered. In this context some of the restrictive assumptions applied to input-output models - such as those relating to fully complementary goods or no economics of scale - can be relaxed.16 GIOAs also link very nicely to other important environmental tools. As soon as sufficiently detailed NAMEA flows are available, GIOAs can be seen as SFAs approximated17 on a meso level. In this sense it even might help to reduce environmental impacts for substances that have been identified as harmful by the Government. In this context we can ask how much of a substance occurs in terms of specific lifecycle stages and production processes, product groups, sectors and activities to substance emissions, accumulations and concentrations in the economy (more detailed questions, which can be raised by SFAs and related methodologies are outlined in Section 4.3.5). In the case of PIOAs the difference with detailed SFAs fully vanishes (see, Pedersen, 1998). Whether or not the results are sufficiently different (and therefore more reliable) to justify the costs of establishing such tables however is questionable and would need to be determined on a case-to-case level.

GIOAs can be further linked with lifecycle methodologies to answer product specific policies such as ‘which of three products performs better environmentally in terms of a set of key pollutants?’ Or ‘where can resource consumption be best prevented in the life cycle?’. This integrated method is called Hybrid LCA (Treloar, 1997; Joshi, 2000; Suh et al., 2004), and will be discussed together with LCAs later. Once input-output methodologies are linked with SFAs and LCAs environmental risk assessment can also be easily related to the results, which opens up a whole new array of possibilities.

It is even thinkable that information about health, exercising, and objective as well as subjective welfare attributes can be introduced into an input-output context (see Keuning, 2000). These are issues that rank right at the top of the governments agenda and have rarely been addressed by quantitiative models. This would go another step further in the analysis of the potential environmental burden in its socio-economic context and would allow modelling efforts starting from such a data framework to provide more policy relevant outcomes.

16 However, it is important to understand that this is only important in the context of policy analysis not for the monitoring and analysis of past flows. 17 It has been explained above that GIOAs use the monetary structure as a basis for modelling intersectoral flows physically. SFAs observe these flows directly.

Appendix SCP001 45/196 The only critical comment about the policy relevance is that even though EIO models have prominently featured on the policy level their contribution to triggering policy action is not always clear. If policy makers have been reluctant to build on the advice of input-output methodologies, it might be related to the vast amount of information they generate. Practitioners have often not been very concerned about presenting them in an easily digestible packaged way as shown in Box 0.5. With the exception of some in the lifestyle movement that has taken off within the input-output

community more recently, authors have sometimes neglected to focus their analysis on specific problems like the supply chain of a particular industry and to provide more specific policy advice. Instead they have tended to focus on all sectors of the economy in a rather general way.

Appendix SCP001 46/196 I.3.4 Life Cycle Inventories

Description Life Cycle Analysis (LCA) is an MFA methodology, which allows decision support in product related policies. Based on a functional system boundary definition LCAs identify and compare physical flows and associated environmental impacts of a product system through all stages of its lifecycle form the cradle to the grave (see Femia and Moll, 2005; Daniels and Moore, 2002; EEA, 1997).

Figure 0.8 - The four discrete stages of a life cycle assessment

LCAs studies often compare the environmental performance of a small number of substitutable products, which provide similar consumption services and therefore provide information for environmentally sound choices. More recently they have also been increasingly used in product design and development processes and for optimisation of production processes associated with a particular product system (see Millet et al., 2006, Wright et al., 2005; Nielsen, P.H. and Wenzel, H., 2002; Saouter and Van Hoof, 2002; Tan and Culaba, 2002; Azapagic and Clift, 1999: Ekvall et al., 1998). LCAs are typically carried out in four discrete, but interrelated stages as shown in figure X. Each phase is briefly described below: • Goal and scope definition: Goal and scope definition is where the LCA is planned with users setting the framework for how the study will be undertaken. • Life cycle inventory analysis (LCI): Life cycle inventory analysis is where the “system” (i.e. those activities under study) is modelled and data gathered and used to compile the data for each part of

Appendix SCP001 47/196 the system modelled. The latter tends to be in the most time consuming phase accounting for 80% of the effort (and cost). Life cycle inventories concentrate on quantifying the use of raw materials and energy as well as the associated releases to air, water and land. • Life cycle inventory impact assessment (LCIA): Life cycle impact assessment is where the result of the life cycle inventory analysis are “converted” into measures of environmental impact; for example, global warming potential; air acidification potential, etc.. • Life cycle interpretation: Life cycle interpretation is the stage when the results cmpiled are examined to fulfilling the aims of the study One of the main adantages of the interpretation phase is that it allows users the identification of those activities, which might be best targeted to bring about the best greatest improvement in performance. Even though it is of importance to provide a general overview of the full method as done above, the initial focus of this report is on physical flows. The corresponding information are compiled in the LCI stage and the discussion will therefore be focussed on this part of the LCA methodology. In later stages some general remarks on impacts assessments will be made (see Section X) and some implications for LCAs will be derived from this discussion. As implied by the name, LCIs try to build-up a systemtic and comprehensive inventory of all physical flows associated with a certain product system. LCI thus involve the identification and quantification of physical inputs and outputs during each life cycle stage. The holistic nature of the lifecycle approach requires not only to consider the direct flows associated with the various life cycle stages, but ideally also the indirect flows occuring in production layers higher up in the supply chain (see Guinee, 1993). Based on these inventories a variety of policy questions relevant for the government’s SCP agenda as shown in Box 0.6.

• Where are (important) physical flows triggered along the lifecycle of a product? Which phase – extraction, processing, distribution, use, wasting – is most relevant for policy intervention? Where are the most relevant policy intervention points? • Which of two similar products is environmentally preferable in terms of the physical flows triggered across the lifecycle? • Which materials should be used to maximise the environmental performance of a product? • In which stages of the lifecycle of a product are the largest potential for maximsing the environmental performance of a product? • How can the life cycle impacts of a product be reduced by design?

Box 0.6 - Policy questions lifecycle assessment

Assessment - Methodology After a long period of standardisation, the methodological framework of LCA can be seen as fully standardised today. The framework is meticulously recorded/reported in the ISO 14040+ series published by the International Standards Organisation as shown in Box 0.7. The LCI methodology is covered in ISO 14041 together with Goal and Scope Definition (see Figure 0.8). Based on clearly defined goals and scope, the data compilation processes consists of six individual steps: • Data collection • Refining system boundaries

Appendix SCP001 48/196 • Calculations • Validation of data • Relating data to the specific system • Allocation

The International Organization for Standardization (ISO) is a worldwide federation of national standards bodies (ISO member bodies). ISO technical committees produce international standards on a variety of topics. Draft international standards adopted by the technical committees are circulated to member bodies for voting. Seventy-five percent of the member bodies voting must approve the Draft International Standard in order for it to become final.

The ISO 14000 series relates to numerous facets of environmental management. ISO 14040 – 14043 were prepared by Technical Committee ISO/TC 207, Environmental Management Subcommittee SC 5, Life Cycle Assessment. While ISO recognizes that LCA is still in a nascent stage of development, ISO 14040-14043 is a consensus-based, voluntary set of standards pertaining to LCA. ISO 14040 provides information on LCA principles and framework, while ISO 14041- 14043 provides additional information regarding the various phases of LCA. The standards are designed to guide the practitioner or analyst and are not legally binding or enforceable. They attempt to bring some consistency and credibility to the field as it emerges and takes shape.

ISO 14040 – Environmental management – Life cycle assessment – Principles and framework: Specifies the general framework, principles, and requirements for conducting and reporting life cycle assessment studies, but does not describe the life cycle assessment technique in detail. ISO 14041 – Environmental management – Life cycle assessment – Goal scope and definition and inventory analysis: Specifies the requirements and procedures for the compilation and preparation of the definition of goal and scope for an LCA and for performing, interpreting, and reporting a life cycle inventory (LCI) analysis. ISO 14042 – Environmental management – Life cycle assessment – Life cycle impact assessment: Describes and gives guidance on the general framework for the life cycle impact assessment (LCIA) phase of LCA, and the key features and inherent limitations of LCIA. It specifies requirements for conducting the LCIA phase and the relationship of LCIA to other LCA phases. ISO 14043 – Environmental management – Life cycle assessment – Life cycle interpretation: Provides requirements and recommendations for conducting the life cycle interpretation in LCA or LCI studies. It does not describe specific methodologies for the life cycle interpretation phase of LCA and LCI studies. ISO 14047 - Technical Report Provides illustrative examples on how to apply life cycle assessments. ISO 14048 - LCA Data Documentation Format Provides guidance on factors to consider when documenting LCA darta. ISO 14049 - Technical Report Provides illustrative examples on how to apply goal and scope definition and inventory analysis.

The current revision of the ISO standards will merge the ISO 14040, 14041, 14042 and 14043 into two standards: ISO 14040 with the general framework and ISO 14044 with all the requirements gathered in one standard. Also the data format standard ISO 14048 is currently under revision. However, there is still a need for further harmonisation of protocols, beyond the ISO 14040-series of standards, to promote sound practices and enhance comparability and data exchange between LCAs (see LCA consultants, 2005).

Box 0.7 - Overview ISO LCI/A standards

Appendix SCP001 49/196 As mentioned previously, ideally LCIs would aim to collect a complete set of physical flow data associated with a particular product system. In practice, this is an unmanageable task due to the complexity and the global interconnectedness of the production processes. Data collection processes must therefore focus on a carefully chosen set of economic processes covering the majority of physical flows. A careful procedure for setting and permanently re-defining system boundaries is outlined in ISO 14041. However, Lenzen (2001) calculated that for most commodities, direct energy requirements account for less than a quarter of total energy requirements. Process analyses including 132 first-order inputs carry truncation errors that are mostly above 50 % and accounting for 17 424 second-order input paths generally carry 30 % truncation error. Hence, even though an average rating is given for the system boundary setting, these system boundaries might still miss a considerable share of the physical flows associated with a specific product system. It can therefore be suggestetd that: • LCIs consistently under-estimate the “true” physical flows of a system and must therefore be seen as lower-bound estimates (see Nielsen and Weidema, 2001); • this under-estimation seems to be (potentially) significant and tends to be higher for product systems with complex supply chains as well as • even though the ISO standards provide guidance in establishing good practice, the system boundaries will differ from study to study. Therefore, cross-study comparisons of results can only be undertaken with great care. This is particularly relevant for the government’s build-up of LCA databases of product-specific information services for private, public and corporate consumers.

Figure 0.9 - Incomplete Coverage of the Supply Chain by tradtional process based LCA methodologies (Source: Lenzen, 2001)

Appendix SCP001 50/196

For LCIs there is no requirement for an establishment of a mass balance. This is mainly due to the size and complexity of the system under consideration and the associated problems posed for data collection. It still leaves out an important physical consistency check of the data.18 However, in terms of the credibility of LCI it is one of the biggest development from the SETAC to the ISO standard that sensitivity analysis has become an integral part of the methodology. They are carried within the inventory phase as well as at the end of the full LCA in the course of the “interpretation” phase (see Figure 0.8).

LCIs (and LCAs) are well established in the literature and the scientific soundness of the ISO-type LCIs has been revealed indirectly through successful peer-review processes many times. It can therefore be concluded that even though there are some structural problems in establishing complete life cycle inventories in a bottom-up procedure, the LCI methodology has been successfully developed during the last decade. Due to the system boundary problem physical flow inventories are not complete and results from studies can only be compared with great care.

Even though LCA is now guided by this comprehensive set of ISO standards, there still remains a large degree of flexibility for conducting LCIs. On the positive side this provides a large degree of flexibility for serving different policy purposes (see Van Berkel, 2002).

Assessment - Data The LCI methodology is rooted in a bottom-up data collection procedure as previously established. Therefore data coverage, quality and robustness ultimately depend to a large extent on the established practice used in individual studies, i.e. what data sources are used, how the data is handled, how the system boundaries are chosen etc.. In the light of the immense literature – there are several journals (exclusively) focussing on LCA/LCIs methodologies - it cannot be in the scope of this report to review all individual studies. Based on samples from this literature some general data issues can be highlighted, which are of relevance for the collection of LCA information for the government’s LCA programme/evidence base (see ERM, 2005).

It should be highlighted first that substantial progress has been made in terms of data quality and robustness of the LCI methodology. Especially with the establishment of the ISO 14048 data

18 This does not mean that mass balances are never established in the course of LCIs. However, they are not required by the ISO standard and often left aside.

Appendix SCP001 51/196 documentation format in LCIs, the use and exchange of LCI data has certainly become easier. However, the actual data collection methodology is not as well established under ISO. Therefore a common protocol based on ISO 14041/49 is currently discussed covering data collection methods, treatment of missing information, specification and interpretation of data formats and nomenclatures, system boundaries, allocation procedures, consistent reporting of flows, and validation procedures (see LCA Consultants, 2006). A proposal for such a protocol is provided by the procedural guideline from the CASCADE project (see Carlson and Palsson, 2003).

It is also important to understand that data coverage and quality depends on the duration of the study and cost of data collection and collation (see Van Berkel, 2002). This is true not only for LCI/As, but all bottom-up data collection processes. LCIs often depend on existing LCI databases for compiling a physical inventory of a product system. These are often used for non-site specific inventory data of standard processes (e.g. transportation, packaging etc.) as well as when data shortages are experienced (see EEA, 1997). Even though significant improvements have been made in the quality of these life cycle databases, they are not easily comparable. This makes it more difficult to compare results from LCIs and has implication for the government’s on-going efforts to compile LCA evidence. For the compilation of LCA evidence for SCP consumer information services (see ERM, 2005), it is important to document which LCI database are being used. Even though this information is not relevant to the user of the information service, it allows for a better control of data quality and comparability. It may therefore be useful to commission a review of existing LCI databases in order to select the ones which provide sufficiently robust data to inform SCP policy in the UK.

Assessment - Policy Relevance LCIs have been frequently carried out for decision support of public policies and private companies (e.g. see Cooper et al. 2005; Tan and Culaba, 2002; Van Berkel, 2002; Europen, 1999; Vigon et al., 2003). According to the ISO standards the main applications of LCI data in decision processes comprise product development and improvement, strategic planning, public policy making and marketing. Table I.5 summarises policy applications for governments and private companies, which can be informed with the results of LCI studies.

LCI/A User Application Example Community Action Plans Incineration versus Recycling Public Transport Systems Government Environmentally Conscious Cars, Office Supplies, Food Public Purchase Consumer Information Ecolabels & Standards Establish Environmental Focus Identification of Areas of Improvement Product-Oriented Environmental Policy Environmental Management Concept Selection Company Design Choices Component Selection Material Selection Process Selection Environmental Documentation ISO 14000 Certification, Ecolabels

Table I.5 - LCA/LCI applications for government and companies (Source: Wenzel et al., 1997)

Appendix SCP001 52/196

In relation to the government’s SCP agenda, LCIs can valuably inform decisions in various areas. For example, results can be applied in areas such as sustainable waste management or integrated product policies. This is because LCIs provide specific information on where raw materials enter the human sphere throughout the life cycle of a product and where wastes arise and therefore help to identify policy intervention points. This information can be used by the government to set product standards, identify and encourage good practice or even to motivate bans of product systems with particularly detrimental effects on the environment and human health. The specific information provided by LCIs can also be applied by companies to improve the environmental performance of a specific product system. This usually includes various aspects of integrated product policy from process technology selection to process optimisation to process design. By making it possible to identify where the use of resources can be best reduced in the life cycle through the support of design and technology solutions, LCIs play an important role on the micro/process level and the move towards a closed loop economy. This directly feeds into the government’s green innovation agenda where LCI can help in the comparison of existing and new designs.

Objective Application Support for Decision Product Development Background for environmental specifications: design strategies, principles and rules. Identifies most important environmental Diagnosis Ecolabelling properties for product category. Community Action Plan Identifies environmentally important product groups. Product Development On-going identification of the best choices from alternative solutions. Cleaner Technology Identifies the best available technology by means of LCA. Selection Community Action Plans Identifies the best community strategy for a certain problem or product Consumer Information Documents potential environmental impacts from a certain product

Table I.6 - Applications for LCI/A data (Wenzel et al. 1997)

There is a further array of applications of LCI data in the empowerment of consumers through the provision of information. Eco-labelling schemes and comprehensive physical inventories for procurement decisions are an important component in the government’s plans to enable private, public and corporate consumers to make better choices. In the capacities outlined above, LCIs also contribute to the general SCP objectives.

Appendix SCP001 53/196

I.3.5 Substance Flow Analysis

Introduction Substance Flow Analysis (SFA) is a methodology, which sets out to trace the flows of a single or a group of (chemically defined) substances between the economy and the environment within a predefined system – usually a geographic region. A substance might be defined as any chemical element or compound composed of uniform units. All substances are characterised by a unique and identical constitution and are thus homogenous. In this context both chemical elements and compounds are correctly addressed as substances. This definition does not require a substance to have a shape, as it would not serve the general aim of SFA to trace the fate of substances through a defined system (see Brunner and Rechberger, 2004). SFAs generally start from a certain environmental problem, identify related “problem” substances and the relevant geographic territory. Problem substances are typically toxic or other hazardous physical flows of low volume, but with a potentially high environmental impact. Good examples are lead, chromium or heavy metals. For SFAs the spatio-temporal context is usually of direct relevance. SFAs follow the pathways of these substances through the system from origin to destination and identify where they accumulate.

Figure 0.10 - A schematic concept of SFAs (Brunner and Rechberger, 2004)

Appendix SCP001 54/196 SFAs do not usually restrict themselves to the human sphere and often also account for the resulting biological transformations and resulting concentrations in the environment (see Bringezu, 2000). In addition SFAs often enable the assessment of the impact of certain substance flows in the human sphere as well as in the environment. Results from these analyses often directly feed in other quantitative assessment, such as those associated with the risks imposed to substance-specific endpoints. SFAs therefore provide information for an overall management strategy of problem substances. Typical general policy questions, which can be addressed by such a management strategy are summarised in Box 0.8.

• Through which economic activities does a particular substance enter the human sphere? Where is it released back to the environment? • What are the pathways of a substance through the defined system? What are the most relevant production processes and in which products do harmful substances end-up? • Where are the most effective policy intervention points for eliminating, reducing or minimizing a particular substance in the supply-chain? • How does a specific substance contribute to a particular environmental problem?

Box 0.8 - Policy Questions - Substance Flow Analysis

Assessment - Method The methodological framework of SFAs has not been standardised yet (see Udo de Haes et al., 1997; Daniels and Moore, 2002a) even though leading researchers have shown increasing efforts to initiative such a process. In the literature the methodology proposed by Bancini and Brunner (1991) and its practical up-date by Brunner and Rechberger (2004) have increasingly enjoyed a paradigmatic or archetypical status in the research community. 19 Their approach has often been referred to as substance flux analysis. In this review these two studies will serve as the foundation for the assessment of the methodological soundness and robustness. However, it is important to mention that even though the authors are usually referred to in the context of SFAs, their own vision of the approach sees the analysis of substances embedded in the analysis of material flows. This will be further discussed later on in the report. The SFA methodology is designed as a flexible bottom-up approach, which can generally be applied to any set of system boundaries in the material as well as the human sphere. On the one hand this flexibility allows practitioners to tailor SFAs appropriately towards any environmental problem. On the other hand, SFAs become increasingly dependent on good practice in drawing considerate and unambiguous system boundaries as a standardised a priori definition cannot be provided. This problem can be related to the identification of relevant substances or to the choice of spatial or institutional boundaries depending on the aim of the project. Even though this problem cannot be fully resolved Brunner and Rechberger (2004) provide clear advice how these problems can be appropriately dealt with as depicted in Figure 0.11.

19 In order to avoid referencing both sources separately we introduce the abbreviation Brunner et al. (1991, 2004)

Appendix SCP001 55/196

Figure 0.11 - Good practice: working procedure SFA (Brunner und Rechnberger, 2004)

While substances are clearly defined by their chemical constitution, there is no common way of classifying human activities in SFAs. This is directly related to the lack of standardisation and the flexible, project specific focus of SFAs. In fact, the appropriate classification of human activities according to established classification system is not even mentioned by Bancini and Brunner (1991) as well as Brunner and Rechnberger (2004). This may not prove a problem for the individual study but certainly provides problems for the comparability of results and compatibility of study results. The lack of standardisation and classification in general also implies that there are no established links to other international standards such as the System of Environmental and Economic Accounting (UN, 2003) or the System of National Accounts (UN, 1993). Hence, the further integration with other quantitative frameworks from other social sciences becomes difficult. For decision-making this has negative effects in terms of the ex-post policy usefulness of studies

As highlighted earlier the robustness of the method highly depends on the good practice of the researcher. General consistency in the quantification of substance flows is ensured through the establishment of a mass balance for goods, which is the basis for a balancing of substance flows.

Appendix SCP001 56/196

Assessment - Data Even though an assessment of the various data sources is not possible as the method itself rather than individual studies are evaluated, typical data sources for SFAs can still be summarised and discussed.20 Typical data sources included in SFAs comprise: • National commodity statistics (production, trade); • Literature (encyclopaedias of chemical technology, branch specific literature, environmental research reports etc.); • Surveys (e.g. Private companies dealing with the actual substance or products containing the substance); • On-site measurements (e.g. industrial emissions, municipal waste water, solid waste, chemical waste etc.) As many data sources are associated with a high level of uncertainty, assessments of data reliability and cross-checking on information and estimates are very important elements in the process (see Hansen, 1997). A general problem experienced throughout SFAs is that substances cannot be directly observed within most processes in the economy. Most of the time they are embodied in materials. Often they only become “visible” when they leave the economy again as pollution; as carbon dioxide emissions when fossil fuels are burned for example. If an attempt is made to follow a substance through the economy, the physical flows can often only be imputed by using (average) concentration coefficients. In this sense, the robustness of results does not only depend on money and time resources (as for all bottom-up methodologies), but also on the quality and availability of appropriate coefficients. Hansen (1997) infers that the error margin in detailed study should not be greater than 20 percent.

Assessment - Policy Relevance The relevance of SFAs for informing SCP policies arises from the interest in targeting specific substances and keeping them under control. These are usually low-volume high-impact flows of toxic or other harmful substances, which cause a particular or set of environmental (or health) problems. In the most general sense SFA studies focus usually on a specific environmental problem associated with a particular (group of) substance(s) and provides information for an overall management strategy for this substance (Femia and Moll. 2005: 42).

20 It should be noted that many of the physical environmental accouts compiled by ONS or the Environment Agency are substance-specific. However, they usually only measure the substances at the system boundaries between the economy and the environment – most commonly when they leave the system again. SFA in contrast follows the substances through the economy and/or the environment.

Appendix SCP001 57/196 Therefore, it might be seen in the wider SCP agenda as a helpful methodology to inform the government’s detoxification efforts. In this context SFAs inform policies on the elimination, reduction, minimisation and recycling of specific substance flows. This is facilitated by the assessment of 1. existing levels and trends in substance emissions, accumulations, and concentrations in the human economy and various environmental media (e.g., the lithosphere and biosphere; see Figure 0.10). This also allows the comparison of anthropogenic substance flows with natural, geogenic flows in order to assess the significance of human induced flows and accumulations (Daniels, 2002: 78); 2. the contribution of specific life-cycle stages and processes, goods, sectors, and activities to anthropogenic flows and emissions (and how these contributions change over time); 3. the qualitative nature of environmental impacts of certain human economy activities associated with a specific substance (Daniels, 2002:77).

SFAs can help answer a long array of substance specific policy questions such as: • How much more of a particular hazardous organic compound - such as PCB’s or PCT’s - has been used in the economy compared to the last reporting period? • Where in the domestic industrial supply chain are the main entry points for a certain heavy metal such as lead or arsenic? • In which products is a particular substance embodied? • Where are certain substances released to the environment, where do they accumulate, and do they re-enter the economy (e.g. embodied in food products like fish). The great methodological flexibility of SFA is a strength and limitation at the same time for their application in the SCP context. On the one hand, the flexible system boundaries allow for the approach to be tailored towards a specific – often local – environmental problem. The results of the studies are often directly used by policy makers to solve or control the problem. Due to this project- specific focus, many one-off studies on the sub-national level can be found. On the other hand the lack of a sufficient methodological standardisation (see Brunner and Rechberger, 2003; De Haes et al., 1997) makes a comprehensive ex-post usage difficult because of a lack of comparability and consistency (see Hansen, 1997). This also applies to SFAs potential to be further used in analytical models in a robust manner. If SFAs are to be used in the government’s SCP agenda, there is a need to establish good practise and standardise UK approach and reporting. This has already been done by the Danish Environmental Protection Agency in the course of their substance management strategy. Insights from the Danish experience are summarised in Box 0.9.

Appendix SCP001 58/196

In Denmark SFAs are a standard tool for identifying sources for release of hazardous substances to the environment and the applications of the substances responsible for the release at the national level. The Environmental Protection Agency (EPA) commissions SFAs, which have been identified as actual or potential hazardous to humans and/or the environment. The results of SFA are used for considerations by the EPA related to the need and instruments for risk minimisation for the substance in question and monitoring of policy success in controlling the substance. For monitoring the success of substance related policies, some SFAs are updated regularly.

After the commissioning of a first series of SFA studies mainly for tracing heavy metals and hazardous organic compounds on the national level during the 1970s and 1980s, the Danish EPA prepared its own reporting guidelines to ensure uniformity and comparability of SFAs undertaken in Denmark. These guidelines inform on issues like methodology, data sources, reliability and confidentiality and presents a standard format for the reports. According to the EPA’s Guidelines a SFA on the national level for a hazardous substance will cover the following issues:

• Production, import/export, and utilisation of raw materials and semi-manufactured goods; • Consumption by use areas in Denmark (consumption is taken as net consumption adjusted for import/export of the substance by finished goods/ products); • Emissions and losses by manufacturing processes and by consumption/ utilisation of finished goods to air, water, soil, wastewater, solid waste and hazardous waste by use areas; • Quantity collected for recycling and emissions related to recycling processes; • Quantity collected with municipal waste water, solid waste and hazardous waste and emissions related to treatment processes and final disposal of residual products from treatment processes; • Consumption and emissions related to use of the substance as natural contaminant in materials like fossil fuels, wood, cement, fertilisers and agricultural chalk.

Box 0.9 - Good Practice: Standardised SFAs in Denmark SFAs can make a significant contribution in various capacities to inform policies on the government’s SCP agenda. In terms of the general SCP objectives substance specific de-coupling targets have already been formulated by the government as for example in the key policy area climate change. Substance specific de-coupling indicators are therefore part of the government’s de-coupling indicator set. While data informing these indicators is regularly collected within the UK environmental accounts, they only identify the source of release of a substance. SFAs can provide a much more comprehensive picture of how these different substances pass through the economy and where the best points for policy intervention might be through the identification of relevant resource and product flows. They are therefore crucial for informing preventive substance specific policies to bring about de-coupling processes. The underlying data for SFAs is regularly collected as part of the UK’s environmental accounts.21 Even though it is the intention to keep these official statistics out of the scope of the SFA assessment they are referred to here because any economy-wide SFA would be able to establish a similar aggregate substance flow account, which could be used for monitoring de-coupling processes. At the same time it needs to be highlighted that for the highly hazardous and toxic substances, which are commonly collected in the course of SFAs, only an absolute notion of the concept of de-coupling (i.e. an reduction in the total amount of substance circulating in the economy) makes sense due to their harmful nature.

21 The UK’s environmental accounts, however, only record the outflows of a particular substances...

Appendix SCP001 59/196 SFAs can also inform policies aimed at reducing key environmental impacts, which are closely related to the release of substance flows to a particular environmental medium. In this sense they are a good tool for informing and controlling government’s detoxification efforts.

For the specific policy objectives it is not difficult to see the role SFAs can play to inform government policy. As soon as specific substance flows become of interest to the various SCP agenda items such as production, sustainable business or waste, SFAs provide a flexible tool for informing these policies. The government might for example want to minimise the flow of lead across the supply chain in the course of “production” policies or trace where flows of toxic materials end-up and whether or not they are appropriately treated. In both cases SFA can provide the required information.

SFAs might be of particular relevance to the SCP product policy agenda, where the reduction of environmental impacts from hazardous substances is of major concern (DEFRA, 2005b). Once a harmful substance is identified, SFAs can trace the substance from source and reveal in which products the substances end-up. This can be the basis for setting product standards or provide crucial information for labelling schemes. If preventive policies are of concern they can help to identify where in the supply chain and in which particular processes hazardous substances enter the lifecycle of a product and where they might be best addressed by policies. To give a more comprehensive understanding of substance flows and to comply with the government’s emphasis on the lifecycle perspective, functional rather than regional system boundaries would be more appropriate as applied in life cycle assessments. In this course Tukker and Kleijn (1998) show how SFAs and LCAs can be integrated to further increase their policy relevance.

Appendix SCP001 60/196 I.3.6 Hybrid Methodologies The different methodologies have all shown particular strength and weaknesses. Bottom-up approaches such as LCI, BMFA/MSA or SFA often work with very accurate and detailed information, but have problems in covering all relevant physical flows throughout the supply chain. In contrast, EIOA as a top-down approach provides a very complete picture of the domestic supply chain at the meso level, but models physical flows on a rather aggregate level based on the assumption that for each sector the amount of pollution/resource use triggered is the same for each unit of output produced. Linking bottom-up 22 and top-down approaches therefore might allow combining the strength of both. The potentials for integrating the different methods will be discussed here. First, an explicit focus will be given to alternative approaches for compiling LCIs using input-output data. Second, these insights will be generalised for the integration with other methods.

Integrating input-output and LCI approaches

Environmental input-output life cycle inventories Originally introduced by Moriguchi et al. (1993), researchers at the Green Design Initiative at the Carnegie Mellon University in Pittsburgh started promoting the use of EIOA models as a cheap and time-efficient shortcut to life cycle data, which helps overcoming the truncation error problem of conventional LCIs by taking the complete domestic supply chain into account. These models have become known under the heading environmental input-output life cycle inventories (EIOLCI) or assessments (EIOLCA). In general, the most basic versions of these models do not differ from normal EIOAs. However, to achieve a level of usefulness comparable to conventional LCIs, they are usually based on much more comprehensive environmental data (resource input as well as waste and pollution output). Moreover, due to the focus on products, which are only captured as products groups within an input-output context, the tables show a very high level of disaggregation (often more then 500 prduct groups/sectors). In this form EIOLCIs take all economy-wide interdependencies in tracing the upstream chain of inputs to ultimate extraction into account, and solve the problem of an arbitrary boundary definition and circularity problem23 conventional LCI/As are prone to (Joshi, 1998).24 The other advantage of this method is that such an assessment is very cheap and fast in execution. However, EIOLCIs themselves give rise to a variety of limitations, which make them only suitable to substitute conventional LCAs in very-specific contexts (e.g. Lave et al., 1995, Joshi, 1998; Lenzen, 2001; Suh et al., 2004; Suh and Huppes, 2006):

22 Bottom-up data and process data will be used interchangeably in the course of the discussion of LCIs. 23 The circularity problem refers to the situation that each industry is dependent directly or indirectly on almost all other industries. The chain of interactions typically expands exponentially and it is impossible to trace interactions (Joshi, 1998: 15). It can therefore be seen as the origin of the boundary problem. 24 In fact this problem is only fully solved for the domestic supply chain. Truncation error still remains for supply chain coverage abroad, which can be significantly reduced by endogenesing imports (see Suh and Huppes, 2005). See Section 6 of the main report for details about how the problem can be most appropriately be dealt with.

Appendix SCP001 61/196 • Aggregation level: Even though the input-output tables used are usually very detailed, each product groups still consists of a large variety of individual products. Due to this aggregation problem EIOLCIs can result in over- or underestimation of the “real” physical flows triggered by a product system. It seems that EIOLCIs are at most appropriate for comparing aggregate, disparate products and processes in homogenous product groups. They are certainly not suitable for looking at individual products within a heterogenous product group or individual products, which differ significantly from the “group average”. • Reference year: The underlying input-output tables are often published with some delay. Therefore the underlying EIOLCI data is often older than the one used in traditional process based LCIs. This means, for example, that they are not suitable to assess new products or production technologies. • Use and End-of-life stages: EIOLCIs might cover pre-comsumer stages within the domestic supply chain very well. However, they do not provide a good coverage of resource use associated with the product use, final consumption and end-of life options such as remanufacturing, recycling or disposal. This is particularly true for all durable goods, which are in use longer than the one year reporting period conventionally covered by EIOLCIs. In general, it might be said that EIOLCIs are generally not suited for substituting conventional process-based LCI approaches. However, if resources are limited and time short, they might give a good first snapshot. For the UK the use of the approach is discouraged to perform actual EIOLCI/As in the light of the relatively high aggregation level (76 sectors) and the limited amount of environmental data available. However, the approach might well be used for scoping the government’s action on the SCP product agenda, i.e. for identifying which product groups might be most interesting to focus on based on the (limited) available environmental data.

Hybrid approaches integrating top down and bottom up data Within the debate the word hybrid approach has been used for describing very different things. Some authors have used it for describing input-output models, where physical data have been directly entered into the table. Others have used it for referring to methodological integration of some kind. In this report the word will be used in this second sense.

Hybrid LCI approaches might be distinguished based on three different level of methodological integration. This distinction has been proposed in a recent article by Suh and Huppes (2005), which will be followed closely in the discussion here (see also De Haes et al. (2004) or Heidungs and Suh (2002)). • Tiered input-output LCI (TIOLCI) • Input-output based hybrid LCI (IOHLCI) • Integrated input-output hybrid LCI (IIOHLCI) In the course of the report it will be referred to hybrid IO LCIs when all three different methods are referred to, while for specific reference the individual names of the different methodologies will be used. All three hybrid IO LCI approaches have in common that they try to overcome the truncation error problem of process based LCIs, while trying to avoid the limitations associated with EIOLCIs.

Appendix SCP001 62/196 TIOLCI combine bottom-up data from traditional LCIs for the end-of-life stages with results from EIOLCAs to overcome their problem of an incomplete coverage of the product life cycle. For key upstream flows further physical data is often entered directly into IO framework (see Bullard et al., 1978).

TIOLCI models are straightforward to carry out and relatively time-efficient. While the lifecycle coverage is completed, these models do not allow dealing with the different stages in one systemic approach and cannot account for interactions between them. Moreover, they suffer from a potential double counting problem for which, however, can be controlled for. Most importantly, TIOLCI models do not address the aggregation error of EIOLCIs. In the light of the UK data situation they are therefore generally not suitable to substitute process-based LCI on a larger scale as they do not provide the required detail to be of any use and to deliver robust results. However, as for EIOLCIs the government might still use them for economy-wide benchmarking efforts of different product groups to scope policy action and the needs for evidence build-up. In this case even though being more work-intensive, they are clearly the preferable option among the two (see Tukker et al., 2005).

The aggregation problem is addressed in IOHLCI approaches (e.g. see Treloar, 1997; Joshi, 2000; Lenzen, 2002;). This methodology uses bottom-up data to disaggregate the sectors/product groups in input-output tables according to project needs. For example, if a project is interested in compiling an LCI of a Ford Fiesta, bottom-up data is used to disaggregtate the automobile industry into Ford Fiesta and other cars. This should be done by using detailed expenditure and sales data on the one hand and emission data for the disaggregated processes on the other hand. The remaining stages in the life cycle need to be added manually as for TIOLCI.

Such a methodology seems to be a robust alternative to conventional LCIs even though it cannot be fully established that they necessarily provide better results as this procedure enters new sources of error. Lenzen (2001) discusses the various error sources and shows that they are likely to be smaller and better controllable than the large truncation error found for conventional LCIs. It might therefore be modestly concluded that they are at least as robust than conventional LCIs. In such a case they still seem preferable as a method, as they provide more comparable results due to their roots in a consistent integrated national accounting framework. They can therefore provide a useful evidence for the government’s SCP evidence base.

However, the “queen’s way” is provided by IIOHLCI (see also Heijungs and Suh, 2006), which fully integrate conventional process based LCI and IO data. By using a matrix representation for compiling the process data instead of a process flow diagram approach (see Heijungs 1994; 1996), the bottom-up LCA data can be readily combined with the input-output matrix in a fully integrated model, which covers all stages of the life cycle in a systemic way. Ultimately such a model can overcome the limitations of both conventional LCIs as well as of EIOLCAs without adding new sources of error into the analysis. The only slight drawback is the labour-intensity of the task, which is comparable with conventional process based LCIs.

Appendix SCP001 63/196 Table I.7 provided by Suh and Huppes (2005) might help the government deciding when each method is most appropriate to inform the policy agenda depending on a selection of variables such as the most appropriate choice of system boundaries, costs or data uncertainty.

In general, it can be concluded from this Section that hybrid LCI approaches provide a feasible way to increase the robustness of LCI evidence. The combination with input-output data forces further standardisation in terms of classifying the bottom-up data and leads to greater comparability across studies as the truncation error or cut-off-point problem can be overcome. Especially for setting product standards, devising labelling schemes and for the using the information otherwise to inform consumers this is of crucial importance. Moreover, additional analytical options are provided by hybrid LCIs, which can help to inform the SCP agenda from LCIs more comprehensively.

Table I.7 - Criteria for choosing LCI methods (Suh and Huppes, 2005)

For the future build-up of SCP evidence it is suggested to encourage the use of such hybrid IO LCI methods to increase the robustness and comparability of LCI evidence. However, as not every hybrid IO LCI approach is more suitable than conventional process based ones, the government needs to be careful about what people “sell” as such. This Section might serve as helpful guidance for distinguishing useful from less useful approaches.

Finally, beside all euphoria for these new approaches the outdated input-output data and the limited availability of emission data in the UK’s environmental accounts need to be re-emphasised. The latter is particularly relevant for IOHLCIs, which sometimes only enter a very limited amount of physical bottom-up data into the input-output framework. Therefore, good practice needs to be established in the bottom-up data collection process, if applied in the UK context. Moreover, UK practitioner so far have little experience with these rather new LCI approaches. This means that there might be a limited

Appendix SCP001 64/196 capacity to apply these methods even though a first UK specific software tool will soon enter the market (see Lenzen, 2006). EIOLCIs or (better) TIOLCIs should not be used for detailed LCIs. However, they could play an important role for informing government as they allow the identification of the most environmentally important product groups. This can help to identify key policy areas and direct the further build-up of more detailed LCI/LCA evidence.

Other hybrid approaches Methodological integration is not only beneficial in the case of LCIs and EIOA. There are further options to increase robustness and/or policy applicability for other MFA methodologies even though LCIs and EIOAs are usually involved in these integration efforts due to some unique features: • LCIs is the only methodology with functional system boundaries. • EIOAs provide a full coverage of the supply chain on the meso level, combine data in monetary and physical units and can link bottom-up data into a standard accounting framework. This can radically increase analytical options and policy applicability, and contribute to increased comparability and robustness of the different MFA methods. BMFA/MSA bottom-up data can, for example, be integrated into IO tables. A detailed description of such an integrated approach can be found in Section 4.12. De Haes et al. (1997) show how the policy applicability of SFAs can be expanded by adopting functional system boundaries. Van der Voet et al. (2005) add a new dimension to EMFA accounts by linking via LCI databases to environmental impacts.

Methodological integration is therefore generally seen as a welcome addition to the standard MFA toolkit. The relevance of applying such hybrid methods will often depend on the policy needs. While hybrid IO LCIs are generally recommended for use in the build-up of new LCI evidence, the application of these methods depends much more on the specific policy context. Therefore, in later sections some other key hybrid methods for informing the SCP agenda will be highlighted.

Appendix SCP001 65/196 I.4 Environmental Impact Assessments

Introduction As requested, this report has so far almost exclusively focussed on the description and analysis of the economy’s “physical inventory” on the input and output side in terms of physical weight. The derived information are particularly helpful when it comes to the design of policies that ultimately aim to effectively reduce or control particular physical flows. With the exception of SFAs, these policies might all be grouped into the wider field of de-materialisation. However, they do not allow for the assessment of environmental flows in terms of their environmental impacts and the prioritisation of material flows in terms of their contribution to particular environmental problems. In this sense the reviewed methodologies do not allow for a comprehensive treatment of the issue of de-toxification (reduction of impact per unit of average material flow) as an integral part of a material flow policy and the associated need for material substitution. Environmental impacts and de-toxification are of direct importance for the government’s SCP agenda (DEFRA, 2003; DEFRA, 2005a). This section of the report will therefore briefly outline how environmental impacts can be linked to MFA methodologies, how this might further increase their policy relevance and what the limitations of such an approach are. In general, some sort of physical inventory must be the pre-condition for any environmental impact assessment and some sort of MFA must therefore be an integral part of their methodology. In turn, MFA methodologies can be linked to impact assessment or supplemented with derived information. Two types of assessments might be distinguished, which will be discussed consecutively: • Full Environment Impact Assessments • Impact Potential Approaches

Full environmental impact assessments Full environmental impact assessments try to evaluate the contribution of human-induced physical flows to particular environmental problems. They attempt to idenify the actual impacts on local or global ecosystems and ultimately how these affect (human) populations. They therefore need to deal with very complex mechanisms of how human-induced physical flows change the state of the environment and translate into various impacts. These mechanisms are often not very well known and impacts usually depend on the prevailing local conditions (see Femia and Moll, 2005).

Full environmental impact assessment are therefore usually associated with a high level of uncertainty at the current state of affairs. They usually focus on very specific physical flows (usually substances) and are often restricted to relatively small spatial entities. This raises problems for the purpose of applying them in the context of a wider SCP material flow policy context. In particular it seems unfeasible, • to cover a wide range of physical flows; • to cover all relevant economic processes (which are usually spatially dispersed).

Appendix SCP001 66/196 Despite the severe limitations associated with integrating full environmental impact assessment models into a broader SCP material flow policy, research aimed at their linkage with MFA tools should still be encouraged on a smaller scale. This is because of the great potential such approaches carry for pushing the boundaries of the discipline. For prioritised environmental problems and physical flows these models could be used to integrate policy advice typically derived from ecological models with that of various MFA approaches. This allows policy makers to think about the impacts on natural systems (and their consequences for society), as well as where the best points for policy intervention might be in the human sphere. SFAs are the most obvious candidate from which such modelling efforts could start from due to their flexible system boundaries and their focus on a (or a few) substance(s). Following the methodology of Brunner and Rechberger (2004) these substances are also readily interlinked with the materials transmitting them throughout the system of consideration. The methodological foundations for integrating full environmental impact methodologies into such frameworks might also be found in economy-wide frameworks on the national and regional level as put forward by authors like Suh (2005).

Impact Potential Approaches

Linking Material Flows and Potential Impacts In the wider context of a well-orchestrated, prioritised MFA policy according to the government’s SCP objectives, the “impact potential approach” (IPA) seems to be much more readily applicable. IPAs attempt to assess the potential (rather than the actual) contribution of physical outflows (residuals, wastes) to particular environmental problems without making the link to the actual impacts on the environment and human society as discussed above. This is usually expressed in so called environmental impact potentials like global warming, acidification or tropospheric ozone forming potential. These indicators are usually based on some sort of equivalence scale which expresses the contribution of different material flows to a particular environmental problem in a single uniform unit. For example, the potential greenhouse effect is normally quantified by using global 25 warming potentials for substances having the same effect as CO2 in reflection of heat radiation. Based on model estimations (Albritton et al., 1996) it has been established that 1 ton of methane is assumed to contribute in the same way to global warming as 24 tons of carbon dioxide. The global warming potential of 1 ton of methane can therefore be expressed as 24 tons of CO2 equivalent. In this way the contribution of all greenhouse gases to global warming can be readily compared and it can inform policy prioritisation based on quantitative scientific evidence.

MFA methodologies can then help identifying which and where certain key material flows can be tackled by policy measures to address a certain environmental problem effectively as well as efficiently. In this context IPAs seem most useful to further increase the policy relevance of MFA

25 Global warming or “greenhouse effect” is the effect of increasing temperature in the lower atmosphere. The lower atmosphere is usually heated by incoming radiation from the sun. A part of the radiation is normally reflected by the soil surfaces, but the content of the various greenhouse gases in the atmosphere reflects the IR-radiation resulting in the greenhouse effect, i.e. the an increase of temperature in the lower atmosphere to a level above normal.

Appendix SCP001 67/196 approaches by providing information that allows the prioritisation of material flows within a particular impact category and the design of material substitution policies.

Leiden List SETAC “default “Nordic List” ISO preliminary SETAC Europe list” (Udo de (Lindfors et al., list (ISO, 1997) (1992) Haes, 1996) 1995) non-renewable abiotioc resources energy and abiotic resources global materials scarce, renewable biotic resources biotic resources global water land land land use local global warming global warming global warming global warming/ global climate change depletion of depletion of stratospheric global stratospheric stratospheric ozone depletion ozone ozone human toxicity human human health, human toxicity global, continental, toxicological toxicological excl. regional, local impact work environment human health, non-toxicological excl. work environment occupational human health, local safety impacts in work environment environmental ecotoxicological ecotoxicological ecotoxicity global, continental, toxicity impacts impacts regional, local photo-oxidant photo-oxidant photo-oxidant phto-chemical global, continental, formation formation formation oxidant formation regional, local acidification acidification acidification acidification global, continental, regional, local euthrophication euthrophication euthrophication euthrophication global, continental, regional, local COD (chemical local oxygen demand) discharge effects of waste local heat on water nuisance (smell, odour local noise) noise local radiation local, regional space requirment local final solid waste regional, local (hazardous) final solid waste regional, local (non-hazardous) casualties local habitat alterations local and impacts on biological diversity

Appendix SCP001 68/196 Table I.8 - Impact categories of LCA approaches (European Environment Agency, 1997) Such (material specific) IPA coefficients can be found for a variety of environmental problems. Lifecycle Assessment databases might be seen as a convenient source for this type of data (see van der Voet; Bringezu). Different LCA standards recommend the inclusion of different impact categories usually varying between 10 and 20 in total numbers. A comparison of four major LCA impact lists can be found in Table I.8.

Not all of these impact categories are as well established as global warming (and CO2 equivalence). This means that the scientific acceptance and level of uncertainty associated with these individual impact categories highly varies (see Femia and Moll, 2005). An assessment of the ability of different methodologies to estimate the various conversion factors in terms of scientific rigour and reliability is not within the scope of this project. If IPAs are to play an important role within the government’s evidence based SCP framework as suggested in various policy documents (DEFRA, 2003; DEFRA, 2005b), an independent review of the underlying methodologies should be commissioned. It is of prior importance for the development of a robust evidence base that good practice is established in the use of IPAs for informing government policy. This will be further discussed below. A good starting point for such a review might be a shortlist of 11 environmental impact categories, which are proposed as sufficiently developed by Guinée et al. (2002) in their discussion of the ISO 14040 standard: • Abiotic Resource Depletion • Land Competition • Global Warming • Ozone Layer Depletion • Human Toxicity • Ecotoxicity • Photochemical Oxidant Formation • Acidification • Eutrophication • Ionising Radiation • Final Solid Waste

On the usefulness and rigour of aggregate measures of environmental impact So far the different environmental impacts considered are related to particular environmental problems. This does allow prioritising material (or substance) flows within a certain impact category, but not across different ones. Therefore, the remaining question is whether such an overall prioritisation of material flows can be achieved via quantitative methods in a scientifically sound way through the aggregation of environmental themes into an aggregate measure of potential environmental impact.

Appendix SCP001 69/196 Lifecycle Assessments traditionally aggregate all different impact categories into a single index (see, Guinée et al., 2002). More recently De Bruyn et al. (2004) and Van der Voet et al. (2005) took LCA aggregation methods and applied them in the more general context of economy-wide material flow policies. As an alternative indicator to indicators such as DMC or TMC, which are purely based on physical weight, they propose an alternative indicator of potential aggregate environmental impact called “environmentally weighted material consumption” (EMC). The authors argue that such an indicator allows us to consider prevention and recycling, alongside material substitution as part of a comprehensive economy-wide material flow policy in a meaningful way. A brief review of the method is provided in Appendix I.6.

Much of the appeal for policy making of such aggregation procedures is that they allow the provision of clear-cut policy-recommendation such as “product A is environmentally preferable to product B” or “material A is less damaging than material B” on a more aggregate level. In this way a prioritisation of material flows can be achieved using quantitative methods. However, the scientific rigour of such approaches is highly questionable. Firstly, there is no commonly accepted weighting procedure to aggregate physical flows across the various environmental themes (impact categories). Therefore, a variety of different quantitative and qualitative (or better semi-quantitative) methods from monetisation approaches to target oriented approaches to stakeholder approaches can be found in the literature (see EEA, 1997; De Bruyn et al., 2004) providing potentially very different weights.26 Therefore, results and policy implications will ultimately depend on which method has been chosen. This points towards a second, more fundamental limitation, which is associated with properties good policy indicators should provide to be used for SCP policies as highlighted in the SD strategy (see DEFRA, 2005b; Worldbank, 2002). Amongst others, any indicator should be meaningful and robust. However, both requirements cannot be met by indicators of (potential) aggregate environmental impact.

There is no meaningful way to talk about aggregate environmental impact in a first instance, as it is an abstract notion. In the real world there is not such a thing. Therefore, aggregate environmental impact must be seen as construct, which deals with all environmental problems at once. However, a meaningful conceptualisation, if not by definition impossible, might go far beyond human apprehension.

Even if such a notion of potential environmental impact could be operationalised in a meaningful way, its unobservable nature would make a robust measurement impossible. This implies that there is no scientifically refutable way to establish any set of weighting coefficients. On the one hand, it would always remain unknown whether or not trends in such an indicator of (potential) aggregate of environmental impact over time would reflect the real developments within the natural system. On the

26 Even the argument of applying equal weights in order not to avoid any value judgement as sometimes proposed needs to be questioned as this requires the same type of decision on how the different impact categories relate to each other in their importance for (human affecting) environmental problems.

Appendix SCP001 70/196 other hand, as different ways of estimating weighting coefficients would lead to different trends in the indicator, there is no obvious way of establishing a robust result for policy advice.27 It might be argued now that similar problems exist in any social science discipline – the reader might only think about the concept of utility in economic theory and its empirical applications. However, it is exactly this that is often criticised and in the case of potential environmental impact it is much more sensible if one refrains from aggregating impacts across environmental themes. In fact, dealing with potential environmental impacts on a more difficult disaggregated level (i.e. various environmental impact categories) might actually be preferable for well-informed policy formulation. This can be seen in analogy to many problems occurring in the course of sustainable development policies. The holistic nature of the concept requires balancing and reconciling different economic, social and environmental policy targets. In this course the power of the concept might be seen in its in-built requirement to consider several issues at the same time. This often requires us to trade-off different policy targets providing incentives for new and innovative solutions. These in an ideal world reconcile the different targets in win-win situations. Bringing this back into a purely environmental context, the problem of pollution created by cars might serve as a good example. If it is the goal to find the most environmental friendly car in terms of pollution from fuel use, a lifecycle assessment might assess for example two competing technologies on the market – diesel and petrol engines. The former causes a lot of harm to human health by emitting a lot of PM10, while the latter is the much less preferable option, if global warming is concerned. An aggregate index of potential environmental impact would add both up based on some weighting scheme. This would lead to a result that petrol engines are less preferable to diesel engines or vice versa.

Despite the convenient advice derived from such a measure, it would cover up the sheer fact that both solutions do not seem to be desirable for society (both causing significant negative externalities). If not already available, research would need to be directed towards the stimulation of (eco-)innovations, which minimise the impacts in both fields. In the case of cars such a solution already exists in hybrid engines, which with emit very little CO2 and PM10 per miles driven. Even though it has already been highlighted earlier that results need to be presented to policy makers as concisely as possible, they can be presented only as concisely as useful (when it comes to their aggregation level).28 In the case of IPAs this should not go beyond individual environmental themes.

27 It could be argued that deriving the coefficients in partnerships with key policy makers in the course of stakeholder workshops is the only viable way of deriving weighting coefficients as this would be a measure of aggregate environmental impact weighted by the perceived political importance. Such a way of aggregating the results could be more easily justified. However, the results would only reflect the perception of the (usually very small) group involved in this weighting exercise and might not be easily generalised. Moreover, even though sophisticated methods exist to translate people’s preferences across problems into numbers, there are problems for the involved participant in perceiving the relationship between the various environmental problems in a way that they can be expressed on a cardinal rather than ordinal scale (i.e. we want to know that global warming is twice as important as euthrophication and three time as important as ozone layer depletion.). This is different, for example, than the expression of peoples’ preferences in economic theory. An ordinal ranking would be less of a problem, but would not help if an aggregate measure of (potential) environmental impact was concerned. 28 If results are more complex the challenge for research rather seem to lie in a clear presentation and interpretation rather than necessarily aggregation.

Appendix SCP001 71/196

Implications for the build-up of a robust SCP evidence base If accepted, the suggestions made in this Section have important implications for the government’s efforts in developing a physical evidence base for SCP policies: • First, it needs to be established which impact categories are developed enough for robust policy advice. • Second, policy advice should not be built on measures of aggregate environmental impact. Therefore, a review of the robustness of IPAs across the various environmental themes is suggested. Such a review should be commissioned to an interdisciplinary research team consisting of ecologists (spcialised on environmental impacts), LCA practitioners and other MFA experts. This review would be used by the government to produce an exclusive shortlist of environmental impact themes, which are considered suitable for informing SCP policies. As aggregate environmental impacts are rejected as robust evidence, further suggestions should be provided on the best way in which policy advice could be derived from the more disaggregated results. This is particularly important for LCA methods, where the aggregation of impacts across environmental themes is common practice. Moreover, LCAs play an important role in different key areas of the SCP agenda such as product policies or the provision of information for more sustainable choices. Hence, it is important for the government to develop its own view on how impact assessments should be carried out in the context of LCAs to inform SCP policies and to establish good pratice guidelines. If it is the goal to “extend the environmental impact assessment of products as a regular feature of good business practice” (DEFRA, 2005b: 47), this should be done in a robust and meaningful way.

Appendix SCP001 72/196 I.5 Policy Analysis

So far this report has been concerned exclusively with ex-post analysis. However, even though it is of crucial importance to monitor success in an ex-post approach and to learn from past experiences through an analysis of historical data, the decisions of policy makers aim to shape the future of society. In this sense decision making is strictly forward looking and it is of crucial importance for policy makers to know how different sets of policies might affect potential futures. Therefore, Proops (1991) stresses in this context that accounting efforts need to lead to „global modelling“. In this capacity further value might be added to the MFA evidence in terms of its policy relevance by informing decision makers on the choice and effectiveness of different policies and ultimately reducing the large uncertainty attached with unknown future states.

Quantiative ex-ante methodologies such as forecasting and scenario modelling have been undertaken in different contexts. The natural sciences have done a great deal to predict the future state of the environment. The immense modelling efforts (of ecologists) in climate change research are one prominent example (for a good summary see the Appendices in Grubb, 1999). These models have contributed substantially to a better understanding of global warming and its consequences and therefore ultimately helped to raise awareness and built capacity on the policy level to take action in this field. From this perspective such models seem to have been of great policy relevance. With the focus of this report on the usefulness of MFA methodologies for decision support, it is more interesting to concentrate on models which are concerned with addressing physical flows in the economic system, i.e. starting from the socio-economic sphere and assessing the physical implications of different policy measures. These models have been pre-dominantly developed in the environmental-economic literature, are directly concerned with policy formulation, and might be generally referred to as policy analysis models. In particular, they provide quantitative decision support for comparing future states of the physical system (so called – what if scenarios) or the effectiveness and efficiency of different policy instruments to achieve a predefined target.

Figure 0.12 - Quantitative policy analysis in a static perspective

Appendix SCP001 73/196 Figure 0.12 sketches the general structure of models for quantitative policy analysis as prominently conceptualised by nobel prize laureate Jan Tinbergen (1956). Through the introduction of mathematical functions, these models establish mechanisms between external factors, which are not under control of decision makers (exogenous variables), policy instruments (variables which are assumed to be under full control of decision makers such as taxes, standards or emission trading schemes) and policy targets (quantitative targets of a policy). Material flows, for example, are not under the direct control of policy makers. The current material flows are external factors from this perspective. However, they can be indirectly influenced through the use of a set of policy instruments. This could be a tax on resource use or a ban of a harmful toxic substance. In both cases action will have been triggered by certain policy targets.

As SD and SCP are holistic policy concepts it is of great importance to be able to account for a list of other (endogenous) variables, which might be indirectly affected by a policy. For example, even though policy makers might be interested in the effectiveness of a carbon tax in environmental terms, it is of key importance to control how this affects production activities in the different sectors, changes the employment situation as well as the income structure within the population. Policy analysis models useful for SCP decision-making should therefore not only focus on the target variables of the particular policy under consideration (here the MFA variables), but also cover a broader range of variables in the economic and social and sphere.

As the ultimate concern of decision-making based on an anthropocentric understanding of SD is human well-being, the models implicitly (e.g. it is assumed throughout that an increase in an environmental indicator like CO2 represents a decrease in or at least threatens human well-being in the long run) or explicitly (through some sort of welfare function) deal with the welfare implications of different sets of policies. The conceptualisation of human welfare is a complex issue and its discussion has filled many bookshelves in libraries over the past decades. It therefore cannot be in the scope of this report add-on to provide a comprehensive overview. Instead the insights derived in the national accounting literature, which is concerned with the quantitative representation of human welfare, will be briefly summarised and linked back to policy analysis in the MFA/SCP context.

Starting off from the discussion of GDP/ GNP as a index of human well-being (e.g. Nordhaus and Tobin, 1972), research tried to adjust these measures during the 1970’s for environmental degradation and some social issues (e.g. Daly, 1994). Different “Green GDP” and other “Social Welfare” measures were the result of these discussions. However, inspired by the social indicator movement, it is commonly accepted today that human welfare is a too complex issue to be measured in a single index (e.g. Juster et al., 1985, Stahmer, 2000, Keuning, 2000). Instead it is an established accounting wisdom that human welfare can only be appropriately represented in a quantitative context through a collection of indicators, in which each monetary and non-monetary welfare attribute is recorded in the most appropriate unit and the most appropriate classification (see Keuning, 1994). The result is an indicator set in multiple units and multiple classifications - ideally integrated in an accounting framework. Although they are not integrated within a larger accounting system, the local quality of life indicators, introduced recently by the Audit Commission (2005), follow this approach.

Appendix SCP001 74/196 For MFA approaches this has two major implications. First, the negative externalities triggered by physical flows should be best represented in physical units. In this sense policy analysis carried out in the MFA context can be seen as good practice in terms of their depiction of “environmental” welfare aspects. In turn a very critical view should be taken towards approaches, which provide policy decision support based on preference or welfare estimates expressed in a single unit such as environmental valuation or any other stated preference approach. The required welfare function often converts all welfare attributes into a single unit (and arbitrarily weight them in this context), and also only covers a very limited number of welfare attributes. If the rigorous criteria applied in the methodological assessment of this review are applied, these approaches would not appear as suitable for robust policy advice.

This point towards a second implication for MFA based policy analysis. Because physical flow variables only represent a small sub-set among all relevant welfare attributes, policies can only be judged in terms of their desirability for society, if they are envisioned within a larger context. These are the “side-effects” of Figure 0.12 mentioned earlier. This means the effects of policy interventions can only be adequately analysed, if the whole system is represented in its economic, social and environmental dimension. Such comprehensive decision support as required through the holistic nature of the SD concept can only be provided through large scale models. In order to deal with such a complex high-dimensional problem, which depicts the interaction of a large number of characteristic variables in the social, economic and environmental sphere, in quantitative terms, a large number of equations are required. The link between the different MFA methodologies discussed in this report and policy analysis varies. In this context it seems helpful to classify the methodologies according to two criteria: • Temporal system boundaries: It seems sensible to distinguish MFA methodologies looking at product systems on the one hand, which are interested in the material flows triggered over the whole life span of a product from the cradle to the grave. As all materials are accounted for when the product goes out of service, there is no future state anymore that is of any interest for policy- makers. Scenarios are therefore only of limited interest as the product system does not further evolve.29 All other MFA methodologies account for the material flows associated with human activities within a certain reporting period (usually one year). Future states in this case are of great interest for decision-making. • Nature of the framework and level of integration: Most of the MFA methodologies such as SFA, EMFA or Bulk Material Flow Analysis methodologies are purely physical bottom-up accounting frameworks. Environmental input-output methodologies are different. They are accounting and modelling framework at the same time.30 In this sense, the capability to carry out scenarios can be seen as in-built. Moreover, they directly interlink physical data with a standard economic accounting and modelling and can be directly used to extend economic policy analysis

29 This is not say that political decisions do not affect the lifecycle of a product. It could be argued that scenarios are relevant to estimate how different sets of policies influence the lifecycle impacts of products. This could be interesting for durable goods, where the use phase emissions would be affected, as well as consumption goods, where the production processes might change. In this case IOLCA methods as well as hybrid LCAs would be certainly much better suited as they are readily integrated into a comprehensive national accounting structure. 30 This does not change through the fact that the data is usually provided by ONS and EIO practitioners often do not engage in the data collection efforts themselves.

Appendix SCP001 75/196 for a physical dimension. Therefore, EIOA are capable of integrating environmental and socio- economic issues within one consistent and comprehensive data framework. In fact, EIOA seems to be the only methodology, which can deal comprehensively with physical flows and accompanying economic and social issues in future scenarios. It needs to be highlighted that EIOA models are by no means the best choice as they are often static in nature even though the estimations of dynamic versions is straightforward (see Miller and Blair, 1985). They further suffer from their restrictive assumptions regarding perfectly complementary (Leontief) production functions, no economy of scales and no capacity constraints in their prediction of future states. While they might still be suitable to answer policy questions in a “what-if” manner such as “how can future carbon emission be affected, if all new homes built from today where BedZED standard” or “how much can freight transport emissions be reduced, if one fifth of the food is produced and consumed locally” they fail to satisfactorily answer questions associated with structural policies such as the implications of a carbon tax inducing economy-wide reshuffling effects (substitution of inputs crucial) or questions where the changes in technology need to factored in. Hence, it can be said that scenarios based on static EIOA are restricted to strategic planning. However, the input-output tables are also a starting point for a large variety of other econometric models, such as liner and dynamic programming models, computable general equilibrium models or simultaneous structural macro-econometric models. In a review of some of the most relevant models in the context of SD, Frohn (1998) concludes that only the latter are sufficiently comprehensive to deal with the complex social, environmental and economic interactions at the same time as required by the holistic nature of the concept. Based on the earlier suggestion for the need of large scale models, policy analysis for SCP might therefore be most appropriately conducted in simultaneous structural macroecononometric models (see Chen et al., 2005; Frohn, 1998) though less complex model (ideally with a large number of variables) can benefit the SCP agenda. The MFA literature in the UK – most prominently represented in the collection of Biffaward studies – has only made a very modest use of scenario analysis to inform policies. Those scenarios have usually been restricted to “what-if” type questions. While this in itself does not constitute a problem, they often miss a link to sets of available policy instruments, which can be used by policy makers to change the current state. These have usually not been provided within the models and quantitative advise on the best choice of policy instruments has therefore often not been provided. Moreover, when scenarios were carried out, models have most often refrained to take into account the social and economic implications of a policy. The debate surrounding the rebound effect, where changes in resource efficiency change relative prices and lead to adjustments in consumer behaviour, which partially off-set the reductions in resource use (Binswanger, 2001; Jalas, 2002; Hertwich, 2005), demonstrate that these links are important for making an informed decision in the spirit of the sustainability concept and also have direct “physical feedbacks”.

The application of MFA evidence in scenario models provides a large variety of options to further improve the policy applicability of the different methods. By applying MFA evidence jointly with other social and economic data the uncertainty associated with future policies can be reduced by providing information about alternative outcomes and the effectiveness of different policy instruments. Scenario models can therefore be seen as one of the most immediate applications of MFA evidence in terms of their policy relevance.

Appendix SCP001 76/196 I.6 Discussion of Study by Van der Voet et al. (2005)

Introduction The study by Van der Voet et al. (2005) sets out to estimate the potential environmental impacts of ecnomy-wide material flows (as recorded in EMFA accounts) of the EU-25 and AC-3 countries/economies. In particular, a methodology is outlined to weigh specific material flows with information on the environmental impact related to these materials, and to use these weighting factors to compose indicators to assess the environmental impact of the material consumption of these economies.

Methodology The methodology applied by Van der Voet et al. (2005) combines information from the ETH life cycle database (Frischknecht, 1996) with the consolidated EMFA accounts previsouly devised in the course of the so called „Zero study“ (see Moll et al., 2003). For the purpose of the study „materials“ are defined neither as raw materials (resources) nor as products, but as „finished materials“. Bases on such a definition „wheat and cotton fibre are then materials, not bread or textile. Glass is the material, not windows or bottles, nor sand [...]“ (Van der Voet et al., 2005: 34). Overall, 31 finished materials were distinguished in the study and their environmental impacts was assessed across 13 (impact) categories in a life cycle approach. This was achieved in two steps: In a first step, the impacts per unit of physical flow were established using the ETH database. In this course „all processes“ (see limitations) required „from the cradle to the grave“ were assigned to the list of finished materials, where the use phase emissions could only be included in a limited sense (for details see, Van der Voet et al., 2005: 35). Impact coefficients are neither time nor location specific – this means the results must be interpreted as impacts of the materials „wherever and whenever they may occur“ over the life cycle of the product. In a second step, the impacts of the various material were linked to the consolidated database of the Zero study (see Moll et al., 2005). To overcome the system boundary problem between the LCA and the material flow data (the former uses the system boundaries of apparent consumption, while the latter uses the DMC system boundaries), the MFA data was translated into finished materials. The resulting system boundaries are the ones of (estimated) consumption of „new materials“ implying that materials embodied in products are excluded. This might result in lower quality results for some materials, in particular the ones which are not produced in the country of concern. However, the authors suggest that the resulting error is relatively small, as products flows are in general small compared to the flows of materials. The overall impact is then calculated by multiplying the total weight of material with the impact coefficients for the various impact categories. Van der Voet et la. (2005) also provide an index of environmentally weighted consumption, which aggregated environmental impacts across categories. Such a procedure us rejected for reasons explained in depth in Section 4.4 (see also Appendix I.4).

Appendix SCP001 77/196

Appeal Van der Voet et al. (2005) provide a unique methodology to link environmental impacts with econmy- wide material flow accounts. This opens up a whole new array for policy advice – most importantly it enables to advise policy makers meaningfully on where the substitution of materials could be a viable policy option. Moreover, it has a general appeal for the SCP agenda as it uses the UK specific MFA data calculated by Moll et al. (2003). In this sense it provides a unique, UK specific source of evidence for informing SCP policies in the UK.

Limitations There are a variety of limitations associated with the attempt to comprehensively assign environmental impacts to the econmy-wide flows of materials. Amongst others these are: • Typical for bottom-up data sources, LCA databases are neither complete in their coverage of processes nor their coverage of materials. While the missing processes in the „pruned“ production tree result in a truncation error of unknown size (for some estimates see Lenzen, 2001) in the estimates, the incomplete list of materials in the database requires further to extrapolate material specific information into larger groups of materials relevant in the economy-wide MFA context (see Tukker et al., 2005 for a similar arguments in a product specific context). • Translating all economy-wide material flows into the list of finished materials enters an additional sources of uncertainty into the analysis. • The ETH database used is rather outdated, considers only process data representative for Western Europe and not specifically for the UK, and no technical progress is considered in the assessment of the lifecycle emissions from the cardle to the grave.

Conclusion The proposed method of Van der Voet et al. (2005) outlines an intersting new route in assessing the environmental impact caused by material consumption throughout the economy. Such a methdology does not only allow for the priotitisation of physical flows in terms of their impact in the various environmental themes, but also to bring in material substituton as a new option of an economy-wide policy. However, even though the the approach developed previsouly in Van der Voet et al. (2003) and De Bruyn et al. (2004) in this sense has made fundamental contribution to the literature, the limitations of the existing study seem to be too fundamental to comprehensively inform the SCP policy agenda. Therefore, the application of the results here for the identification of key materials can only provide a first indication of which materials are of strategic materials (see Section X) in the spirit of the EC‘s Thematic Strategy. Moreover, it provides a good starting for building up a modelling framework that allows the prioritisation of physical flows for a targeted SCP material flow policy. Further inspiration in this context can be sought from the study by Tukker et al. (2005).

Appendix SCP001 78/196

II Review of Biffaward Studies (Section 5)

II.1 Tables: Detailed assessment criteria, policy agendas, list of Biffaward studies

• Table II.1 - The detailed assessment criteria • Table II.2 & II.3 - Policy agendas • Table II.3 – List of Biffaward studies

Criterion Description Specification B1 System Boundaries Good A clear and useful boundary A system boundary has been explicitly discussed has been defined and: 1. Is unambiguous 2. Distinguishes economic/human systems 3. Is relevant to the policy objective of the study 4. Is understandable to any professional user Satisfactory A boundary has been • A system boundary has not been explicitly consistently followed defined but a sensible boundary has been consistently followed

Boundaries Boundaries OR • A system boundary has been explicitly discussed but fails to achieve one or more of the four criteria above Unsatisfactory Arbitrary and not • Weaknesses in the definition of the boundary satisfactorily defined mean that the objectives of the study may not be achievable OR the recommendations may not be valid or supportable D1 Year of Dataset Good 2001+ • The year of the dataset is taken to be the Satisfactory 1999 or 2000 earliest year of the main31 contributing data Unsatisfactory Pre 1999 sources • The year is taken as the last year for time series studies D2 Updatable Good • The study intended that the data would be

Data Data Designed to be updated updated and the model re-used Satisfactory • All of the main datasets as used are updatable – the results could be recreated by Applied datasets are updated another expert if supported by the original authors Unsatisfactory • It is unlikely that the main datasets could be One off data sets updated or used again to recreate the results

31 Main datasets are defined as datasets which might affect the recommendations including the quantitative assumptions and factors

Appendix SCP001 79/196 D3 Reliable Good Uses official government • The main datasets comply with this statistics or observed data submitted to regulators Satisfactory Unofficial data of good • The trustworthiness of the data is tied to the quality from trustworthy relationship of the bodies providing the data sources and the objectivity of the study. Unsatisfactory Concern about data source • The level of uncertainty may make some of and method of construction the recommendations invalid D4 Available Good • All of the main datasets are in the public domain and available to any user OR Published/offered • The authors offered the data during the consultation process. Satisfactory • All data and factors used to generate the Available on request results would likely be available to all users of the data in practice and on request. Unsatisfactory • It is thought likely that not all of the data and factors as used to generate the results would be available in practice on request or Data not available on request • The report was assessed not to be properly documented AND the authors failed to offer the data during the consultation process. D5 Documented Good • Users are able to understand how the strengths and weaknesses in the input data Well documented & impact on the recommendations understandable to any reader • The results would likely be able to be re- generated by an expert practitioner Satisfactory • An attempt has been made to document the Documented process which generated the dataset and the main procedures have been communicated. Unsatisfactory • The main process which generated the Not properly documented datasets or results is not clear. D6 Real/modelled Good Most data represent real • All of the main input datasets represent real observations observations relating to mass Satisfactory • The assumptions made when generating the Most data are modelled but input datasets are NOT significant enough the assumptions used are for them to be needed to be discussed in explicit, limited and conjunction with the recommendations for reasonable the recommendations to be valid. Unsatisfactory • The assumptions made when generating the input datasets are significant enough for Most data are modelled but them to be needed to be discussed in the assumptions used are conjunction with the recommendations for either not explicit or the recommendations to be valid OR problematic • Any proxy or modelling procedure relating to the main data which is not based on a physical material flow (i.e. financial) M0 Method(s) used (The main approach or methods as applied) M1 Method Robustness Good Method is robust • High quality results are produced using the most appropriate and transparent methods Satisfactory Minor flaws in the approach • Approaches and methods are credibly Method Method applied and transparently discussed, and limitations are taken into account of in the recommendations

Appendix SCP001 80/196 Unsatisfactory Serious flaws in the approach • Serious flaws in the approach (e.g. lack of transparency or credibility in the use of approaches and methods). M2 Objectivity Good There is no potential conflict • Study is clearly independent of those with a of interest amongst the material interest in its outcome studies funders Satisfactory There is a material interest • There is a material interest amongst one or amongst one or more of the more of the studies funders BUT this is not study’s funders apparent in the report Unsatisfactory There is an apparent material • The objectivity of the study appears interest amongst one or more compromised of the study’s funders M3 Peer reviewed Good Full methodology published • An academically credible method not subject in academic journal; widely to widespread fundamental dispute accepted Satisfactory • Some kind of independent peer review Some kind of peer review process has been established by the study process; method authors (independent review group or invited accepted/adopted by at least reviewers) OR one other research group • Method accepted/adopted by at least one other research group Unsatisfactory • Not independently reviewed and method not Not independently reviewed used by any other research group OR and method not used by any Method fundamentally contested other research group •

C1 Non-direct flows (see below) Good • Accounts for hidden flows (e.g. unused material such as mining waste) and indirect flows (life-cycle material requirements upstream of the product, including process Yes material requirements abroad) • Accounts for memorandum items for balancing (all materials associated with economic flows, including for example oxygen required for combustion) Satisfactory • Uses a boundary which distinguishes between the economy and the environment

At least defines boundary and • Accounts for indirect flows. This is NOT required for some material studies (i.e. accounts for indirect flows carbon/nitrogen) and some waste flow studies where the accounting for a flow is all that was intended. Consistency Consistency Unsatisfactory • It may be noted that all but one of the primary and secondary studies fall into this Fails to account for indirect category flows or define a boundary in Material studies have been excluded from terms of the economy • assessment against this criterion for the reasons given above C2 Consistency with standard classifications Good • The report explicitly demonstrates that the Relates to SIC codes results relate to SIC codes Satisfactory Approximately relates to SIC • Most of the results could be related to SIC codes codes. Unsatisfactory • Large parts of the results do not relate to SIC Does not relate to SIC codes codes s u l R1 Comparability

Appendix SCP001 81/196 Good Project results can be • The quantitative results could be added to compared to other findings existing data sets or the results from similar and can be used to extend studies existing data sets Satisfactory • The quantitative results are comparable to Results can be compared to existing data sets or similar studies BUT are existing findings thought to be in different formats Unsatisfactory • The type of outcome from the study means that the results cannot be compared to Results are not comparable to existing data sets or similar studies OR existing data sets • The results are in unexplained conflict with existing data sets or similar studies, or poorly understood P1&2 Policy relevance (SCP & general) - See policy agendas tables (Tables B2 & 3) Good • The recommendations can GUIDE policy Feeds in directly to a policy makers in the future formulation of policy in agenda at least one policy agenda Satisfactory Provides contextual • The recommendations provide useful information of use to policy INFORMATION to policy makers in agenda relation to one of the future policy agendas Unsatisfactory Not useful in relation to • The information provided in the report is of policy agenda little or no relevance to future policy making P3 Accessibility Good Results/data are easily • The report is well written and can be accessible and transparent for understood by a relevant policy maker direct use within the policy making process Satisfactory Results/ are accessible and • The report can be understood by a relevant transparent and can be used policy maker in the policy making process with an appreciation of the study’s weaknesses Unsatisfactory Results/data cannot be used • The report is not understandable to a relevant in the policy making process

Policy relevance Policy relevance policy maker OR • The study has been assessed as unusable in some other way. P4 Endorsement/use Good A decision-making authority • A statutory decision making body was contributed to the study and formally involved with the generation of the has officially adopted its results AND results or recommendations • The results have been formally used by a statutory decision making body OR • The recommendations have been formally adopted by a statutory decision making body Satisfactory A decision-making authority • A statutory decision making body has made has used the results use of the results or recommendations whilst formulating policy Unsatisfactory No official use of results or • No use in formulating policy recommendations

Table II.1 - Assessment criteria

Appendix SCP001 82/196 Public Private

Global European UK/DA National Regional Local Sector Company Decoupling/'green Regional (RDA-led) growth'/eco- Identification of Production focus on resource efficiency/resource waste/inefficiency efficiency – productivity Sustainable Corporate reporting

business guidelines/regulations Integrated product policy Products Market transformation (e.g. WEEE ELV Directives)

Regional Economic Green 'challenge' Strategies (RES) & Innovation funds Innovation and industrial collaboration.

Centres of Green Green public Procurement Procurement public procurement Excellence procurement Information provision, Regional spatial Consumption economic instruments strategies Waste management Waste management Waste planning planning Note: The regional policy agendas were contributed by policy experts after the project workshop and have therefore not been referred to in the assessment notes.

Table II.2 - Sustainable Consumption and Production (SCP) policy agendas – (Used in the assessment of Criteria P1)

Appendix SCP001 83/196

Public Private

Global European UK/DA National Regional Local Sector Company EU SDS/ 6EAP/ Sustainable Sectoral Corporate Comparison of Thematic development Regional SDS sustainability sustainability/CSR national SDSs strategies strategy (SDS) strategy strategy Resource Resource Part of Resource accounting accounting accounting Environmental methodologies/ Benchmarking methodologies/ methodologies/ Management indicators indicators indicators System Local Strategic Modelling policy Regional Meeting sectoral Partnerships Benchmarking Benchmarking impacts (also by Development targets (e.g. (Local sector/region) Agencies recycling) authorities) Sectoral analysis/ Eco- Sustainable prioritisation (i.e. industries/Indus-trial Community where are the symbiosis Strategies biggest impacts) Identification of Freight traffic

environmental limits planning

Table II.3 - Table II.3 Other environmental policy agendas - (Used in the assessment of Criteria P2)

Appendix SCP001 84/196

Primary production studies B/1127 4Sight Rocks to Rubble B/2034* Pigs B/1639* Poultry Industry Secondary production studies B/1601 Chemicals B/1406 Construction B/1412 Paper and Board Industry B/2043 Electricity Generation B/1505 Ceramics B/1489 Glass Manufacture in UK B/1584 Foundry B/1705 Iron, Steel and Aluminium (Economic) B/1698 Iron, Steel and Aluminium (MFA) Tertiary service sector studies B/1635 National Health Service - Material Health B/1524 Public Sector B/2068 Education B/1224 Schools B/1536 Financial Sector B/1634 Exhibition Industry Product studies B/1663 Automotive B/1776 Magazine Publishing B/1640 Housing B/2222* Clothing and Textiles B/1837 Newspapers UK B/2182 Food and Drink Processing Consumption studies B/2230 Ecological Budget UK B/1831 Scotland B/1996 South West - Stepping Forward B/1946 Northern Ireland - Northern Limits B/1699 South East - Taking Stock B/2011 Wales B/1646 London - City Limits B/1303 Isle of Wight - Island State B/1195* Tourism Waste studies B/1783 Waste Data Flow B/1425 Producer responsibility (REMAT) B/1352 Tyres B/1716 Transport Options for Scottish Waste B/1850 EuroCharge B/1413 Civic Amenity Sites B/31 Waste from electrical and electronic equipment B/1597 Agricultural Waste (Mass balance)

SCP001 Page 85 of 196 B/1265 Agricultural Waste (Strategy) B/1826 Thermal Methods of Waste Treatment B/1815 Wood combustion in furniture manufacturing B/1410 Furniture Packaging B/1424 Solvent Waste in Furniture Manufacturing B/1448 Glass from Fluorescent Tubes and Lamps B/1411 Timber Waste B/1271 Packaging Material studies ** Methane B/1355 Carbon UK B/1445 Nitrogen UK B/2173 Wood

Table II.4 - The Biffaward studies

Source: http://www.massbalance.org/projects/ * These studies were not available in time to be assessed and will not therefore be assessed ** The methane project was not listed on the Mass Balance website but was provided by Louise Barcroft of Biffa.

SCP001 Page 86 of 196 II.2 Assessment notes made for each study

Primary production studies

B/1127 ‘4 - Sight Rocks to Rubble’

Organisation: Sustainability North West

Stated objectives: This project forms part of a large initiative (RIFAS) in partnership with Sustainability North West to provide an integrated framework for auditing sustainability not only in the NW region but throughout the UK. This part of the initiative will establish a model for sustainability applicable at a regional level. Outcomes: The project successfully manages to construct a mass balance and explores an appropriate sustainability strategy for the sector.

Boundaries – Boundaries are implicitly defined although not explicitly discussed

Data – There are a number of concerns relating to the sourcing and construction of data. The source and construction of the data sets are not transparent in the main report. The difficulties are discussed in the notes section (6.1) it is reported that the author is ‘unable to offer assurances as to the validity of the data used’. A valiant attempt has clearly been made with patchy data but the methods used have not been fully explained. Method – The methods used to construct the data set are not transparent.

Classification – Data approximately consistent with SIC classifications

Results – The result are clear and broadly comparable to similar data sets. Some work would be required to validate the way in which the data was constructed and relate this to other data sets.

Policy Agendas: Policy relevance - The results of the study are relevant to policy makers at a regional and perhaps a national level. The results feed into meeting SCP objectives at a company and sector level and provide information for national sustainable strategy.

Policy recommendations -

SCP001 Page 87 of 196 B/1815 - Benchmarking wood waste combustion in the UK furniture manufacturing sector

Organisation: British Furniture Manufacturers Ltd

Stated objectives: This project seeks to quantify the mass balance associated with the on-site wood combustion units and to produce benchmarking and best practice information on boiler use and control, in order to optimise boiler performance. By ensuring that energy is properly utilised and emissions are minimised, the project will seek to make wood combustion a more sustainable waste management option.

Outcome: The objectives of the study have largely been achieved. However, the purpose of the report clearly has the dual purpose of meeting these objectives as well as a lobbing exercise during the review of PG1/12(95), and therefore it may lack objectivity in its conclusions. However, the main part of the report represents a highly focused and useful reference document for the sector. The mass balance data is of limited use outside the sector and is based on data of limited reliability.

Boundaries – The boundaries were not discussed but it has been assumed that a boundary of furniture manufacturers has been followed. The audits included site outside this boundary which were only excluded when the results were scaled up.

Data – There are serious faults with the data, the most significant of which is the lack of sampling technique. This means the scaled up mass balance figures have to be considered as modelled results.

Method – The methods used (site audits) combined with a lack of understanding of how representative the sample sites are means that there are serious flaws in the approach taken.

Consistency – The data (consisting of the flows of wood and emissions) is consistent with SIC classifications and other larger data sets.

Policy Agendas: Policy relevance - The results of the study have moderate relevance to waste management policy makers at a national level and feed directly into the review of PG1/12(95) on emissions of plants from 0.4 to 3MWth which are regulated locally. It therefore seems likely that the report will have a limited future application in policy once the review has been completed and that is why a green assessment did not apply.

Policy recommendations – No explicit recommendations were provided. However, the implication of tightening emission regulators was briefly discussed in the conclusions. It is considered that this was done without full objectivity and did not consider the possibility of better management of existing plant.

Outcomes from recommendations –

SCP001 Page 88 of 196 The British Furniture Manufacturers Ltd reported that: [The report] led to revisions contained in process guidance note PG1/12(04) and has been used by a number of sites to optimise their wood combustion process in order to reduce emissions.

SCP001 Page 89 of 196 B/1355 - Carbon UK

Organisation: Industrial Sustainable Development Group

Stated objectives: This research will analyse total carbon budgets of the waste management industry by region, compare total carbon costs against other sectors and evaluate the potential for developing tree replanting schemes in the vicinity of or on capped landfills to offset these emissions through carbon sequestration

Outcome: The report manages to cover a much wider scope than the objectives set out. The result is a relatively comprehensive account of the stocks and flows of carbon in the UK which later goes into carbon management in more detail.

Boundaries - Some confusion in describing boundaries due to the use of the term ‘responsibility for emissions’ led to confusion regarding whether carbon embedded in imports were being assessed (p.4). Data - Data sources were clearly described and official data sources were used almost entirely. Some cases of data from different years being used. Data set could be re-created if required if a constant bio-stock assumption was to be acceptable (p. 16-24) Method – The methods used are transparent and robust. Only hypothetical concern about the objectivity of the study due to TXU funding

Consistency – Although the report does not draw a distinction between the environment and the economy, the way the data is set out would permit the boundary to be set ex-post. Indirect flows have not been provided. Data approximately consistent with SIC classifications

Results – The result are clear and comparable to similar data sets. Good use is made of the results in the findings and conclusions.

Policy relevance – The results of the study are relevant to policy makers at all levels and provide information for meeting a number of the SCP objectives.

B/2222 Clothing and Textiles Organisation: Cambridge University Environmental Projects Ltd

Stated objectives: The aim of this project is to conduct an extended mass balance study of the UK clothing and textiles sector, and to evaluate the potential for changed scale supply chains to improve company and UK competitiveness while significantly reducing the need for sending waste to landfill.

Outcome: Report Expected: 30th June 2006

SCP001 Page 90 of 196 B/1489 Glass Manufacture in UK

Organisation: British Glass Foundation

Stated objectives: The principal objective of the project is to promote sustainable development. The means by which this element of work will contribute to this objective is by the provision of sound information which will assist regulators, businesses and entrepreneurs in their decision- making processes.

Outcome: The objectives of the study have been achieved. The report is generally well written and its contents are very relevant to waste management SCP policy. The boundaries are clearly defined. Data availability was clearly a limiting factor in this study. The researches involved have largely overcome the challenge, in part at least through their association with the relevant trade association. However, the basis in which this data has been provided was to keep its contents confidential. The data availability has therefore assessed as being unsatisfactory. The recommendations are targeted at a range of audiences. There is perhaps excessive focus on recommendations for improving data collection However, there are clear recommendations in relation to recycling and bottle banks which are not sufficiently tied back to the results.

Boundaries – The boundaries are clearly defined.

Data – Data availability was clearly a limiting factor in this study. The researches involved have largely overcome the challenge, in part at least through their association with the relevant trade association. However, the basis in which this data has been provided was to keep its contents confidential. The data availability has therefore assessed as being unsatisfactory. The process of generating the data sets is only partly documented and only discussed in general terms in section 7.1. There are a number of high level estimates were needed to fill gaps in data collection.

Method – The methods used are transparent and robust.

Consistency – Data is consistent with SIC classifications and has been awarded a good assessment even though not explicitly discussed as the sector is clearly defined by the SIC classification sector.

Results – The results are clear and comparable to similar data sets.

Policy Agendas:

Policy relevance – The reports contents are very relevant to waste management SCP policy at a national and local authority level and they feed directly into meeting SCP waste management objectives. The study would also provide very useful information to the sectors when drawing up their sustainability strategy.

Policy recommendations – The recommendations are targeted at a range of audiences. There is perhaps excessive focus on recommendations for improving data collection However, there are clear recommendations in relation to recycling and bottle banks which are not sufficiently tied back to the results.

SCP001 Page 91 of 196 Outcomes from recommendations - The British Glass Foundation reported that: The report has been particularly useful for our industry in our discussions with government in respect of the sustainability issue. It eventually formed the basis of the pilot study that the Glass Industry did for DTI's "Changing Patterns

SCP001 Page 92 of 196 B/1705 Iron, Steel and Aluminium

Organisation: University of Surrey Environmental Body

Stated objectives (Material flow analysis) The main aim of the project is to conduct a comprehensive Material Flow Analysis of the iron, steel and aluminium industries in the UK. This will contribute to the development in methodology and understanding of waste material flows to provide more effective use of waste materials and reduce impact on landfill

Stated objectives (Economic): The objective of this project is to investigate the economic values associated with the material flows and stocks of iron, steel and aluminium in the UK so as to map the value chain corresponding to the material flows and to see how these values influence material recovery, productivity and waste minimisation. Outcomes: The objectives have been achieved and in methodology terms at least, exceeded. The combined report sets out a time series methodology which is able to track iron & steel (and aluminium) flows over time and explain their stocks within the UK economy. It is evident that the authors of the report had a clear grasp of the use of the methodology that they were using. The only report available via the Biffaward website (the executive summary) did not deal with data issues sufficiently. The more detailed report was made available during the consultation of authors. Overall however, the results and findings do not relate to any future policy issue sufficiently for a good assessment to be given. They would however provide a very clear example of the issues associated with ‘Decoupling/'green growth'/eco-efficiency/resource productivity’ and would provide very useful information to the sector(s) when drawing up their sustainability strategies.

Boundaries – The boundaries were clearly set. Data – The final year for the time series data was 2001. The only report available via the Biffaward website (the executive summary) did not deal with data issues sufficiently. The more detailed report as was made available during the consultation of authors shown that the much of the data came from the Iron and Steel Statistics Bureau (ISSB) in the UK which is dedicated to the collection and publication of data on the UK iron and steel sector. .This has been assessed to be unofficial but largely a reliable source of data. The generation of prompt scrap was needed to be modelled, although as this was a smaller part of the over dataset, this was considered to be an acceptable level of modelling.

Method – The methods used are transparent and robust. The combined report sets out a time series methodology which is able to track iron & steel (and aluminium) flows over time and explain their stocks within the UK economy. It is evident that the authors of the report had a clear grasp of the use of the methodology that they were using. This report has broken new ground in terms of its methodology.

Consistency –Data is consistent with SIC classifications and has been awarded a good assessment even though not explicitly discussed as the sectors are clearly defined by the SIC classification sector.

SCP001 Page 93 of 196 Results – The result are clear and comparable to similar data sets.

Policy Agendas: Policy relevance – The report provides a very clear example of the challenges of ‘Decoupling/'green growth'/eco-efficiency/resource and labour productivity’ which need to be met in a competitive global market and therefore provide information for production SCP policy objectives. Additionally, the study would provide very useful information to the sector(s) when drawing up their sustainability strategies. Policy recommendations – Overall the results and findings do not relate to any pressing policy issue.

Outcomes from recommendations – This study appears to have been very usefully within the sector but does not appear to have been adopted or used by and statutory decision making body in relation to SCP policy. Colin Honess of Corus responded:

‘We have greatly benefited from that work... From our point of view, it has been an extremely useful study as it has provided us with some 'hard' data (albeit for 2000), from which we have been able to supply information on the UK's steel flow balance, for various market sectors… The Biffaward project has provided us with an excellent starting point for further work, together with the realisation that this type of work is essential for promoting the excellent Recyclability of Steel. This will probably include a data update for 2005/2006, so that we have 'recent' mass flow figures for the UK.’ Peer-reviewed articles of the Economic project have been published, accepted for publication, or submitted as follows: ƒ Dahlstrom, K. & Ekins, P. 2005 ‘Eco-Efficiency Trends in the UK Steel and Aluminium Industries: Differences Between Resource Efficiency and Resource Productivity’, Journal of Industrial Ecology, Vol. 9 No.4 (Fall), Special Issue on Eco-Efficiency and Industrial Ecology, pp.171-188

ƒ Ekins, P. 2005 ‘Eco-Efficiency: Motives, Drivers and Economic Implications’, Journal of Industrial Ecology, Vol. 9 No.4 (Fall), Special Issue on Eco-Efficiency and Industrial Ecology, pp.12-14

SCP001 Page 94 of 196 B/1445 Nitrogen UK

Organisation: Horticulture Research International

Stated objectives: To prepare a mass balance of nitrogen flows within the United Kingdom, with particular reference to those in landfill sites

Outcome: The objectives of the study have been fully achieved and put into their scientific, policy and environmental context. The results relate to SIC code classifications. The report successfully compares the various sources of nitrates and thereby provides for the identification of policy action. The results of the study are relevant to policy makers at a national and European level and feed directly into meeting SCP waste management and sustainable agricultural production objectives.

Boundaries – The boundaries of the study have been clearly defined and consistently followed. Data – The results of the study are based on largely official and reliable data sets although the construction of the data sets has only been partly documented. There are been a number of nitrogen retention assumptions which have needed to be made to estimate nitrogen outputs from the agricultural sector. Although these assumptions are necessary and the model used appears to be reasonable, the results for this and some other sectors do represent modelled outputs.

Method – The methods used are transparent and robust. It is not clear if the people who are acknowledged on the first page have had a formal review role required for a satisfactory assessment for the peer review criteria.

Consistency –The results are consistent with the Eurostat guidelines. The data relating to Nitrates relate to SIC classification codes. Additionally they can be compared to and integrated with other similar material national studies with little interpretation or additional work.

Results – The results are clear and well presented and accessible to a policy audience. An understanding of the environmental impact of nitrogen in its different forms needs to be carefully understood from the text.

Policy Agendas: Policy relevance - The results of the study are relevant to policy makers at a national and European level and feed directly into meeting SCP waste management and sustainable agricultural production objectives. The results also feed directly into sectoral analysis/prioritisation at a national level.

Policy recommendations - The report does not provide explicit recommendations but intends to provide information.

Outcomes from recommendations -

SCP001 Page 95 of 196 B/1424 - Solvent Waste in Furniture Manufacturing

Organisation: BFM – The trade association for British Furniture Manufacturers

Stated objectives: The main aims of this project were to (as provided in report): ƒ Generate information on the relative performance of different wood coating sub sectors

ƒ Quantify solvent emissions by coating type and sub-sector.

ƒ Quantify levels of solvent waste generation by sub-sector and analyse the disposal and recovery routes.

ƒ Develop solvent usage and solvent waste generation benchmarks.

ƒ Produce guidance outlining the best operating practices and encourage the uptake of these practices.

ƒ Contribute to the review of PG6/33(97) - the wood coating process guidance note.

Outcome: These objectives were largely achieved. The approach used relied upon scaling up from audits of about 10% of total estimated usage. Information about how the audits were undertaken was not made available. This represents a relatively focused report in terms of its policy usefulness.

Boundaries – The boundaries of the audits were not available but were apparent from the report

Data – The scaled up data was modelled but the method used was transparent and reasonable. Information about how the audits were undertaken was not made available. Method – The methods used were clear but not documented.

Consistency – The Data relates to a SIC classification code

Results – The results are clear and comparable to other recent data sets. With assistance from those who undertook the audits, the results would be likely to be validated.

Policy relevance – The results of the study are relevant to policy making at a sector and company level in the ‘Identification of waste/inefficiency’ and provide useful information to both the industry and the regulator. Reported outcomes – The British Furniture Manufacturers Ltd reported that: The project report still represents the most detailed survey of UK wood coating operations. It has been used for benchmarking purposes by a significant percentage of the UK wood coating sector as well as by their regulators. It has informed the revision of process guidance note PG6/33(04)

SCP001 Page 96 of 196 B/1411 Timber Waste

Organisation: Timber Industry Environment Trust (TIET)

Stated objectives: The objectives of the project are to encourage, inform and enable users and suppliers of timber to reuse the amount of timber in the waste stream and to implement more sustainable waste management practices. The aim is to develop a well-researched central information resource.

Outcome: The study has not been assessed, as no quantification of arisings has been provided. The recommendations appear to be well thought-out and based on some experience of the sector but are not supported by sufficient evidence.

SCP001 Page 97 of 196 B/1448 Waste Glass from Fluorescent Tubes and Lamps

Organisation: National Centre for Business & Sustainability Ltd

Stated objectives: To determine current practices and technologies worldwide for glass recycling. To carry out an extensive analytical survey of the glass recyclate. To determine ways to improve mercury recovery in the current fluorescent tube recycling process. To identify other potential new markets for the waste glass

Outcome: The objectives of the study can only be said to be partly achieved as the report fails to quantify the bulk arising of the material under consideration in terms of mass (i.e. tonnes of glass from fluorescent tubes). Estimates of numbers of fluorescent tubes are provided (100 million, no reference or year provided) and a substance flow analysis (SFA) is provided for the mercury released into the environment. It is this latter analysis which has been assessed. Beyond this, the report does manage to discuss the potential applications and identify the relevant policy and technical challenges related to this material with some success. However, it has not been possible to recommend this report for use in policy making in its present form. It is possible that relatively modest additional work might change this assessment by providing an assessment of the arisings of glass that can be understood by those outside the fluorescent tube sector.

Reported outcomes: The National Centre for Business & Sustainability Ltd reported that: The project provided a useful assessment of the barriers and opportunities for recycling glass from fluorescent tubes and lamps and provided recyclers with a matrix of possible options for recycling. The work informed ongoing research into alternatives for recycling contaminated glass in other areas including CRTs.

SCP001 Page 98 of 196 B/2173 Wood

Organisation: Timber Industry Environment Trust (TIET)

Stated objectives: The aim of the project is to evaluate the flow of embodied wood through the major wood-using industries in the UK (furniture, joinery, pulp and pulp logs, packaging) and the associated mass balances and impacts of these sector-specific flows. In addition, it aims to integrate findings with previous studies on wood in construction and paper industries in order to evaluate the total impact of the use of wood in the UK.

Outcome: The objectives of the study have been achieved. The data is taken from other Biffaward funded studies, as well as ONS sources. The report manages to provide an appropriate mass balance of wood in the UK including an estimate of the increase in embedded wood in 2002. The study performs well overall and shows that there are already extensive systems in place to recover wood waste from the various sectors. This is perhaps only partly the case in the construction sector where lack of time and space does not always allow for the proper management of waste. There are, however, no obvious direct uses for the study in policy making. The data would provide information towards waste management SCP objectives by assisting the various sectors in their sustainable development strategies and meeting their recycling targets.

Boundaries – The boundaries are clearly set around five sectors within the UK. This of course limits the coverage of the study.

Data - The data is taken from other Biffaward-funded studies, as well as ONS sources. There were some concerns reported relating to the quality of the data from the sectoral studies and the ONS data. However, the resultant mass balance represents the best possible with present data sources. It is believed that the study could be updated with the support of the researchers involved. Method – The methods used are transparent and robust. The report manages to provide an appropriate mass balance of wood in the UK including an estimate of the increase in embedded wood in 2002. The Biffaward studies which were used to construct the results have been reviewed. However, as additional work has been undertaken to produce the results, the results can not be said to be reviewed.

Consistency – Data sets out how the data relates to SIC, PCC and ICN.

Results – The result are clear and comparable to similar data sets. However, much of the results are presented in terms of percentages not mass. This means for example that the reader is somewhat misled about the scale of the flow the pie charts relate to. Also, the less than comprehensive coverage of the study (broadly set by the sectoral boundaries of the other wood Biffaward studies) limits the studies relevance.

Policy Agendas: Policy relevance - There are no obvious direct uses for the study in policy making. The data would provide information towards waste management SCP objectives by assisting the various sectors in their sustainable development strategies and meeting their recycling targets.

SCP001 Page 99 of 196 Policy recommendations – The study performs well overall and shows that there are already extensive systems in place to recover wood waste. There are no recommendations provided.

Outcomes – The Timber Industry Environment Trust reported that: Whilst the mass balance project produced the most complete set of wood flow data for the UK industry, there were still uncertainties over some of the data. The project demonstrated how difficult it is to obtain reliable data from a large and fragmented industry that is dominated by a large number of very small companies. The mass balance summary report has already been downloading over 500 times since October. The availability of the report has also been posted on the main TRADA website. It is too early yet to try to assess any impact on the industry. The following hits were received Sept 05 – 21st Feb 06 for the contributing reports:

Mass Balance Summary 711 Mass Balance Report – Construction 294 Mass Balance Report – Packaging 240 Mass Balance Report – Joinery 200 Mass Balance Report – Furniture 190 Mass Balance Report – Pulp 110

SCP001 Page 100 of 196 Regional Studies

B/2230 - Ecological Budget UK Organisation: CURE/SEI Stated objectives: A detailed material flow analysis and Ecological Footprint of the UK by region. New MFAs will be carried out and existing work will be brought up to the same level. The aim will be to use the results to improve resource efficiency through waste minimisation, increased recycling and energy efficiency.

Outcome: This assessment is based on working papers 3 and 4 as well as the ‘Counting Consumption’ results report released on the 7th of March 2006. These reports successfully set out how detailed resource flow analysis and the ecological footprint of the UK by region was generated, and provides a detailed description of the data sources used. The default data sets used in REAP uses official, reliable and annually updated data sets. Considerable care has been taken in the construction of the data sets and the methods used are transparent and reasonable. The output from the model represents a significant step forwards in understanding the material basis of the UK economy.

Boundaries – The boundaries were assumed to be “consumption by households and public as well as commercial services, wherever the consumed products may come from and wherever environmental impacts may occur ("consumer responsibility")” (Report 3, p.4). This boundary was clear, appropriate and in the model at least has been consistently followed.

Data – The default data sets used in REAP uses official, reliable and annually updated data sets. The model is intended to be updated and re-used, however it is not clear from the results report what year the results represent (assessed here to be 2001). Considerable care has been taken in the construction of the data sets and the methods used are transparent and reasonable. The process has been properly documented, however due to the complex ‘black box’ nature of the model; it is not considered likely that any [professional] reader could fully understand the process as would be required for a good assessment. Furthermore and as already reported, the year the data is not well presented. At a national level the material flows, and to a lesser extent the footprint results, do represent real material flows. However, to achieve the various regional breakdowns, modelling or proxy techniques were required. This is similar to the other regional studies.

Method – The methods used are transparent and robust and if combined with the proposed data and model will represent a significant advancement in this area. Perhaps the most substantial step forward is in terms of indirect emissions of CO2 in traded goods.

Consistency – The model is consistent with SIC codes and with Eurostat guidelines.

Results agendas Policy relevance - The likely conclusions will contribute towards a number of SCP policy objectives at a national level associated with production (resource productivity), procurement and consumption (information provision, economic instruments). Via SCPnet, the results will assist in

SCP001 Page 101 of 196 the implementation of SCP at a regional level. The outputs of the model represent significant progress in understanding the material basis of the UK economy.

The results were endorsed by Jonathan Porritt and Alice Owens (SCPnet) during the launch event at the Methodist Central Hall in London on the 7th of March 2006.

B/1303 - Isle of Wight - Island State Organisation: Environmental Body, Imperial College (Best Foot Forward)

Stated objectives: The Island State project is an ecological footprint analysis of the Isle of Wight. In effect, this will measure and communicate a range of resource use and associated environmental impacts. Firstly, a mass balance assessment will be carried out. The Isle of Wight is perfectly placed to carry out this research as the project partners will be able to analyse every material that arrives on or leaves the island

Outcome: This represents an ambitious and mostly successful attempt to map all of the material flows within an island. The challenges and deficiencies in data construction are transparently discussed. The key data concern relates to food flows due to the lack of co-operation from retailers. The key methodological concern relates to the lack of accessibility of the model and its boundaries to a non-expert reader.

Boundaries - Some confusion in describing boundaries of indirect resource and energy requirements. The reader needs to look in detail at the methodology to understand what is being reported.

Classification - No apparent attempt to use SIC or other data classification system although the products being reported on could be broadly related to SIC classifications. Data - Data sources were clearly described and attempts were made to use official data sources. Problems were transparently discussed and cross-checked. Data set could be approximately re- created if required. A number of estimates and modelling assumptions were required to fill in data gaps. Unlike most of the other regional studies, this project involved considerable primary research and hence added to data availability.

Method – Some concern over the ‘black box’ nature of the method. However, the study was undertaken objectively.

Results – The method and results are in line with similar footprint studies and therefore could be said to be comparable to existing work.

Policy relevance – The results of the study are relevant to policy makers on the island and to a lesser extent (due to its island nature) to other areas.

Outcomes – The report has been used to inform various policies on the island including renewable energy and Local Agenda 21 strategies. An ecological footprinting working party was set up at the council and a 4-year update recently completed for the Island’s Council. Additionally a footprint

SCP001 Page 102 of 196 trust set up on the island to take forward many of the issues raised by the report. Reference person provided: Wendy Perrera, Isle of Wight Council

SCP001 Page 103 of 196 B/1646 - London - City Limits Organisation: Best Foot Forward Stated objectives: The City Limits project set out to achieve the following objectives: ƒ To quantify and catalogue the energy and materials consumed by London and Londoners, and where possible map the flows of these resources.

ƒ To calculate the ecological footprint of the citizens of London.

ƒ To compare the ecological footprint of Londoners with other regions.

ƒ To compare the ecological footprint of Londoners with the globally available 'earth share' to estimate ecological sustainability.

ƒ To quantify the ecological sustainability of a range of improvement scenarios.

ƒ To assess the availability and quality of data required to carry out this type of analysis, and in certain instances make recommendations to improve data requirements for resource flow and ecological footprint analyses.

Outcome: The objectives have been achieved. This has been confirmed by Ivon Good of the Chartered Institution of Wastes Management (CIWM (EB)). This project represents an extensive and ambitious project and the results are very powerful at informing regional sustainable development policy. However there are a number of assumptions and compromises required to do this and the extensive database does not provide a transparent model.

Boundaries – The responsibility principle was used for the footprint and the geographical principle used for the resource flow. Data – A number of assumptions and modelling assumptions were required to construct the data sets but these were reasonable and largely done transparently. Overall, much of the underlying data would have to be considered as modelled. Method – The general methods used are transparent but the model represents a ‘black box’ which is hard to validate.

Consistency –Data is broadly consistent with SIC classifications but does consider indirect material flows so is to some degree consistent with Eurostat.

Results – The results are clear and consistent with similar data sets generated using similar models.

Policy relevance – The results of the study are relevant to national policy to some degree and provide information to consumers for meeting SCP objectives. They also are very relevant to regional SD policy and provide targets in consumption areas required to comply with global land constraints.

Policy outcomes: Completed in September 2002, the results were handed to the Greater London Authority for use in developing and monitoring progress towards reducing the environmental impacts of London’s wastes. A number of key issues were incorporated into the ensuing waste strategy for London.

SCP001 Page 104 of 196 ƒ Ecological Footprint adopted as a headline sustainability indicator by the GLA and study extensively referenced in ‘managing resources’ section of London Sustainable Development Strategy (2004). Waste and CO2 selected as indicators32.

ƒ Footprint adopted and updated (3 year update) commissioned by London Sustainability Exchange – supported by London Remade and London First.

ƒ City Limits study now part of Edexcel school curriculum in London33.

Endorsements City Limits reviewed by GLA Economics whose purpose is stated as providing a “firm statistical, factual and forecasting basis for policy decision-making by the GLA group”. Conclusions included: “The ecological footprint concept is part of this new paradigm and is an excellent instrument to make complex problems understandable to… diverse policy stakeholders”

“This study of London's footprint is particularly important because it is the first such analysis of a major world city. For the first time we have an overall picture of London's metabolism, how resources are used and where action might be taken to increase our efficiency and become more sustainable....I welcome the publication of this study and commend it to everyone involved in achieving my vision of making London an exemplary., sustainable world city."

Ken Livingstone, Mayor of London

32 http://www.london.gov.uk/mayor/sustainable-development/susdevcomm_indicators.jsp 33 University Study text (i.e. http://www.nottingham.ac.uk/sbe/planbiblios/bibs/Greenis/A/03a.html)

SCP001 Page 105 of 196 B/1946 - Northern Ireland - Northern Limits Organisation: Arena Network Northern Ireland

Stated objectives: The aim of the project is to provide a resource Mass Balance and Ecological Footprint for Northern Ireland leading to a range of improvement scenarios.

Outcome: The objectives of the study have been achieved. The report sets out future scenarios which are partially grounded in social and environmental trends and limits, whilst some others are somewhat arbitrary. The report’s recommendations are specific, well-presented and very much rooted in regional policy agendas. Additionally, there are extensive and detailed recommendations on how to improve data availability which highlights the challenges met and therefore the potential weaknesses in the underlying data sets.

Boundaries – The boundaries of the region were explicitly defined. Data – The data comes from mostly official and reliable sources and the approaches used to construct the data sets have been transparently presented. As discussed in the recommendations section and as with similar regional studies, much of the data has been modelled on UK data. Method – The combination of mass balance, ecological footprint and sustainability assessment provides a powerful combination. The eco-footprint model represents a ‘black box’ which would be time-consuming to validate. Consistency – The results are presented in a way which would allow comparisons with SIC classifications. Indirect flows are considered through the eco-footprint results hidden flows are not.

Results – The result are clear and comparable to similar data sets.

Policy Agendas: Policy relevance - The results of the study are relevant to policy makers at a regional level as they provide information for SCP objectives of sustainable consumption though the provision of information to regional stakeholders and policy makers. The results could also feed directly into the provinces sustainable development strategy.

Policy recommendations – The report sets out future scenarios which are partially-grounded developments in social and environmental trends and limits whilst some are somewhat arbitrary. The report’s recommendations are specific, well-presented and very much rooted in regional policy agendas. Additionally, there are extensive and detailed recommendations on how to improve data availability, which highlights the challenges met and therefore the potential weaknesses in the underlying data sets.

Outcomes from recommendations – ARENA Network Northern Ireland reported that: One of the main purposes of conducting the research was to have it established a key indicator in the strategies of Government. Follow up meetings with the department concerned and the minister responsible have ensured that Ecological Footprinting will be an indicator within the emerging Sustainable Development strategy for Northern Ireland

SCP001 Page 106 of 196

SCP001 Page 107 of 196 B/1831 – Scotland Organisation: Viridis

Stated objectives: The project is a mass balance study of Scotland and forms part of the Sustainable Resource Use Programme by Biffaward. It is proposed to carry out a resource flow and Ecological Footprint Analysis of Scotland. This is a baseline study intended to deliver a platform from which further research, study and data improvements can be made.

Outcome: The objectives of the study have been successfully achieved in a transparent and well- presented way. The combination of mass balance, ecological footprint and sustainability assessment provides a powerful combination of ‘real’ material flow data with the perhaps more readily useable concept of global hectors required to meet the material responsibility.

Boundaries – The geographical boundaries were described although the boundaries were complex for the fishing industry.

Data – The data sources are largely reliable and official. Overall the data collection represents an impressive achievement. The key weakness of the data set is that the production data is based (i.e. modelled on) worker productivity from UK-wide data. This issue has been transparently discussed in the report. Much of the input consumption data does need to be considered as modelled. Method – The combination of mass balance, ecological footprint and sustainability assessment provides a powerful combination. The eco-footprint model represents a ‘black box’ which would be time consuming to fully validate. However, the approach has been well documented and has been reviewed by others.

Consistency –The results are presented in a way which would allow comparisons with SIC classifications. Indirect flows are considered through the eco-footprint results and hidden flows within the mass balance work.

Results – The results are clear and consistent with similar data sets generated using similar models. The report provides a very useful comparison of similar geographical results.

Policy Agendas: Policy relevance - The results of the study are clearly relevant to a range of policy areas at a Scottish level. As the dissolved Scottish government holds a much greater range of policy responsibility than the English regions, the results are likely to be even more useful than the studies for the English regions. In terms of SCP policy, the results provide very useful information for policies which aim to decouple production with material use, as well as manage overall consumption and have been adopted by Scottish executive and has therefore ben assessed to have fed directly into the process. The report could also feed into providing targets for future sustainability strategies in Scotland.

Policy recommendations – The study does not explicitly provide policy recommendations but does explore a number of scenarios based on social trends and sustainability constraints (one planet living).

SCP001 Page 108 of 196 Outcomes from recommendations –

The results of study have been taken forward by the Scottish Executive, SEPA and WWF Scotland: • The 3-year Scottish Global Footprint Project, funded by WWF Scotland, the Scottish Executive, Scottish Power, SEPA, North Lanarkshire Council, Aberdeen City and Aberdeenshire Councils has as one of its core aim the development of policy scenarios to reduce Scotland’s footprint 34. • The Scottish Executive has formally adopted the footprint as an indicator and has committed to replicating the Scotland’s Footprint study in 2008 35. • Scotland’s Footprint is in Our Future, Scotland’s Sustainable Development Strategy (2005) and reducing the footprint is stated as a key objective of the strategy 36. • The Scottish Executive also followed up Scotland’s Footprint with a review of the impact of Scotland’s five largest cities. Endorsements –

Scottish Exec/SEPA/WWF Scotland (extract from report Foreword, Calum MacDonald (SEPA), Simon Pepper (WWF Scotland): “…Scotland’s Footprint is seen as a timely report. The report provides a basis for analysing Scotland’s current production and consumption patterns and the use of resources in the Scottish economy.”

Prof George Fleming: “a valuable tool in the debate on the future direction of Scotland's economic, social and environmental policies.”

34 www.wwf.org.uk/filelibrary/pdf/scotlandfootprint.pdf 35 http://www.scotland.gov.uk/News/News-Extras/footprint 36 http://www.scotland.gov.uk/Publications/2005/12/1493902/39032

SCP001 Page 109 of 196 B/1699 - South East - Taking Stock

Organisation: EcoSys Environmental Management & Education

Stated objectives: Mass Balance and Ecological Footprint study as baseline study to: draw together existing data from a variety of sources, identify data gaps, source currently unavailable data on resource flows through the SEEDA region; provide an integrated data set to inform other projects and policy fora such as market development initiatives, business efficiency initiatives, etc.; engage policy makers and public in process by using Ecological Footprint for sustainability assessments; demonstrate the effect of waste market development and environmental improvement initiatives on the sustainability of the SE region; provide a replicable model for other regions

Outcome: The objectives of the study have been achieved. The report sets out future scenarios which ground developments in social and environmental trends and limits. Beyond this and perhaps in addition to similar studies, the report’s recommendations are specific, well-presented and very much rooted in regional policy agendas. However, the underlying data is not discussed at all in the main report. A brief report of the key sources of underlying data and key weaknesses would have greatly strengthened the validity of the recommendations for the majority of readers who realistically will only read the summary report.

Boundaries – The boundaries of the region were not explicitly defined which might be problematic for readers unfamiliar with the boundaries for the SEEDA region. Data – The underlying data is not discussed at all in the main report but discussed in some detail in other parts of the report. This separation is sensible and permitted a much more policy-focused report. However, a brief report of the key sources of underlying data and key weaknesses would have strengthened the validity of the recommendations. The full technical report did appear to fully document how the data sets were constructed but perhaps in too much detail for the reader to gain an overall impression. Overall it would appear that this report fared similarly to the other regional studies in their data collection and needed to construct production data based on employment numbers.

Method – The combination of mass balance, ecological footprint and sustainability assessment provides a powerful combination. The eco-footprint model represents a ‘black box’ which would be time-consuming to validate.

Consistency – The results are presented in a way which would allow only approximate comparisons with SIC classifications. Indirect flows are considered through the eco-footprint results and hidden flows during the mass balance work.

Results – The result are clear and comparable to similar data sets.

Policy Agendas:

SCP001 Page 110 of 196 Policy relevance - The results of the study are relevant to policy makers at a regional level and feed directly into almost all of the seven SCP objective stages of material cycles. The results also feed directly into the regions sustainable development strategy.

Policy recommendations – Like similar regional studies, the report sets out future scenarios which ground developments in social and environmental trends and limits. Beyond this and perhaps in addition to similar studies, the report’s recommendations are specific, well-presented and very much rooted in regional policy agendas.

SCP001 Page 111 of 196 B/1996 South West - Stepping Forward

Organisation: South West England Environmental Trust

Stated objectives: The aim of the project is to provide an integrated Resource Flow data set to inform other projects and policy fora such as market development initiatives, business efficiency initiatives, target-setting etc as part of a Mass Balance and Ecological Footprint study.

Outcome: The objectives of the study have been successfully achieved in a transparent and well- presented way. The combination of mass balance, ecological footprint, sustainability assessment and future scenarios provides a powerful combination of ‘real’ material flow data with the perhaps more readily useable concept of global hectors. The underlying approach is very similar to the Scottish study which was also undertaken by Best Foot Forward.

Boundaries – The geographical boundaries were described.

Data – The data sources are largely reliable and official. As with any regional study, a complete data set represents an impressive achievement. As with the other regional eco-foot printing studies, the key weakness is that the production data is based (i.e. modelled on) worker productivity from UK wide data. This issue has been transparently discussed in the report. The underlying input data does however need to be considered as largely modelled.

Method – The combination of mass balance, ecological footprint, sustainability assessment and future scenarios provides a powerful combination. The eco-footprint model represents a ‘black box’ which would be time-consuming to fully validate. However, the approach has been well documented and has been reviewed by others.

Consistency –Data is consistent with SIC classifications and indirect flows are considered through the eco-footprint results and hidden flows during the mass balance work.

Results – The result are clear and consistent with similar data sets generated using similar models. The report provides a very useful comparison of similar geographical results.

Policy Agendas: Policy relevance - The results of the study are clearly relevant to a range of policy areas at a regional level. In terms of SCP policy the results provide very useful information for policies which aim to decouple production with material use, as well as manage overall consumption. Beyond this and more widely the report could feed into providing targets for future sustainability strategies in the South West.

Policy recommendations – The study explores a number of scenarios based on social trends and sustainability constraints (one planet living). These are well researched and grounded and represent a useful output for policy makers.

Outcomes from recommendations –

SCP001 Page 112 of 196 Report endorsed by SW RDA and SW Regional Assembly and South West Observatory is promoting footprint. Provisional results referenced in SW Integrated Regional Strategy (which makes SCP a key goal).

Endorsements –

“In making the best use of available data the report is an important aid in quantifying … environmental impacts. … A welcome key outcome from the report is the identification of the areas that can most readily be addressed…” Extract from report Foreword, Chanel Stevens, SW Regional Assembly

“Stepping Forward provides a lot of that data for decision makers in the south west - about energy use, material consumption, transport, water and land use, and so on. It takes it out of the zone of technical experts and policy wonks, and pitches it directly at all of us, both as consumers and citizens.” Jonathon Porritt, SDC (see www.steppingforward.org.uk)

SCP001 Page 113 of 196 B/2011 Wales

Organisation: WWF-UK

Stated objectives: The aim of the project is to provide a material flow analysis and ecological footprint of Wales and two sub-regions (Cardiff County Council and Gwynedd Council). The project will be used to build the capacity of institutions and agencies and to influence and inform national and regional policy. Outcome: The objectives have been achieved. This is a well-executed study which extends the analysis to include material flow analysis and input-output analysis. The underlying data (from CACI, see section A.3.2 of main report) are modelled consumption patterns based on socio- economic indicators.

Boundaries – The geographical and ‘responsibility’ boundaries are clearly defined and followed. Data – The year of data was not easily found. The year of 2001 was taken from p.4 of the summary report with reference to travel. The underlying consumption data (from CACI, see section A.3.2 of main report) is a modelled data set on consumption patterns based on socio- economic indicators.

Method – The methods used are transparent and robust and represent a noticeable refinement on the Ecological Footprint methodology. Consistency – The results are only broadly consistent with SIC classifications and largely consistent with Eurostat guidelines as they also include indirect material flows.

Results – The result are clear and comparable to similar data sets from other regional Ecological Footprint studies.

Policy relevance – The results of the study are highly useful and relevant to policy makers at a Regional (Welsh) level and provide information for policies intended to meet the consumption part of SCP objectives. In addition, the results will feed directly into the Regional (Welsh) sustainable development strategy.

Sectoral studies

SCP001 Page 114 of 196 B/1597 Agricultural Waste (Mass balance)

Organisation: C-Tech Environmental Group

Stated objectives: It is proposed to undertake a study of the waste streams and energy inputs for major sectors of agriculture, and follow the fate of these streams, from their arisings to final handling. The study will be broken down into three parts. The first section will evaluate the existing operation of farms and the drivers for change. The second part will investigate the potential technologies and techniques which can be used to reduce energy and waste arisings and also the technologies which can take the waste and generate useful by-products or energy. Finally, the results of the work and supporting economic evaluations will then be disseminated to the agricultural community and waste management industry via editorial, website and preparation and distribution of a specific publication to promote the results of the study.

Outcome: This study largely followed its objectives and provides a valuable point of knowledge of the direct material basis of agricultural production, and how to manage waste arisings. However, the failure to define a boundary that included off-site inputs (of at least electricity production) means that the later analysis of waste management options does not have a sufficient material basis to make comparisons.

Boundaries – The boundaries were not explicitly set. This was important for this study as the distinction between agricultural production as an economic entity and ‘the environment’ is often a fine one. In addition, the boundary, as apparent, only considered on-site material requirements which limited the sophistication of the later energy analysis.

Data – The data was from an official source and assumed to be published, based on mostly observations and is periodically updated. The level of documentation of how the data was generated was limited.

Method – The methods used are transparent and largely robust but did not attempt to balance inputs with outputs

Consistency –Data was approximately consistent with SIC classifications but did not meet the requirements to be consistent with the Eurostat guidelines.

Results – The result are clear and assumed to comparable to similar data sets.

Policy relevance – The results of the study feed directly into national SCP waste management planning. The study could also feed directly into regulating the sector as well as helping the sector further develop its sectoral sustainability strategy. The study could be applied to the identification, assessment and target setting parts of the policy-making cycle.

Outcomes – It is believed that this study has been influential in policy making within the Energy sector. 900 paper copies (plus unquantified pdf downloads) have been distributed directly to relevant persons and at numerous events, bringing the mass balance data as well as policy and legislation

SCP001 Page 115 of 196 information to the attention of Government Departments, Local Councils, Power companies and trade associations.

Also the Director of Environmental Services at the Business Partnerships Executive complemented the report and who together with a colleagues in the rural development partnership were considering embarking on a project. Also the local Agenda 21 group based in South Shropshire were interested in the report and it recommendations.

The studies in Agriculture have prompted further R&D funding activity from DEFRA looking at the potential for Anaerobic Digestion to generate energy from waste. Calls for proposals have recently been issued by DEFRA with work due to start early next year. Also, it has also been reported (by RSWT) that the report led to the setting up of the Agricultural waste stakeholder forum as launched by the Prime Minister.

SCP001 Page 116 of 196 B/1265 Agricultural Waste (Strategy)

Organisation: Westcountry Rivers Trust (The Environment Agency) Stated objectives: To facilitate the development of a national strategy for agricultural waste, with specific objectives to collate information on waste arisings and practices, assess the options, review experience in other Member States, and recommend a strategy Outcome: The stated objectives have all successfully been achieved and the data appears to represent the best practicable outcome from the exercise.

Boundaries – The boundaries have not explicitly been set although the farmyard boundary appears to have been followed.

Data – The data collection relies upon surveys and statistics and the use of waste generation factors. The results are therefore to some extent modelled although this would appear to be the only practicable way of generating such a data set for the sector. The age of the results and the difficulty involved with updating the exercise does represent a noticeable weakness in the study. Method – The methods used are transparent although a large number of assumptions have needed to be made.

Consistency –The data is only broadly consistent with SIC classifications as the waste arisings do not relate to production type and some farm diversification activities would not fit within agricultural production.

Results – The result are clear and could be compared to similar data sets. Policy relevance – The study was set out and manages to develop a strategy in response to European waste legislation. The results of the study are therefore highly relevant to policy makers and could feed directly into waste management planning at a national and local level. The report also feeds directly into sectoral sustainability strategies.

SCP001 Page 117 of 196 B/1663 Automotive

Organisation: Viridis

Stated objectives: The project is a mass balance study of the motor industry. It will look at areas other than new vehicle manufacture and ELV recycling where research activity is presently concentrated, including routine servicing and repair, accident and damage repair, third party component manufacture, component reconditioning, component and materials recovery, recycling and re-use (including oils, batteries, tyres etc). Outcome: The objectives of the study have largely been achieved, although a full set of data inputs for the full product life cycle was not (for example energy inputs into ELV processing). Wide and extensive data sources were used as explicitly discussed and documented in an annex. As presented, the reliability of some of the data is questionable, although the bulk of the data has come from official and reliable sources. The mass balance method used was transparent and robust objectively carried out. The scope of the analysis includes the use of vehicles as well as the oxygen used in combustion with was required to generate a mass balance. The results were clearly presented and discussed and the recommendations supported by the results. It is clear that the sector is taking the lead in terms of sustainability. The results of the study are relevant to SCP policy and would feed directly into resource productivity policies at a national level and integrated product policies at European level. The study would (and no doubt has) feed into the sectoral analysis at a national level and the sector's sustainability strategy. Boundaries – The boundaries have been explicitly defined and followed. The scope of the analysis includes the use of vehicles.

Data – Wide and extensive data sources were used as explicitly discussed and documented in an annex. As presented, the reliability of some of the data is questionable, although the bulk of the data has come from official and reliable sources.

Method – The mass balance method used was transparent and robust objectively carried out. The scope of the analysis includes the use of vehicles as well as the oxygen used in combustion with was required to generate a mass balance. The study was conducted with help of an advisory group and has therefore assessed to have been partially reviewed. Consistency & Results –The data is intended to be consistent with SIC classifications. Indirect flows where not however accounted for. The results would also feed into the overall database of material codes envisaged by the Biffaward programme.

Policy Agendas:

Policy relevance - The results of the study are relevant to SCP policy and would feed directly into resource productivity policies at a national level and integrated product policies at European level. The study would (and no doubt has) feed into the sectoral analysis at a national level and the sector's sustainability strategy.

Policy recommendations – The results were clearly presented and discussed and the recommendations are supported by the results. The wide scope of the analysis (to include product

SCP001 Page 118 of 196 use) allowed for a wider perspective to be considered. It is clear that the sector is taking the lead in terms of sustainability.

Outcomes from recommendations – As reported by Viridis: The mass balance data in the report provides important information on resource use and the technical recommendations made will help to improve the resource productivity of the motor industry. The main conclusion of the study was that vehicle use accounts for 98% of the emissions generated by the UK motor industry, thus reinforcing the message that reduction in road transport needs addressing to ensure the UK can meet its Kyoto protocol’s commitments.

SCP001 Page 119 of 196 B/1505 Ceramics

Organisation: South West England Environmental Trust

Stated objectives: To provide information on resource flow of raw material and processing energy in the ceramics industry using the mass balance approach Outcome: The objectives of the study have been achieved. This is an impressive study which has faced a number of challenges in terms of the collection of data from a diverse sector. The use of sample site questionnaires to collect the data and scaling up the outcome was the only realistic solution in this case. An attempt was made to validate the data as collected although the possibility of a biased self-selected sample was not discussed anywhere in the report. The researchers involved clearly understood the mass balance principles as well as the objectives of the Biffaward programme. The results were clearly presented and discussed but the recommendations were very weak relative to the rest of the report. A modest re-consideration and re-drafting of the report may well make this report significantly more useful to policy makers.

Boundaries – The boundaries were clearly defined.

Data and methods – The study faced a number of challenges in terms of the collection of data from a diverse sector. The study used sample site questionnaires to collect the data and then scaled it up. An attempt was made to validate the data as collected although the possibility of a biased self-selected sample was not discussed anywhere in the report. The lack of mass balance was convincingly explained by the evaporation and lack of measurement of water.

Consistency –Data was consistent with SIC classifications overall but the sub sectors used to collect the data did not match SIC codes exactly. Results – The results are clear and comparable to other sector-level mass balance studies.

Policy Agendas: Policy relevance - The results of the study are relevant to SCP policy and would provide information towards resource productivity policies. The study would also feed into the sectoral analysis at a national level and the sector's sustainability strategy.

Policy recommendations – The results were clearly presented and discussed but the recommendations were very weak relative to the rest of the report and failed to make any reference to the analysis. The report might have further explored the potential for better utilisation of other sectors’ waste as an input, as well as what would be required to divert the sector’s waste away from landfill. A modest re-consideration and re-drafting of the report may well make this report significantly more useful to policy makers.

Outcomes from recommendations -

SCP001 Page 120 of 196 B/1601 Chemicals

Organisation: South West England Environmental Trust

Stated objectives: To provide information on resource flow of raw material and processing energy in the chemical industry mass balance study Outcome: The report easily achieved the modest objectives as stated. The report and the underling data it reports on represents a valuable and reliable resource.

Boundaries – Clearly defined

Data – Well documented and based on reliable observations. It has been assumed that the data sets could be updated if required. Method – The methods used are transparent, robust and based on commonly used approaches to mass balance and material flow analysis.

Consistency – The study is entirely consistent with SIC classifications and is consistent with Eurostat guidelines within the boundaries of the study.

Results – The result are clear and comparable to similar data sets from other sectors.

Policy relevance – The conclusions and recommendations perhaps let down what is otherwise an excellent study. The results of the study are relevant to policy makers at national level and provide contextual information for meeting resource productivity and waste management SCP objectives. More generally the results provide information for sectoral analysis for both government and the sector.

SCP001 Page 121 of 196 B/1406 Construction

Organisation: Viridis

Stated objectives: To study the whole life effects of use of natural resources, process burdens, performance criteria and end-of-life re-use/recycling in order to identify ways of reducing waste, improving environmental performance and recycling materials

Outcome: The objectives of the study have largely been achieved. The end-point of waste streams is not, however, considered. Otherwise this is a very thorough and useful report which systematically presents the data mostly by SIC codes. The only criteria preventing a green assessment was the age of the data set, 1998.

Boundaries – The boundaries were clearly defined.

Data – The data was considered a little out of date although to collate such a full data set for any year is recognised as an achievement. The extensive range of data sources made it difficult to assess reliability, availability and level of documentation. Overall it was clear that official or reliable data sources had been sought and the methods of construction and estimation had been made transparent. Method – The methods used are transparent and robust.

Consistency –Data consistent with SIC classifications and would have been fully consistent with Eurostat guidelines if memorandum items for balancing (such as oxygen) were included. It is possible that this might been upgraded to a good assessment if the underlying workings were made available.

Results – The result are clear and comparable to similar data sets. Policy relevance – The results of the study are very accessible and feed directly into SCP policy in terms of production and resource productivity at a national, sector and company level. Additionally the study provides information for sectoral analysis and the identification of significant impacts.

Outcomes from report – For the first time, this report identifies the UK’s construction industry’s resource use and thereby contributed to the overall debate of sustainability in the construction industry. The authors were awarded the ICE Halcrow Prize 2004 for the joint paper “Resource flow analysis: measuring sustainability in construction”, following the conclusion of research project.

SCP001 Page 122 of 196 B/2068 Education

Organisation: Waste Watch

Stated objectives: The aim of the project is to obtain reliable data on resource consumption, use and disposal from the education sector in order to obtain an accurate picture of the sector's impact. Benchmarking performance will help management to improve efficiency by setting targets and applying best practice. Targeted training and a resource management toolkit will facilitate the delivery of best practice throughout the sector and increase awareness of resource management issues.

Outcome: The report successfully manages to cover the ground as set out in the objectives and has clearly achieved a lot whilst working with a fragmented and heterogeneous sector, a sector which does not make extensive measurement of such flows. The main weakness in the study relates to the approach to data collection as discussed below in the section on data.

Boundaries – The boundaries were not explicitly defined but a consistent boundary does appear to have been followed. It was unclear under what circumstances halls of residence were included. Data – The approach taken (audits of sample institutions) would appear to be the only practicable way of collecting data within this sector. Care was taken to ensure a representative sample was used in terms of region and type of school. However, as acknowledged on page 12, the participating institutions were self- selecting and therefore may have been more environmentally diligent. This might have been assessed by comparing the level of environmental management of participating institutions with the wider sector and then perhaps controlled for using statistical techniques with reference to the level of environmental management.

Method – The methods used meant that a mass balance was achieved although there was a lack of exploration of the influence of stocks. Consistency –The results are consistent with SIC classifications but do not follow the Eurostat guidelines.

Results – The result are clear but only partly comparable to similar data sets (such as B/1224). The results could be revised using the approach as discussed although this would likely be an extensive exercise to do ex-post and would require access to the full set of workings and data.

Policy relevance – The results of the study have a number of applications in terms of providing a benchmark for other institutions. Furthermore the data would appear to be of a suitably high enough quality to provide information for public procurement policies and sectoral analysis. However, the weaknesses as identified would prevent the results being used directly to inform policy without further work.

SCP001 Page 123 of 196 B/2043 Electricity Generation

Organisation: C-Tech Environmental Group

Stated objectives: The aim of the project is to develop a mass balance for the generation of electricity, such that the real environmental cost of the various methods of generation is transparent. This information will be normalised and the essential information can be fed into the other existing mass balance studies to enhance their accuracy and reliability.

Outcome: The outcome represents a detailed mass balance of electricity generation but could not be said to have achieved its environmental objective (make transparent the real environmental cost of the various methods of generation) as the study did not include fuel cycles in the analysis. It does however represent a very useful reference document which begins to provide some indication of the environmental costs.

Boundaries – Not mentioned or explicitly defined beyond the title of the report. This leads to some confusion as to what is actually being reported (the report refers to fuel cycles but does not include them in the scope).

Data – The report would have been improved by a section discussing data sources and their weaknesses. The sources as used and method of construction are however a sound way of modelling emissions.

Method – The methods used are transparent and robust. Consistency –Results are consistent with SIC classifications. As the report includes the use of water and air this means that the results are also consistent with the Eurostat guide.

Results – The result are clear and comparable to similar data sets and would need little or no work for them to be used to inform policy on the understanding that fuel cycles are not included.

Policy relevance – The conclusions of the report are a little weak and do not make the most of policy useful results. The results of the study are highly policy- relevant due to the comparison of environmental emissions from nuclear and competing technologies (i.e. the energy White Paper). The study would however only be informative within such a debate and would fail to answer the crucial question relating to the uranium fuel cycle. Beyond this the results have some contextual applications in SCP objectives (i.e. resource productivity) and could feed directly into sectoral analysis. Outcomes – It is believed that this study has been influential in policy making within the Energy sector. 900 paper copies (plus un-quantified pdf downloads) have been distributed directly to relevant persons and at numerous events, bringing the mass balance data as well as policy and legislation information to the attention of Government Departments, Local Councils, Power companies and trade associations. The Mass Balance was included in the IEA (International Energy Agency) Clean Cola Centre Coal Abstracts database.

SCP001 Page 124 of 196 B/1634 Exhibition Industry

Organisation: Midlands Environmental Business Club

Stated objectives: An overall programme is being developed to minimise the waste of the UK exhibition industry and to operate to a uniform standard related to the values of sustainable development. Outcome: The study is the first stage in a process towards a sustainable development strategy for the sector. The study set out to understand attitudes and waste management practices within all parts of the sector. A rough estimate of the waste arsing from the sector was made through interviews. However, due to a number of limitations in the approach taken, this figure is not considered to a robust estimate for policy purposes. The report ends with an industry policy statement/agreement from the three relevant trade associations.

Boundaries – The boundaries were not discussed and there was some uncertainty as to whether a boundary had been defined and followed. For example, it was not clear if delegates waste disposed of off site was included.

Data & methodology – The data was generated through a reasonably large number of interviews (202). This represented about 90% of venues, 60% of organisers and 30% of contractors. These could be considered as a sufficiently large enough sample size. Furthermore, a profile of contractors contacted was provided. However, this profile was not compared to a profile all of the contractors who work in the sector, it has not therefore been able to assess whether the sample is a representative one. The process has been well documented and, with the exception of the high level estimate of arisings, represents real material flows rather than modelled data.

Consistency – The data is consistent with SIC classifications (74.84/3 Activities of exhibition and fair organisers) but does to cover indirect flows. The quantitative findings are comparable to some similar studies from other sectors are likely to be in different formats

Policy Agendas:

Policy relevance - The results of the study have moderate relevance to waste management policy makers at a local level. The study feeds directly into the company’s sustainability strategy. Policy recommendations – The study is the first stage in a process towards a sustainable development strategy for the sector. The study set out to understand attitudes and waste management practices within all parts of the sector. The report ends with an industry policy statement/agreement from the three relevant trade associations.

SCP001 Page 125 of 196 B/1536 Financial Sector

Organisation: Waste Watch

Stated objectives: The objective is to put waste management issues in context by highlighting the inter-relationship between waste management practices and broader environmental issues eg. stimulating the market for raw materials though the consideration of recycled content in purchasing decisions eg paper and carpets. This will also identify opportunities for positive relationship partnerships with organisations outside the sector and will complement the on-going research to footprint other sectors, particularly the complementary public sector project and the material-specific studies on paper and fibre.

Outcome: The objectives of the study have been achieved. The study sets out to provide direct flows of material though a number of case study sites. The study will have been of use to many within the sector, or any office-based sector, as an environmental management performance benchmark. The analysis also reveals a number of interesting dynamics such as the move to electronic offices and higher density rates have in fact increased paper use. The study does not attempt to scale up for the sector as a whole, and there is no discussion of how the sites were selected and the potential for a biased sample of the most environmentally performing sites. It is not therefore considered that this study can be used to inform policy.

Boundaries – The boundaries of the sector were defined as the accountancy, banking, insurance and pension sectors in the UK. The material boundaries of the sector were not defined; although a boundary of direct operational flows were followed. Materials used in renovation were sometimes excluded. Data & method – It is likely that standards of environmental management will have improved since 2000 when the survey was conducted. The study does not attempt to scale up for the sector as a whole, and there is no discussion of how the sites were selected and the potential for a biased sample of the most environmentally performing sites.

Consistency –Data is consistent with SIC classifications

Results – The result are clear and comparable to similar data sets, although as later studies from similar sectors use the ecological footprint to measure consumption, the results are not fully comparable to some other studies.

Policy Agendas: Policy relevance - It is not considered that this study can be used to inform policy. Additional work looking at the sample of surveyed sites may be able to validate the sample and permit scaling up of the data. However, this would take a moderate amount of additional work, may not permit scaling up and would produce results based on a survey conducted in 2000.

Policy recommendations – The analysis also reveals a number of interesting dynamics such as the move to electronic offices and higher density rates have in fact increased paper use.

SCP001 Page 126 of 196

SCP001 Page 127 of 196 B/2182 Food and Drink Processing

Organisation: C-Tech Environmental Group

Stated objectives: The aim of the project is to carry out a mass balance study for the food and drink processing sector and to publicise the results to stakeholders and policy makers through generation of a high quality report and websites.

Outcome: The report successfully generates a mass balance of food entering and exiting this diverse sector. The potential for double counting is carefully managed and it is apparent that the concepts behind mass balance are fully considered. The results strictly follow SIC codes as allocated to the sector. The results (based on 1999 data) are a little old for future policy purposes as the sector was undergoing a large number of challenges and changes when at the time of the study. Care was taken to use official statistics where possible although it has not been possible to verify to what extent other data has been used as the process has not been fully documented. Water and energy are not fully included. The report is highly focused proving guidance for the sector to manage the many regulatory changes in waste management practices. The report is relevant to policy makers at national and European level, provides information for meeting waste management SCP objectives and could feed directly into the sector’s sustainability strategy.

Boundaries – A boundary was defined which is distinct from the agricultural waste study. The boundary of the study does not account for imports and exports of ingredients across the UK boundary.

Data - The results strictly follow SIC codes as allocated to the sector. The results (based on 1999 data) are a little old for future policy purposes as the sectors were undergoing a large number of challenges and changes when the report was being written. Care was taken to use official statistics where possible although it has not been possible to verify to what extent other data has been used as the process has not been fully documented. Method – The potential for double counting is carefully managed and it is apparent that the concepts behind mass balance are fully considered.

Results & Consistency – The results strictly follow SIC codes as allocated to the sector. Water and energy are not fully included so the results are not consistent with Eurostat guidelines and would be unlikely to be comparable with the results from other sectors. It is therefore likely that a moderate amount of work would be required to finish off the data sets.

Policy agendas –

Policy relevance - The results of the study are relevant to policy makers at national and European level and provide information for meeting waste management SCP objectives and could feed directly into the sector’s sustainability strategy.

Recommendations - The report is highly focused, providing guidance for the sector to manage the many regulatory changes in waste management practices. No explicit policy recommendations are provided although the report successfully sets out the likely outcome from planned regulations.

SCP001 Page 128 of 196 Outcomes - The findings of the study have been formally presented at three food industry events dealing with aspects of education or preparation for impending legislative changes or preparation for participation DEFRA consultations in this area. One output was the preparation of a roadmap for the chilled food sector. The study has also provided contributory reference material to other speakers on a similar number of other occasions. 1000 paper copies of the report (plus an unquantifiable number of pdf downloads) have been distributed at numerous food and general industry events, bringing food waste policy and legislation to the attention and up the agenda of individual food processors, and groups.

The study has informed steering provided to a current DEFRA funded PhD studentship in the area of Sustainable Value Stream Mapping for the Chilled Foods Industry.

SCP001 Page 129 of 196 B/1584 Foundry

Organisation: Castings Technology International

Stated objectives: This project, will map the flows of materials within the foundry industry from their extraction, through manufacture and use to final disposal or reuse. This will be carried out for the full range of cast materials taking into account production methods and volumes

Outcome: The objectives of the study have only been achieved in part as the scope of the analysis does not include the extraction of material. The results of the study set out to provide benchmarking information to the sector to identify waste/inefficiency. The data is based on surveys of about 18% of the sector. The risk of bias was explored and managed. As neither the year of the data survey nor the year of the report’s publication was provided, it has not been possible to provide an assessment for the age of the data. The study made every effort to ensure the results followed a common format and the data is consistent with SIC classifications. The results are clear and usable with similar studies such as the iron, steel and aluminium sector report.

Boundaries – The boundaries or the scope of the analysis are not explicitly discussed, although the metal sectors included are defined and the boundary of foundry sites is consistently followed. Data – The data is based on surveys of about 18% (by mass) of the sector. The risk of bias (by site size and location) was explored and managed. The quality of the data can therefore be considered as satisfactory. Neither the year of the data survey nor the year of the report’s publication was provided. Some of the energy data was reported to be from 2000 but it was not made clear if this was consistent with the rest of the data. It has not therefore been possible to provide an assessment for the age of the data. Method – The mass balance method used is transparent and robust.

Consistency –Data is consistent with SIC classifications

Results – The results are clear and usable with similar studies such as the iron, steel and aluminium sector report.

Policy Agendas: Policy relevance - The results of the study set out to provide benchmarking information to the sector to identify waste/inefficiency. The mass balance results would also provide information to policy makers at a national level in meeting SCP resource productivity objectives and if they were to undertake sectoral analysis.

Policy recommendations – The report provides very clear recommendations on how better to collect and present future data sets relating to the sector.

SCP001 Page 130 of 196 B/1410 Furniture Packaging

Organisation: Furniture Industry Environmental Trust

Stated objectives: To critically evaluate the way in which the furniture industry packages its products in order to reduce waste and implement more sustainable packaging procedures. Outcome: The stated objectives were largely achieved. This is a highly sector-focused report which has little policy benefit outside the sector. Although total tonnages are estimated, the analysis does not extend beyond the ‘sight’ of the sector to consider production or disposal of packaging material. The study presently fails to be suitable for policy use due to the lack of documentation of the data collection.

Boundaries – Not discussed

Data – Based on audits, details not provided.

Method – Scale up from 22 site audits (16.7% of market) based around standardised products. This provided an estimate for total tonnages.

Consistency – Consistent with Eurostat within the boundaries of the study (i.e. not the production of the packaging). Results – The result are clear but have not been compared with other data sets.

Policy relevance – The results of the study are relevant to policy makers at company level and feed directly into the identification of waste inefficiencies. The report also provides information for the planning of waste management at a national and local level and provides national policy makers with information about sector analysis and prioritisation.

Reported outcomes –The Furniture Industry Environmental Trust reported that: The project has had numerous benefits to the industry with awareness of the 'true' cost of packaging and packaging waste encouraging the industry to reconsider the way they pack and transport their goods. It is evident that more furniture manufacturers are using less packaging, by moving away from the total concealment in cardboard to using top and bottom caps and edge protectors with shrinkwrap to the outside. Within the office furniture and kitchen sector there is also more evidence of using returnable packaging such as blankets. This has been most appropriate for those companies who utilise their own installation teams, ensuring improved handling of goods and returned packaging.

SCP001 Page 131 of 196 B/1640 Housing Organisation: BioRegional Development Group

Stated objectives: To promote widespread use of reclaimed materials in mainstream construction, by collating (in a mass balance format) and disseminating information on the materials and methods used at Beddington Zero Energy Development (BedZED) Outcome: The objectives of the study have been achieved. The report is well written and is informative in an area which has a great deal of policy relevance due to the interest in sustainable communities. The data has been taken from a number of varied unofficial sources as required for such a study. However, the data is considered to be reliable and as well documented as might be expected for such a study. The assessment method used (BRE’s Ecopoints) is considered to be somewhat of a ‘black box’ although there has been an effort to reduce this weakness. The standard set by the BedZED scheme is sufficiently widely recognised by policy making organisations that the study has been considered as endorsed.

Boundaries – The boundary of the analysis was not discussed although a sensible boundary has been consistently followed.

Data – The data has been taken from a number of varied unofficial sources as required for such a study. However, the data is considered to be reliable and as well documented as might be expected for such a study.

Method – The method used (BRE’s Ecopoints) is a recognised approach to use for such applications and has been applied transparently and robustly. The weighting applied to generate a single score for very different impacts is presented. Overall, any such approach is going to be somewhat of a ‘black box’ requiring the reader to trust in the model and who are applying it. Consistency – Overall, most of the results could be related to SIC codes

Results – The result are clear and comparable to similar data sets.

Policy Agendas: Policy relevance - The results of the study are most relevant to policy makers at a national level and they feed directly into meeting SCP objectives (resource productivity) and are central to the achievement of sustainable communities which is a very real and important policy challenge. The results are also considered relevant enough to provide a benchmark and feed into the government’s sustainable development strategy.

Policy recommendations – The recommendations are well written and presented, although targeted at house builders rather than policy makers.

Outcomes from recommendations - The study led to the formation of a trading company: BioRegional Reclaimed. The company specialises in reclaimed building materials for construction projects. It sources and supplies reclaimed materials to order.

SCP001 Page 132 of 196 B/1776 Magazine Publishing

Organisation: University of Leeds Environmental Trust

Stated objectives: The project will perform a mass flow analysis of the periodicals publishing sector in the UK. Mass flow analysis quantifies all physical flows of materials through an economic sector, from extraction of raw materials through production and processing, consumption, recycling, and final release of materials to the environment.

Outcome: The objectives of the study have been achieved either through this study or through reference to other Biffaward-funded projects. This is a well-researched study which clearly had the support of the sector involved. The weakness in the methodology, the use of sample sites, was sufficiently managed and the risk of bias was discussed and the data partly validated through checking flows through the supply and waste chain. The recommendations are well though out and relevant to both the sector and policy makers. One of the recommendations feeds directly into local authority waste management planning.

Boundaries – The boundaries are clearly described and consistently followed.

Data – The risk of bias in the selection of samples was discussed and the data partly validated through checking flows through the supply and waste chain. Unlike other sample questionnaire based studies, it was considered that the relationship the researchers had with the surveyed sites means that the results could be updated with the researched support. Also, as the data was validated and checked elsewhere in the mass balance model, the results are not considered to be modelled.

Method – The methods used are transparent and robust. Consistency –Data was consistent with SIC classifications as far as was possible.

Results – The result are clear and fit into and complement other Biffaward-funded studies (i.e. newspaper and paper). However, as water and gas use was not fully captured, it is considered that considerable amount of time would be required to finish of the data set.

Policy Agendas: Policy relevance - The results of the study are relevant to policy makers at a local level and feed directly into meeting local SCP objectives as they highlight the low rate of magazine recycling. This is due to the misconception that the high levels of ink in magazines means that they cannot be recycled through domestic recycling routes. The study would also feed into the sectoral analysis at a national level and the sector's sustainability strategy.

Policy recommendations – The recommendations are well though out and relevant to both the sector policy makers. The scope of the recommendations is wide and has an appreciation of the complexities of the issues considered. Perhaps an exception is the recommendation relating to the use of LPG in delivery vehicles. It would appear that more thorough analysis might be needed before this is rejected as uneconomic if this has not already been done.

SCP001 Page 133 of 196 B/1635 National Health Service - Material Health

Organisation: Royal Society of Wildlife Trusts

Stated objectives: To undertake a baseline environmental impact study of the NHS using materials flow analysis and Ecological Footprinting methodologies. This will identify opportunities for materials efficiency and waste minimisation. It will also provide a sustainability indicator and be used as an environmental communications and performance monitoring tool

Outcome: The objectives of the study have been achieved. The study has been undertaken using a similar approach to many of the later regional studies, mass balance and ecological footprint analysis with scenarios. The overall approach is found to be sound and fit for its purpose. The data is considered to be reliable and well documented coming from mostly official NHS sources. Some modelling was adopted to estimate the mass of products consumed. However, as spending data was only used to apportion the product data not to derive it, it is not therefore considered that input data was modelled. The reduction scenarios were very well constructed and would represent challenging but achievable targets for the NHS. The report does not make any specific policy recommendations although its content highlights the very low level of recycling in the NHS. The study would provide feed-in directly to ‘information provision, economic instruments’ in relation to consumption SCP objectives, and national waste management planning at a national level. The report would also be very useful for any government department undertaking ‘sectoral analysis/ prioritisation’.

Boundaries – The boundaries of the analysis are properly discussed.

Data - The data is considered to be reliable and well documented coming from mostly official NHS sources. Some modelling was adopted to estimate the mass of products consumed. However, as spending data was only used to apportion the product data not to derive it, it is not therefore considered that input data was modelled. Method – The study has been undertaken using a mass balance and ecological footprint analysis with scenarios. The overall approach is found to be sound and fit for purpose.

Consistency – The data is consistent with SIC classifications where possible via the use of ProdCom data for the input of products. The results of the study are also consistent with and comparable to a large number of other consumption- based ecological footprint studies undertaken in the previous few years (methodologically this may not be the case for the earlier studies).

Results – The result are clear and consistent with and comparable to a large number of other consumption-based ecological footprint studies undertaken in the previous few years (methodologically this may not be the case for the earlier studies).

Policy Agendas: Policy relevance - The study would provide feed-in directly to ‘information provision, economic instruments’ in relation to consumption SCP objectives, and national waste management planning

SCP001 Page 134 of 196 at a national level. The report would also be very useful for any government department undertaking ‘sectoral analysis/ prioritisation’.

Policy recommendations –The report does not make any specific policy recommendations although its content highlights the very low level of recycling in the NHS. The reduction scenarios were very well constructed and would represent challenging but achievable targets for the NHS.

Outcomes from recommendations - ƒ The Royal College of Nursing Congress adopted a motion to reduce the amount of waste produced by the NHS quoting data from Material Health in support of this 37

ƒ Report has made it into University modules on Sustainability & health 38

ƒ Presented at Architects for Health event 39

ƒ Transport data from MH used as evidence base in new NHS Health Development Agency document on transport and health 40

ƒ Study has made it into Audit Commission document on Quality of Life Indicators 41

ƒ Referenced as evidence base in NHS PASA sustainability report 42

Endorsements – Extract from report Foreword, Sir Nigel Crisp, Chief Executive of the Department of Health and NHS & Sir Liam Donaldson, Chief Medical Officer, Department of Health. Extract from Foreword to Material Health:

“This study provides the first complete overview of the impact of NHS activities in England and Wales on the environment and will be a valuable tool in the debate on the future direction of the NHS’ economic, social and environmental policies.”

“We recognise that the NHS, as a large organisation, will have a powerful role in helping to deliver a more sustainable future …In this study, we have for the first time an overall picture of the resources consumed in the delivery of healthcare in England and Wales. Quantifying the environmental impact (‘ecological footprint’) of the NHS allows us to see where improvements could be made, in terms of promoting more sustainable patterns of consumption and reducing the quantity of waste that is generated, whilst still improving the social and economic circumstances of the country’s population. …”

37 http://www.rcn.org.uk/news/congress2005/display.php?ID=1481 38 www.nottingham.ac.uk/sbe/ planbiblios/bibs/Greenis/A/23.html 39 http://www.architectsforhealth.com/library/event-oct2004-a.html 40 www.renewal.net/Documents/ RNET/Research/Makingcaseimproving.pdf 41 www.audit-commission.gov.uk/Products/NATIONAL-REPORT/ 0D488A03-8C16-46fb-A454- 7936FB5D5589/QofL2005.pdf 42 http://www.pasa.nhs.uk/sustainabledevelopment/2004/performance/ataglance/environmental.htm

SCP001 Page 135 of 196 B/1837 Newspapers UK

Organisation: National Society for Clean Air and Environmental Protection

Stated objectives: The project will effect interlocking mass balances of material flows across the UK newspaper publishing sector on an annual basis and disaggregated by geographical region. The mass flows will cover all activities from the production of raw materials, through production, processing, consumption, recycling, and final release of materials to the environment.

Outcome: The objectives of the study have been achieved. This is a well-researched study which has been objectively undertaken. The methods used highlighted the weaknesses in the data available. However, the results of the study can be considered as the best available in this area. The underlying approach to generating much of the data is a model as there are only a few real mass measurements in the results. This has led to the mass balance or the loop not meeting as has been discussed in the report. The sources of the data are varied, although largely reliable and extensively documented. The methods used are transparent and robust and are very relevant to what is being studied. The general findings and conclusions of the report have been taken to be the studies recommendations. These are very relevant and useful to local authority waste management planning. The subject area is very policy relevant to waste management SCP objectives at a local national level. There is also useful information within the report for examples of industrial symbiosis.

Boundaries – The boundaries are clearly described and consistently followed. The inconsistency in the input data was also transparently discussed.

Data – The underlying approach to generating much of the data is a model as there are only a few real mass measurements in the results. This has led to the mass balance or the loop not meeting as has been discussed in the report. The sources of the data are varied, although largely reliable and extensively documented. Method – The methods used are transparent and robust and are very relevant to what is being studied.

Consistency –The results are explicitly consistent with SIC classifications. Non direct or life cycle flows were partially considered.

Results – The result are clear and fit into and complement other Biffaward-funded studies (i.e. magazine and paper).

Policy Agendas:

Policy relevance - The subject area is very policy relevant to waste management SCP objectives at a local national level. There is also useful information within the report for examples of industrial symbiosis.

SCP001 Page 136 of 196 Policy recommendations – The general findings and conclusions of the report have been taken to be the studies recommendations. These are very relevant and useful to local authority waste management planning.

B/1271 Packaging

Organisation: University of Leeds Environmental Trust

Stated objectives: The project is intended to provide baseline data on packaging and packaging waste flows against which targets and progress toward targets can be monitored. The data will encompass the flows of primary, secondary and tertiary packaging waste from the household, commercial and industrial sectors. The project will consolidate existing data sets and where necessary initiate studies to collect additional data. Outcome: It is not easy to determine from the report to what extent the objectives have been achieved. The first phase report is a very long (>200 pages) and detailed account of the issues and data sources available, as well as what a mass balance of packaging in the UK would involve. There isn’t however any central place where the scope of this first stage is set out. The report would be greatly improved with a more focused summary. It will be a real challenge to generate a mass balance of packaging in the UK and the report successfully identifies the gaps in data, as well as provides a robust mass balance framework. The report concludes that the present availability of packaging data is not fit for implementing the various directives and regulations. This has been assessed as directly relevant to meeting national waste management objectives. Boundaries – The boundaries are defined.

Data – It will be a real challenge to generate a mass balance of packaging in the UK and the report successfully identifies the gaps in data, as well as provides a robust mass balance framework. Results & consistency – The report makes every attempt to be consistent with SIC classifications although would likely diverge is some parts of the mass balance. Non-packaging or processing material requirements are not considered within the study.

Policy Agendas: Policy recommendations – The report identifies a large number of the gaps in data and concludes that the present availability of packaging data is not fit for implementing the various directives and regulations.

Policy relevance – The report provides a very useful model for anyone attempting to construct a mass balance of the packaging in the UK However, the recommendation that the present data is not fit for purpose is taking as the most relevant policy outcome and is directly relevant to meeting national waste management objectives.

Outcomes from recommendations -

SCP001 Page 137 of 196 B/1412 Paper and Board Industry

Organisation: South West England Environmental Trust

Stated objectives: To provide information on resource flow of raw material and processing energy in the paper industry- mass balance study. Outcome: The objectives of the study have been achieved. This represents a well-defined and perhaps one of the best presented reports in the overall programme. The shortcomings in available data have been transparently discussed. The most obvious shortcoming in the report is the absence of recommendations, and therefore its relevance to policy.

Boundaries – The boundaries are clearly defined, throughout and consistently followed. Data – There are a number of acknowledged weaknesses in the data sets as used. The main data sources (from the Paper Federation’s 2000 Environmental Enquiry) only represented ~1/3 of the total industry and therefore the data had been extrapolated from this. This has been accepted as a broadly acceptable basis for data although the data needs to be considered as partly-modelled. The other (IPPC applications) are also acceptable even though the data has not been validated. The use of 1997 energy data is however the key weakness of the data. Overall most of the data represents one-off data sets and therefore is not updateable.

Method – The methods (mass balance) used are transparent and robust and accepted.

Consistency –The results are intended to be consistent with SIC classifications. Only some indirect flows are (i.e. electricity) and no hidden flows are included. Therefore the results cannot be considered to be consistent with Eurostat guidelines.

Results – The results are clear and largely due to their relative simplicity, could be compared to similar data sets.

Policy Agendas: Policy recommendations – There are no explicit recommendations in the report. However, the results are presented and explained very effectively. One insight from the report is the implications of increasingly high levels of paper recycling and the corresponding increase in waste produced by the sector due to lower grade inputs.

Policy relevance – Leading on from the above, the results of the study provide information to policy makers responsible for planning waste management at a national level. The results would also feed into any process of the sector developing a sustainable development strategy. Outcomes from recommendations -

SCP001 Page 138 of 196 B/2034 Pig

Organisation: RACEnvironment Ltd

Stated objectives: The objective of the project is to track and measure the flow of resources in a sector of the food production industry, in order to identify a strategy for the reduction of its impact on the environment. The outputs from the project will be used to inform, educate and increase awareness throughout the full range of stakeholders.

B/1639 Poultry Industry

Organisation: RACEnvironment Ltd

Stated objectives: The project process is to compile a resource flow analysis of the UK poultry industry, from production, through processing and retail, to the consumer. It will provide a holistic overview of poultry meat production and an ecological footprint analysis for one sector of the food industry.

SCP001 Page 139 of 196 B/1524 Public Sector

Organisation: Waste Watch

Stated objectives: We consider an environmental footprint to include the identification and quantification of environmental impacts, such as resources consumed, waste generated, energy consumed and atmospheric emissions. For a range of public sector bodies, representatives of central government, local authorities and executive agencies, we will document and normalise this data, by functional work practices and report this within an inventory table. Outcome: The objectives of the study have only partly been achieved as the footprint has not been calculated as only direct consumption has been assessed. The study is policy- relevant and recommendations are well written and presented. However, the approach required to achieve the results (scaling up of sample questionnaires) represents a significant limitation in how the study can be used to inform policy. It is recognised that there would have been few alternative options open to the study team when generating the dataset in this sector. It is this issue alone which has led to the red assessment being awarded.

Boundaries – The boundaries are only discussed in Annex B. This is considered to be a weakness as the boundaries of the public sector would not be obvious to many readers. Overall however, the boundary meets the criteria for a good assessment. The exclusion of devolved administrations was accepted as a minor weakness. Data and Method – As acknowledged in the report, the data and the method of its construction is very week. The sample selection was done based on the likely level of the corporation, rather then how representative the case study might be. This makes the scaled up results unreliable. For example, the input into the public sector of more than one computer and nearly one phone per employee (reported to be 1.1 and 0.96 respectively) in any one year is implausible, although of course possible. Consistency –Data is consistent with SIC classifications and the results might have been comparable to other similar studies which only consider direct flows.

Results – The result are clear and comparable to similar data sets.

Policy Agendas:

Policy relevance - The results of the study would have been relevant to policy makers feeding green public procurement SCP objectives.

Policy recommendations – The recommendations are well written and presented and are relevant and useable as they do not directly rely on the studies data.

SCP001 Page 140 of 196 B/1224 – Schools

Organisation: Southampton Environment Centre

Stated objectives: To assess: the environmental impact of the schools sector arising from use of material and energy resources; the opportunities / limits to reductions motivated by cost savings alone; barriers to further improvements; policy suggestions to overcome barriers.

Outcome: The report successfully manages to cover the scope of the objectives whilst working with a fragmented and heterogeneous sector that holds few records. The main weakness in the study relates to the unclear boundary of the analysis as well as the approach to data collection (as discussed below in the section on data). The report did however provide an extensive and informative briefing for LEA’s and headmasters on how to manage energy and resource flows in their school.

Boundaries – The boundaries of the study have not been made clear and there is concern that the boundaries of the factors used may not be consistent. For example, it was not clear whether ‘transport’ related to transport of pupils to school or not. It was also inconsistent about when material used in producing products offsite was being included. Data – The approach taken (case study schools in a certain area) would appear to be the only practicable way of collecting data within this sector. However, there were a large number of mostly ad-hoc assumptions made to construct the data sets. For example it was (implicitly) assumed that Grant Maintained (GM) schools were representative of the entire sector. There was insufficient analysis of the implications of this and how the sample data relates to the sector more generally. Method – The methods used as presently documented contained flaws.

Consistency –The results are consistent with SIC classifications but it has not been possible to determine with any confidence whether the results follow the Eurostat guidelines. Results – The results are clear but only partially so, comparable to similar data sets (such as B/2068).

Policy relevance – The results of the study have a number of applications in terms of providing a benchmark for other institutions. Furthermore the data would appear to be of a suitably high enough quality to provide information for public procurement policies and sectoral analysis. However, the weaknesses as identified would prevent the results being used directly to inform policy without further work. A thorough investigative and comparative analysis would need to be undertaken on the calculations of this and the Education sector study B/2068 for either to be used to inform the formulation of policies.

SCP001 Page 141 of 196 B/1195 Tourism (Benchmarking Tool) Organisation: WWF-UK

Stated objectives: To develop a web-based benchmarking and educational tool which will introduce environmental management to hotels and identify effective strategies enabling hotels to achieve waste reductions of up to 50% within five years

B/1352 Tyres Organisation: Viridis

Stated objectives: This application is Stage 1 of a larger multi-stage project covering database, statistics, current practice, awareness campaign, improved technological processes and new applications for recovery, including demonstration and pilot projects.

Outcome: The combined work (one report for the EU and one for UK) is a reasonably comprehensive mass balance of tyres from their manufacture, use and reuse to final disposal. Although there are a number of data issues, these were handled transparently and the results therefore represent the best available picture for the UK. The presentation of the results is very useful and shows the clear policy focus of the project. Boundaries – The boundaries were clearly defined

Data – It was not clear what year the results were supposed to represent. It was taken to be 2000, the year much, but not all, of the data was taken. The approach calculating the input of mass, multiplying the number of tyres by factors and therefore a modelled data set, was not made explicit enough.

Method – The methods used (mass balance) are transparent and robust. Consistency –The data is consistent with Eurostat and with SIC classifications.

Results – The result are clear and comparable to similar data sets.

Policy relevance – The results of the study are very accessible and feed directly into waste SCP objectives, specifically waste management planning at a national level. The results of the EU and UK studies provide contextual information for a number of more general polices including understanding the emerging policy impacts of EU waste directives.

Outcomes form report – This was the first product mass balance completed within the Biffaward programme. It was commissioned in anticipation of the ban of whole and shredded tyres “Civil engineering applications of tyres”.

SCP001 Page 142 of 196 Waste management studies

B/1716 Best Integrated Transport Options for Scottish Waste Organisation: Viridis

Stated objectives: To provide a tool for planners to ensure that waste management facilities can be located to minimise waste transport journey distances and emissions. The project will cover waste movements throughout the whole of Scotland.

B/1850 EuroCharge Organisation: CIWM (EB)

Stated objectives: The project proposes to carry out an examination of the current practices of direct charging of householders for waste management services across Europe. The research will include a summary of the legal position in each EU country, a breakdown by municipality of charging policies and methodology, measurements of social impacts and up to ten detailed case studies. Outcome: The assessment was undertaken on the shorter 39-page report but accepts the larger 258 page report as a full documentation of the approach. The report meets its stated objectives. The results are relevant to but do not attempt to contribute towards material accounts. It has therefore been a probabilistic to assess the study.

Boundaries – The boundaries of the analysis were not defined. Data – The data was case study and survey (qualitative) based. The year was taken to be the year of publication (2003). The data was well documented, and therefore available, and largely based on real observations. Method – The analysis relied upon a significant number of assumptions. The results however represent the best available analysis.

Consistency – Data is broadly consistent with SIC classifications but not with Eurostat guidelines.

Results – The results are clear and comparable to similar data sets in so far as they are based on European case studies.

Policy relevance – The recommendations are very relevant but could be written in a more neutral way. They feed directly into SCP waste management planning at a national and local level and inform the modelling of policy impacts. The subject of this study remains very policy relevant.

SCP001 Page 143 of 196 B/1413 National Assessment of Civic Amenity Sites

Organisation: Future West

Stated objectives: To research the extent to which non-household waste is entering Civic Amenity Sites, aiming to identify mechanisms which encourage/enforce the proper disposal of waste. It will also investigate other reasons for the recent increase in waste entering Civic Amenity Sites.

Outcome: The objectives of the study have been fully achieved. The study faced the very considerable challenge of estimating the waste from a group of users (traders) who take active steps to avoid classification and without being able to weight any of the waste being disposed of. The data relies upon the often subjective assessment of surveyors on site. Extensive design measures were taken to ensure that the data represented the best available estimate of the proportion of waste coming from trade sources. The key weakness of the study is the lack of winter survey data which means that the study was unable to scale the results up for the year. It is not thought possible to update this survey now. The recommendations of the study are well written and set out and provide a useful analysis of why traders are using civic sites and the best ways of tackling this. The most innovative solution seems to be to allow traders to use civic sites and only pay for non-recyclable waste disposal. The results of the study are very relevant to waste management planning at a local authority, and to a lesser extent, at a national level.

Boundaries – The boundaries are clearly set around waste being disposed of at the 10 or 11 civic amenity sites in the Bristol area.

Data & method – The data relies upon the often subjective assessment of surveyors’. Extensive design measures were taken to ensure that the data represented the best available estimate of the proportion of waste coming from trade sources. After detailed analysis of variance, it was found that so-called objective criteria for assessing whether users were traders were in fact more subjective than surveyors’ own so-called subjective assessment. The study then relied upon the surveyor’s observation of the waste being disposed of by type and subsequent estimation of its mass to estimate cost of the illegitimate use of these sites by traders. The resulting approach and data represents a best available. The study was unable to scale up the results into an annual result as the survey was only undertaken in the summer.

Consistency – The data is only very broadly consistent with SIC classifications

Results – The results are clear but only comparable to data sets from civic amenity studies. The key weakness of the study (the lack of winter survey data) is not solvable as a repeat survey would not provide comparable results as too much time has elapsed.

Policy Agendas: Policy relevance - The results of the study are very relevant to waste management planning at a local authority, and to a lesser extent at a national level. It has not been possible to identify policy relevance beyond waste management policy.

SCP001 Page 144 of 196 Policy recommendations – The recommendations are good and well set out and provide a useful analysis of why traders are using civic sites (lack of alternatives which avoid paying the landfill tax) and the best ways of tackling this. The most innovative solution seems to be to allow traders to use civic sites and only pay for non-recyclable waste disposal. This would allow small traders to contribute towards the original objective of the landfill tax (i.e. reducing the amount of waste going to landfill).

SCP001 Page 145 of 196 B/1826 - Thermal Methods of Municipal Waste Treatment

Organisation: C-Tech Environmental Group

Stated objectives: The aim of the project is to critically evaluate a range of techniques from published data on gasification, pyrolysis, plasma gasification, microwave pyrolysis, hydrolysis and encapsulation, to provide a reference document for the waste management industry, government and other stakeholders.

Outcome: These objectives were achieved and therefore represent a useful reference document. The outcomes contribute towards constructing material balance accounts but do not do this directly.

Boundaries – The boundaries of the emission factors generated are defined.

Data – Method – The data provided is largely based on data from manufacturers of the equipment, and therefore there is a potential interest in the outcomes.

Consistency – Does not contribute towards constructing material balance accounts

Results – The results are clear and equivalent to existing data. Policy relevance – The results of the study are relevant to policy makers at a national level and provide information for waste management planning as well as a number of more general policies .

Outcomes - Numerous developments and processes are under consultation by councils and local authorities to use technologies reviewed in the report (Allington, Bexleyheath, Dorking, Colnebrook, Edmonton, Marchwood, Portsmouth, Sheffield, Wrexham, etc). There is an increasing need to reassure the local communities of the viability and safety of these processes and the report seeks to address to this in simple layman's terms.

The County Environmental Director of Devon County Council considered the report a comprehensive well –researched and evidenced summary of thermal methods. Future Energy Solutions commented that the document was of particular interest.

1800 paper copies of the report (plus unquantified pdf downloads) have been distributed directly to relevant persons and at numerous events, bringing the data as well as policy and legislation information to the attention of Government Departments, Local Councils, trade associations, etc.

SCP001 Page 146 of 196 B/31 - Waste from electrical and electronic equipment

Organisation: Industry Council for Electronic Equipment Recycling Ltd (ICER)

Stated objectives: The project will identify the amounts and type of electronic and electrical equipment entering the waste stream, how much is currently being recycled and what needs to be done to achieve more recycling.

Outcome: The objectives of the study have been achieved, although ‘what needs to be done to achieve more recycling’ was not discussed explicitly. Overall this is a very good study and relevant to a major area of European legislation (the WEEE Directive). The data is somewhat out of date (1996 & 1998) for an area which is changing rapidly. The data capture is varied but is sound and all efforts have been made to ensure that the approaches are sound. The modelling approach estimates of arisings based on assumed disposal rates which will arise from new purchases, is a sound approach although the results it produces need to be considered as modelled rather than measured. The results provide a very useful baseline for implementing the WEEE Directive in the UK and are relevant to integrated product policy and waste management SCP objectives.

Boundaries – The boundaries of the study are set around the waste streams as set out in the WEEE Directive.

Data - The data is somewhat out of date (1996 & 1998) for an area which is changing rapidly (i.e. the rates of recovery are increasing due to regulatory response, consumption is increasing and the material content of WEE is changing fast). The data capture is varied but is sound and all efforts have been made to ensure that the approaches are sound. Method – The modelling approach provides an estimate of WEEE arisings based on purchases and replacement rates and is a sound approach which makes use of the reliable data. The results it produces need to be considered as modelled rather than measured as the assumptions made, although reasonable, cannot be considered to be limited.

Consistency – Data is only broadly consistent with SIC classifications.

Results – The results are clear and comparable to similar data sets relating to WEEE arisings in Europe.

Policy Agendas:

Policy relevance - The results of the study are relevant to policy makers at a national and EU level. The results provide a very useful baseline for implementing the WEEE Directive in the UK and are relevant to integrated product policy and waste management SCP objectives. The report would provide information for several sectors’ sustainable development strategies. Policy recommendations – The report does not provide very many explicit recommendations although the results identify those categories of equipment which were not being recycled at high enough rates to comply with the WEEE directive.

SCP001 Page 147 of 196 B/1425 REMAT - Meeting Producer Responsibility Obligations

Organisation: University of Surrey Environmental Body Stated objectives: The overall aim of this project is to determine and define the collective user requirements for corporate environmental reporting and to devise environmental reporting standards and templates to meet producer responsibility obligations (PRO) and which will be adaptive to future integrated product policy (IPP) developments. Further, the project will examine how these could be integrated into business practice and implemented within an e-commerce/ e- management framework for easy implementation and use. Outcome: The objectives of the study have been achieved. The output, REMAT, is a computer- based model developed after extensive consultations with stakeholders. REMAT is intended to assist business and regulators implement the various producer responsibility Directives (ELV, WEEE and RoHS) by providing a data integration model. The absence of quantitative data to be assessed led to some difficulty during the assessment which has been resolved in the following way: the qualitative design specification data collected during the second and third stages of consultation have been taken to be the study’s data. REMAT has been assessed as the study’s method recommendations. The results as would be generated by REMAT are taken to be the results. The consultations were undertaken with careful forethought and the data from them fed directly into the construction of REMAT. REMAT is a large and complex model which has undergone extensive planning and consultation to ensure it is fit for purpose. REMAT has been assessed as methodologically sound. REMAT is highly policy relevant to policy makers at a national and EU level and would feed directly into meeting integrated product policy and waste management SCP objectives. Boundaries – The boundaries of the study are provided as the producer responsibility for material flows in relation to the End of Life Vehicle (ELV), WEEE and Reduction of Hazardous Substances (RoHS) Directives. Data – The qualitative data collected during the second and third stages of consultation have been taken to be the study’s data. The first stage of consultation has been excluded as it involved confidential talks with the study’s steering group. Inclusion of this round would have led to a red overall assessment being provided due to the lack of availability of the data, an outcome which has been assessed as inappropriate. The consultations were undertaken with careful forethought and the data from them fed directly into the construction of REMAT. It is considered that a further round of consultations could take place to update the model at a later date. It is considered that the data is updateable.

Method – REMAT is a large and complex model which has undergone extensive planning and consultation to ensure it is fit for purpose. It is not been possible to validate the REMAT in detail from the report, although it has been possible to confirm that the approach taken is methodologically sound.

SCP001 Page 148 of 196 Consistency –Data is broadly consistent with SIC classifications as it relates to the recycling of waste.

Results – The results would be generated in a useful format and could be compared to similar data sets relating to producer responsibility.

Policy Agendas: Policy relevance – REMAT is highly relevant to policy makers at a national and EU level and would feed directly into meeting integrated product policy and waste management SCP objectives. The results could also feed into modelling policy impacts.

Policy recommendations – REMAT is taken to be the study’s recommendation.

Outcomes from recommendations -

SCP001 Page 149 of 196 B/1783 Waste Data Flow

Organisation: CIWM (EB)

Stated objectives: To carry out a 3-stage research and development project to develop a national information intranet of waste flows within the public sector, ultimately replacing the annual manual surveys undertaken by various Government departments and agencies

Outcome: The study as presented on the CD provided by Biffa ‘Waste DataFlow: An introduction for local authorities – From Numbers to Knowledge’ successfully describes (in an audio visual format) how the proposed system will map all municipal waste data flows in the UK. The proposed outcome will represent a substantial improvement in the recording of waste flow data and will feed directly into meeting the requirements of European Union Directives including the Landfill Directive.

Boundaries – The boundaries are clearly defined as the UK and municipal waste flows Data – The proposed data requirements will be almost entirely ‘real’ data based on observations of mass. It has not been possible to explore the accuracy of the waste stream classification of existing data sets but it is likely that this process will both standardise and improve waste classification. Method – The methods are transparent and robust.

Consistency –Data consistent with SIC classifications as it is only concerned with the waste at the point of disposal so indirect material flows are not relevant. Results – The results will be broadly comparable to similar data sets and comply with a standard classification across the EU.

Policy relevance – The project has been developed in collaboration with (and by implication endorsed by) a range of government stakeholders. The proposed outcome will represent a substantial improvement in the recording of waste flow data (and therefore waste management planning) at local authority level and will also feed directly into meeting the requirements of European Union Directives including the Landfill Directive at a UK level. In addition, the outcome will provide information for future national sustainable development strategies.

SCP001 Page 150 of 196

SCP001 Page 151 of 196 III Development of an Indicator for Emissions and Impacts associated with the Consumption of Imported Goods and Services (Section 6)

III.1 Specific assessment requirements

Criteria similar to those used in the assessment of Resource Flow studies (see Section 6) were used with some specific adaptations as defined in the table below.

Assessment Criterion Good Satisfactory Unsatisfactory of… Clearly defined Boundaries of system boundaries Defined and Arbitrary or not national Clearly defined matching standard consistently satisfactorily economy and or established followed defined international methods trade Boundaries allow Boundaries do not Comparison Comparisons are comparisons with allow comparisons with other only partly Comparisons results from with results from studies with possible or require different studies or different studies or same recalculations methods methods boundaries Approach covers all indirect Approach partly Approach only

Boundaries Coverage of emissions/impacts Coverage of covers indirect covers direct indirect impacts caused by analysis emissions/impacts emissions/impacts upstream production

INFO: study considers trade in …

INFO: number of world regions / countries

Year(s) 2001+ 1999-2000 prior to 1999 Year analysed The method allows The study has The method does for the calculation produced a >4 not allow for the of long-term trends Time series years trend of calculation of time Time span without major embedded series without adaptations or emissions major adaptations limitations Designed to be Applied datasets One off data sets,

Data Updatable Datasets applied updated are updatable not updatable Uses official government Concern about Only uses official statistics and/or data source and Reliable government unofficial data of Datasets applied method of statistics good quality from construction trustworthy sources

SCP001 Page 152 of 196 Assessment Criterion Good Satisfactory Unsatisfactory of… As above for main Uses robust real components but Real data and/or data and weak on more estimation Real or estimation disaggregated procedures not Datasets applied modelled data procedures (fit for elements (fit for robust even for all declared main purposes but main components purposes) not all) Error The study estimation/calculat describes in No mention of Uncertainty of ion integral part of general terms the Description of uncertainty or modelled data methodology or uncertainty of the errors margin of error explicit sensitivity data that it testing produces

INFO: Number of economic sectors

INFO: data sources

Robustness Approach is Approach has Approach has (credible & credible and minor flaws or serious flaws or Method reliable) reliable limitations limitations The study has been The material The study's funding funded by public interests of the means that there is Objectivity money or there is study's funders are Funding source potential for no potential apparent in the material interests conflict of interest report

Documentation Well documented Not properly Description of of assumptions & understandable Documented documented method and limitations to any reader Some kind of peer

Method robustness review process; Not independently Full methodology Peer reviewed method reviewed and published in Publications / and width of accepted/adopted method not used by academic (peer citations understanding by at least one any other research reviewed) journal other research group group Uses mainly methods, Consistent with Not consistent with Consistent with frameworks or parts of the UN any part of the UN Method UN SEEA data consistent SEEA framework SEEA framework with UN SEEA Partly consistent With other with & referenced established Yes No Method to established methods/studies method/study

Employs Provides own internationally classification Does not relate to With standard standardised system, which Consistency standard Classification classifications classification relates to standard classification codes system (SIC, CN classification etc.) system Method is in Study describes Method is not Applicability to principle application to the applicable to the Application the UK applicable to the UK UK UK

SCP001 Page 153 of 196 Assessment Criterion Good Satisfactory Unsatisfactory of…

INFO: Indicators (environmental impacts embedded in trade)

Work is required to Results conflict Comparability Method designed make results with existing with other to provide comparable to and findings and/or Results

results comparable results usable for existing cannot be added to data sets existing data sets Results Method can be only with a standardised to Yes - without considerable provide fully major difficulties No Results amount of work comparable (<1 year work) (>1 year work) results

Table III.1 - Criteria used in the assessment of studies on embedded emissions

SCP001 Page 154 of 196 III.2 Review of existing approaches described in the literature

III.2.1 Studies not involving input-output calculations It is possible to estimate the direct emissions associated with the production of imports to the UK without applying input-output (IO) analysis. However, such an approach only covers the direct sectoral emissions at the source level, unless estimations of indirect emissions are done separately. In a single-region model, direct emission factors identical to the national production are assumed for all imports, whereas in the multi-region case these can be distinguished depending on the imports’ origin. The principal data requirements for this type of estimation are: • energy use matrix: fuel types per sector, incl. households,

• CO2 emission factors: amount of CO2 emitted per unit of fuel used,

• or, if available: direct CO2 emissions matrix: per sector, incl. households, • total intermediate and final imports (monetary or physical data) per sector in the single-region case, • shares of imports per sector (based on monetary or physical data) in the multi-region case. In the following, recent studies employing approaches without IO calculations are reviewed, together with the outcome of the assessment. Because of the limited scope of this study, only major and recent approaches have been assessed but other studies dealing with the problem of embedded environmental impacts are cited as well.

Muradian et al. (2002) use the World Bank’s Industrial Pollution Projection System database of pollution intensities in order to estimate trade balances of embedded emissions of Japan, the USA and Western Europe. These intensities are multiplied with UN trade statistics, so that indirect effects are not accounted for. As expected, the ‘environmental terms of trade’ of the three regions with the developing world are positive, i.e. environmental load is displaced abroad. Using the same database and similar pollutants, Lee and Roland-Holst (1997) estimate embedded pollution in trade between Indonesia and Japan.

Reference assessed Muradian et al., 2002 (based on the World Bank’s IPPS database) Verdict Score 15, unsatisfactory Main strengths Presents time series and documents assumptions and limitations well. Main weaknesses Only looks at the eleven most polluting sectors; serious limitations and diminished applicability due to inconsistent data sources.

In a UK study (Harris, 2000) the CO2 contents in imports to the United Kingdom have been estimated by applying three alternative process-type approaches. While the first approach uses only UK energy data, approaches two and three include foreign energy use and emissions data.

SCP001 Page 155 of 196 The calculations are based on sectoral trade and fuel use statistics of the exporting countries. The three approaches produce substantially different results. In summary, the study has found that the use of energy consumption data with subsequent conversion into CO2 emissions does not provide a satisfactory basis for the assessment of air emissions arising from the production of imports to the UK. Rather, the development of consistent environmental accounts data for a range of key trading partners is seen as the way forward, with finer detail on the production, consumption and emissions of particular industries, for a range of years. In this project an assessment was undertaken of the third method (Harris, 2000) where other countries’ environmental accounts were used, in conjunction with trade statistics, to calculate the share of atmospheric emissions attributable to the production of imports to the UK.

Reference assessed Harris, 2000 (Method 3: using other countries' environmental accounts) Verdict Score 21, unsatisfactory Main strengths reliable data, UK application Main weaknesses indirect emissions only partly covered (via use of electricity), restricted data availability

An example for a study dealing with a greenhouse gas other than CO2 is Subak (1995). Subak investigates methane (CH4) embodiments in the most CH4-intensive agricultural goods – rice, meat and dairy products – traded from developing countries to the USA, UK, Germany, Japan, France and Canada, and thus the possibility of CH4 leakage. However, Subak only considers emissions occurring on-site in the respective agricultural industries. Thus, she arrives at a lower limit of 1200 kt CH4 embedded in agricultural products supplied to the above six countries, representing between 1-2% (Canada, USA) and 14-23% (Germany, Japan, UK) of national CH4 emissions. These leakage figures are likely to increase if subsidies to agricultural sectors in industrialised countries are removed, and exports of agricultural products by developing countries increase.

Hence, future CH4 leakage may be of a scale sufficient to undermine regional abatement goals. Subak (1995) states that “an estimation method based on attributing all embedded emissions to the country of import opens up a range of new technical difficulties”. In theory, a database called ‘Energy Analysis Programme (EAP)’ developed by IVEM, Groningen could be used to derived total energy multipliers of traded goods which can subsequently be turned into embedded CO2 emissions. The Dutch EAP was originally developed by Benders and colleagues (2001) and Wilting and colleagues (1999). The EAP is based on a hybrid energy analysis, which is a combination of physical-chemical process analysis and economic input-output analysis. The EAP calculates the primary energy requirement in the lifecycle of a final product. The energy intensity is then calculated by dividing these energy requirements of consumer goods and services by the monetary unit of the selling price.

Carlsson-Kanyama et al. (2002, 2005) describe such an adaptation and apply EAP data for an estimation of indirect and direct energy requirements of city households in Sweden. However, no embedded energy was calculated for traded goods and services and therefore the study was not assessed in this project.

SCP001 Page 156 of 196 In order to adapt the Dutch EAP data to the UK the following additional data would be needed (see Carlsson-Kanyama and colleagues (2002): • national IO tables, • specific life cycle inventory data for energy inputs for materials, if available, • energy demand for transportation (if transport is to be investigated), • ERE (energy requirements for energy) values of energy carriers43. For a model that picks up differences in embedded emissions in imports, EAP values for at least the major trading partners of the UK would have to be calculated. Another drawback with the EAP is that it treats all goods, even imported goods, as though they were produced domestically.

Reference assessed Carlsson-Kanyama et al., 2002 (use of EAP, based on Benders et al., 2001 and Wilting et al., 1999) Verdict (was not assessed, as study does not calculate embedded emissions) Main strengths - Main weaknesses The major drawback of the EAP approach is that it will never be possible to adapt it completely to economies other than the Netherlands because essential life cycle inventory data are likely not to be available for all countries and industries required.

A recent study by Shui and Harriss (2006), examining embedded CO2 emissions in trade between

China and the US found that about 7% to 14% of China’s current CO2 emissions were a result of producing exports for US consumers and that during 1997–2003 US CO2 emissions would have increased from 3% to 6% if the goods imported from China had been produced in the US.

The method employed in this study does not involve input-output calculations directly but uses emission factors from a database constructed with the help of input-output analysis. The CO2 embodiments in exports and imports have been estimated by multiplying the dollar value of each export (import) product by the corresponding CO2 emission factor for the same kind of product produced in the exporting (importing) country. Trade data were obtained from the US Census

Bureau. The CO2 emission factors for US exports to China are derived from the Economic Input Output-Life Cycle Assessment (EIO-LCA) software developed by Green Design Initiative at Carnegie Mellon University. The EIO-LCA was developed based on US economic input-output data. Due to the lack of a similar data for assessing environmental impacts of Chinese products, the CO2 emission factors for Chinese exports to the US are based on CO2 emission factors that have been corrected for differences in the fuel mix of the manufacturing sector in China and the US. Reference assessed Shui and Harriss, 2006 (based on EIO-LCA database) Verdict Score 12, unsatisfactory Main strengths presents time series

43 The energy intensities in EAP are calculated as primary energy. For this several ERE values of energy carriers are needed. The ERE values include the energy needed for drilling, extracting, and transporting the fuels to where they are used.

SCP001 Page 157 of 196 Main weaknesses traded commodities not defined; data sets not reliable for countries other than US; results not comparable; not applicable to the UK

As part of the Ecological Budget UK project, Wiedmann et al. (2005) describe a method to estimate CO2 emissions embedded in imports to the UK. In order to compute the CO2 emissions embedded in UK imports by industry and world region, value-based trade tables from HM Revenue & Customs44 were used together with the emission factors estimated by Ahmad and

Wyckoff (2003). The latter authors computed embedded (direct + indirect) CO2 emissions per US dollar (adjusted for purchasing power parity at 1995 levels) by some industries for a number of OECD and non-OECD countries. These countries were allocated to eight world regions.

Embedded CO2 emissions were then estimated per US dollar of imports adjusted for purchasing power parity (PPP) at 1995 price levels by these eight world regions for twelve industry groups. The basic data and sectoral layout were adapted from Ahmad and Wyckoff (2003).

The report finds that a total of 147 Mt of CO2 were embedded in UK imports in the year 2000. This is 24% of all domestic emissions in 2000 which were 606 Mt (ONS, 2004)45. The greatest contributors are the sectors ‘Other metal products and machinery equipment’ (32% of total embedded emissions), ‘Chemicals’ (18%) and ‘motor vehicles, trains, ships and planes’ (12%).

Reference assessed Wiedmann et al., 2005 (based on data from Ahmad and Wyckoff, 2003) Verdict Score 22, unsatisfactory Main strengths Applies multi-regional trade model; UK application Main weaknesses uses a one-off, not updatable data set

It should be mentioned here that Ecological Footprint (EF) analysis, as introduced by Wackernagel and Rees (1996), also provides a measure of some environmental impacts embedded in the trade of goods. The EF measures human demand on bioproductivity by assessing how much biologically productive land and sea area is necessary to maintain the consumption of a given human population. The calculation of Ecological Footprints starts from the consumption of resources in terms of mass units and transforms this mass into land appropriation in a second step (Monfreda et al., 2004). A considerable share of the Footprint consists of the hypothetical land area that is required to absorb carbon dioxide emitted from the combustion of fossil fuels.

National Footprint Accounts (NFA) are generated annually by the Global Footprint Network for most countries of the world (WWF, 2004; GFN, 2005). They account for the consumption of land by the countries’ residents wherever this land might be located. The Footprint associated with products imported from foreign countries, for example, is fully added to the consumers’ Footprint account. Therefore, the concept of EF analysis strictly follows the principle of consumer responsibility.

44 Statistics and Analysis of Trade Unit (http://www.uktradeinfo.com) 45 If it was assumed that imported goods were produced with the same emission intensity as products manufactured in the UK, then the result for total embodied emissions in imports would be 88 Mt or 15% of domestic emissions in 2000 (SEI, 2005, own calculations).

SCP001 Page 158 of 196 In its current state the method to generate National Footprint Accounts (Monfreda et al., 2004; Wackernagel et al., 2005) can only provide a rough estimate of land appropriation associated with the trading of goods. Using FAOSTAT data (FAO, 2005) on domestic production, imports, exports and yields for a number of primary and secondary products form agriculture, forestry and fisheries, the accounts estimate the apparent net consumption of a nation and the associated use of 46 47 land. The national energy Footprint is calculated via CO2 emissions data from IEA or CDIAC . For the trade balance of manufactured products, embedded energy data from disparate sources are used to convert their quantities into energy equivalents. These values are then assigned CO2 equivalents and subsequently energy Footprints.

Recent improvements of the NFA feature the exhaustive use of global trade data from UN Comtrade (2005) in SITC classification on a 4-digit level and improved embedded energy data for over 600 commodities (Wackernagel et al., 2005). While the method is practical for computing the apparent resource consumption of 150 countries in the world in long-term time series, there are still fundamental shortcomings in the methodology: • For domestically produced bio-products national conversion efficiency factors are used to calculate the Footprint, whereas average global conversion efficiency factors are used for imports. The Footprint of exported products from biological resources is weighted in proportion to the amount of products imported and produced domestically and their respective conversion factors. • Manufactured products have the same embedded energy regardless of the country of manufacture, i.e. the same energy intensities for imports and exports are used and they are the

same for each country. For the conversion into energy Footprints via embedded CO2 emissions, world average carbon dioxide intensity is used for all imports, whereas for exports of manufactured products the average carbon dioxide intensity of the exporting economy is used, reflecting the national fuel mix for energy production. • Imports and exports of services are not included in the NFA analysis. This means that any direct and indirect resource use and/or pollution embedded in trade flows of services are not accounted for. More generally, since only the total imports and exports from and to the rest of the world are listed for each country and thus no trade supply chains are identified, no distinction can be made as to where or how the imported products are produced. Hence, no account is taken for differences in production technology in trading partners, or, specifically, the direct and indirect Footprint intensity of trade flows of goods and services (with trade in the latter, and associated Footprints, neglected all together).

III.2.2 Studies involving input-output calculations

Introduction

46 International Energy Agency, Paris, France 47 Carbon Dioxide Information Analysis Center, Oak Ridge, Tennessee, USA

SCP001 Page 159 of 196 Environmental extended input-output analysis (Leontief and Ford, 1970; Victor, 1972; Miller and Blair, 1985) is a well established approach that allows resource flows and environmental impacts to be assigned to categories of final consumption. Some more recent examples for the use of environmental input-output analysis include analyses of international trade (Proops et al., 1999; Ahmad and Wyckoff, 2003; Meyer et al., 2003a,b; Ferguson et al., 2004; Peters and Hertwich, 2005, 2006a), estimation of land use changes in China (Hubacek and Sun, 2001) as well as pollution attribution and calculation of regionally specific fuel use (Turner, 2003) Further applications assess the environmental impacts of spending options (Lenzen and Dey, 2002) and explore the interdependence of industries in terms of environmental pressure and resource depletion (Lenzen, 2003). Material flow calculations at the national and international level (Moll et al., 2002; Hinterberger and Giljum, 2003; Giljum and Hubacek, 2004; Suh, 2004a) and Life Cycle Assessments (LCA) have also been combined with input-output analysis (Hendrickson et al., 1998; Joshi, 1999; Lenzen, 2002; Suh and Huppes, 2002; Suh, 2004b; Suh et al. 2004). The clear and main advantage of IO analysis is the fact that – once the calculation framework is established – it automatically accounts for the mutual interrelationships of economic activities and is therefore able to assign all indirect environmental impacts arising out of intermediate transactions (supply chains) in production processes. Single-region models cover national inter- industrial dependencies whereas multi-region models also include trade between different economies and thus allow following up supply chains even outside the national territory. The basic data requirements for IO based models are: • national (monetary) input-output tables, • energy use matrix: fuel types per sector, incl. households,

• CO2 emission factors: amount of CO2 emitted per unit of fuel used,

• or, if available, direct CO2 emissions matrix: per sector, incl. households • matrix of intermediate and final imports (in monetary units) per sector in the single-region case, • bilateral import coefficient matrices (based on monetary data) in the multi-region case.

The review of IO based approaches is divided into single-region, multi-region and simulation models. Again, only major and recent approaches have been reviewed and assessed.

Single-region input-output models An early study attempting to quantify “environmental loading” of traded products is Walter’s (1973) examination of the pollution content of American trade. Even though the author uses input- output coefficients to allocate environmental control cost to industries, the analysis falls short of all but first-round effects, since no matrix inversion is carried out. The first author to use the Leontief inverse in order to investigate embedded-factor trades was Fieleke (1975), who determined the US trade deficit in embedded energy. Shortly after, Bourque (1981) calculated the embedded-energy trade balances between Washington State and the rest of the United States. Due

SCP001 Page 160 of 196 to a lack of sectoral data in physical units, he uses “energy equivalents”, that is dollars of output and the number of employees in energy-producing industries.

A comparative input-output study for Germany and the UK is presented by Proops et al. (1993)48. The authors employ a single region IO model, assuming domestic production efficiencies for all imports, to calculate time series of CO2 emissions from 1968 to 1984 (UK). Three types of CO2 emissions for which a country is held responsible are being distinguished: domestic CO2 emissions to meet domestic final demand (“domestic”), foreign CO2 emissions to meet domestic final demand (“direct imports”) and foreign CO2 emissions to meet imported final demand (“indirect imports”). From 1968 to 1984 the domestic CO2 proportion fell from 87.4 to 81.4% whereas the proportions for directly and indirectly imported CO2 increased from 4.1% to 8.9% and 8.5% to 9.7%, respectively.

Gale (1995) investigates the effect of Mexico’s participation in the NAFTA agreement on CO2 emissions, by estimating changes in Mexican imports, exports and import-competing goods, and subsequently inserting adjusted figures into an augmented input-output model. Gale’s preliminary results show that even though tariff elimination gives rise to an overall 12% increase in Mexican

CO2 emissions, half of this increase is compensated by shifts in the production structure away from pollution-intensive sectors.

In a study of international trade flows, Wyckoff and Roop 1994 estimate the amount of CO2 emissions in imports of 21 different groups of manufactured goods to six of the largest OECD countries (Canada, France, Germany, Japan, the UK and the USA). The model is based on individual input-output tables for these countries and bi-lateral trade matrices (‘linked single- region model’). Their findings suggest that about 13% of these countries’ total CO2 emissions are embedded in imported manufactured products and they conclude that measures of greenhouse gas abatement policies will be less effective if they solely rely on domestic emissions.

Wyckoff and Roop (1994) also test their model in terms of variability towards sector aggregation.

The results of the 33 sector model are compared to a 6 sector model. This test reveals that the CO2 embedded in manufactured imports to Canada, France, Germany, Japan and the UK from the USA are about 30% less when calculated using a 6 sector model than when calculated using a 33 sector model.

Reference assessed Wyckoff and Roop, 1994 (six OECD countries, manufactured products) Verdict Score 25, satisfactory Main strengths Early study with robust methodology and data from official data sources. Could be repeated with more up-to-date data. Main weaknesses Data out of date; only manufactured products.

The effect of imports and foreign emissions on the life-cycle CO2 emissions of German production was examined by Wenzel (1999). This author carries out a life-cycle analysis of the CO2 requirements of passenger cars, computers and food items. His assessment involves the

48 For other applications see also Gay and Proops (1993) and Symons et al. (1994).

SCP001 Page 161 of 196 identification of countries of imports origins, transport distances and modes, and foreign energy production processes, in particular electricity. While transport emissions are modelled on the basis of process-chain data, embedded emissions are calculated using input-output analysis. Wenzel finds that, in spite of long distances, CO2 emissions from transport form a relatively minor part of total emissions (1-2% for cars and computers, and around 6% for food items). If, however, foreign energy production is explicitly taken into account, CO2 requirements change significantly (9% for cars and computers; food is not examined). Wenzel concludes that within effects of trade on CO2 emissions, and within reduction potentials, differences in production structure are more important than increased transport requirements.

Reference assessed Wenzel, 1999 (German products) Verdict Score 12, unsatisfactory Main strengths This is the only study analysed that provides results on the impacts of freight transport. It also takes into account different prices for energy carriers and differences in the efficiency of electricity production in other countries. Main weaknesses While the IO analysis follows the classical procedure, the approach requires a large amount of additional data from various sources for electricity production and freight transport which is not very practical and which raises some doubt about the reliability of the method.

The problem of ‘territorial’ or ‘attributable’ emissions (Proops et al. 1993) is also addressed by Kondo et al. (1996) and Munksgaard and Pedersen (2001), who demonstrate the differences between CO2 accounts assuming producer and consumer responsibility. The latter authors highlight the significant changes that Denmark’s CO2 trade balance underwent between 1989 and 1994. Eder and Narodoslawsky (1999) examine several criteria for inter-regional consumer and producer responsibility in their augmented input-output-based case study of a small Austrian region. Energy and/or CO2 emissions embedded in imports have also been estimated by Common and Salma (1992), Schaeffer and Leal de Sá (1996), Bicknell et al. (1998), and Frickmann Young (2000). Using structural decomposition analysis in a single-region input-output model, Munksgaard et al.

(2000) have analysed the factors affecting the development in CO2 emissions from private consumption in Denmark over the period 1966 to 1992, distinguishing between direct and indirect

CO2 emissions. The decomposition of CO2 emissions from Danish household consumption is performed at the global level, i.e. emissions from imported commodities as well as commodities produced in Denmark are taken into account.

The study finds that indirect emissions accounted for a major part of growth in total emissions from households, although CO2 emissions from direct consumption have exceeded emissions from indirect consumption during the whole period albeit to a diminishing extend. Indirect CO2 emissions, i.e. emissions from non-energy commodities, increased by 15%, mainly due to overall growth in private consumption. This growth was partly offset by energy conservation in production.

SCP001 Page 162 of 196

Reference assessed Munksgaard et al. 2000 (household consumption) Verdict Score 26, satisfactory Main strengths This study not only calculates embedded emissions but also demonstrates a general strength of IO analysis which is that it can be used to perform additional analyses such as structural decomposition analysis of changes over time. High sector disaggregation. Main weaknesses Data out of date.

Using a single-region input-output model, Jacobsen (2000) examines the relation between trade patterns and the energy consumption in Danish manufacturing industries. His results show that manufacturing sectors, such as chemicals or paper production, can be affected in opposite directions as a result of changes in trade patterns. Another interesting aspect of his analysis is the dependence of the results on the aggregation level of the input-output data: impacts of trade pattern on energy consumption obtained from a 27-sector model differ significantly from those obtained from a 117-sector model. This dependence is due to the aggregation of sectors with very different trade developments and energy intensities.

A paper by Sanchez-Choliz and Duarte (2004) describes the sectoral impacts that Spanish international trade relations have on levels of atmospheric pollution using a single region input- output model. The exports and imports of the Spanish economy are evaluated in terms of the direct and indirect CO2 emissions (CO2 embedded) generated in Spain and in the rest of the world. Their model consists of 18 economic sectors, including transport and other services. It uses a domestic and imports technical coefficient matrix and distinguishes direct and indirect emissions of domestic production, imports, exports and imports that are re-exported. The emission coefficients however are identical in all cases, assuming that traded goods were produced with the same technology as in Spain.

Four sectors are identified as main contributors to overall CO2 emissions in Spain, either by inducing the bulk of territorial emissions from the national economy or by purchasing production inputs and final goods from other countries. These sectors are food, construction, transport materials (vehicles) and other services. Exports of embedded CO2 emissions are mainly concentrated in the basic sectors of the Spanish economy, which are also responsible for the greatest amount of direct emissions, mining and energy, non-metallic industries, chemicals, and metals.

Reference assessed Sanchez-Choliz and Duarte, 2004 (Spanish economy) Verdict Score 25, satisfactory Main strengths Robust method with reliable data; could be adapted to the UK case. Main weaknesses Focuses on exports, more than imports; sector disaggregation low (18); no mention of uncertainties, assumptions or limitations.

SCP001 Page 163 of 196 Ferng (2003) adopts a single region input-output model to estimate the CO2 emissions according to the principles of consumption and production benefit (= consumer and producer responsibility). Ferng suggests using an emission-responsibility share, Φ, whose value is between zero (100% production benefit) and one (100% consumption benefit). That way, embedded emissions can be estimated according to different shares of responsibility for direct and indirect emissions, demonstrated in a case study for Taiwan’s economy in 1996. However, the paper does not explicitly address emissions embedded in trade and therefore was not assessed in the scope of this study.

Recently, a single region IO model has also been applied by the Centre for Environmental Strategy at the University of Surrey to calculate CO2 emissions embedded in imports to the UK (Carbon Trust, 2006). The study suggests that the UK is a net importer of carbon intensive products and services from abroad and quantifies the consumption emissions as being 11.7 MtC greater than the emissions from all UK production. However, the project team was not able to obtain a published description of the methodology (‘UK Carbon Attribution Model’) in the time frame of this project, despite several requests.

Most authors listed above carry out an input-output analysis of a closed economy, and subsequently apply multipliers obtained from this model to exports and imports. In this approximation, the imports structure does not enter the direct requirements matrix, and is hence not reflected in the multipliers. In contrast, Proops et al. (1993), Lenzen (1998), Kondo and Moriguchi (1998), and Machado et al. (2001) incorporate the imports matrix into the input-output framework. Their analysis showed that Australia, Brazil, Germany and the UK are net CO2 exporters, and that Japan is a net CO2 importer. Kondo and Moriguchi (1998) compare in detail sectoral CO2 intensities calculated in the open- and closed-economy approaches. Machado et al. (2001) apply a commodity-by-industry IO model in hybrid units (energy commodities in physical units and non-energy commodities in monetary units) to the Brazilian economy in 1995. They suggest that Brazil’s international trade policy should incorporate environmental concerns in order to harmonise the country’s trade targets with its environmental priorities. Similarly, De Haan

(2002) examines the Dutch trade balance in terms of emissions of CO2, acids, nutrients, and solid waste, using the domestic requirements as well as the imports matrix. Export surpluses are reported for all indicators. Reference assessed Lenzen, 1998 (Australian analysis) Verdict Score 24, satisfactory Main strengths Robust approach with high sector disaggregation. Main weaknesses Data out of date; no description of data uncertainties.

Reference assessed Machado et al., 2001 (Brazilian economy) Verdict Score 22, satisfactory Main strengths Flexible hybrid unit IO model. Main weaknesses Assumptions, limitations and error estimation not adequately documented; sector disaggregation relatively low (14 industries, 19 commodities).

SCP001 Page 164 of 196 The key strength of IO analysis is that it is a multi-sectoral framework, where interdependencies between sectors that are distinct in their production technology (such as resource and pollution intensities) can be quantified. The fact that each trading partner is only represented with one production sector in the models of Proops et al. (1999) and Imura and Moriguchi (1995) is a major limitation. Imura and Moriguchi (1995, p.190) do explain that a full multi-sectoral inter-regional input-output approach is not practical due to the lack of consistent data. Thanks to improvements in data availability and quality, the situation has changed in the last few years and more sophisticated models have been described recently (see next section).

In general, imports to one country come from a number of different regions with different technologies. In turn, each of these regions also places import demands on foreign economies. Thus, embedded production factors may continue far upstream in an international supply chain in the same way that inter-industry demands continue far upstream on the domestic level.

The mathematical formulation to analyze this problem comprehensively becomes more complex. A truly multi-region (MRIO) model is needed where inter-regional trade flows are internalised within the intermediate demand.

Multi-region input-output models

To set up national CO2 accounts in multi-region input-output models, additional data are required for the estimation process. These are: • national input-output tables, energy use matrices and fuel coefficients as defined above for at least one additional country and • bilateral import coefficient matrices (based on monetary data). Two types of multi-region input-output models can be distinguished, linked single-region models and true multi-region models. In the first case, national IO tables are exogenously linked with bilateral trade data for different countries or regions and embedded emissions are calculated for each national economy separately49. This approach only captures the last stage of an international supply chain of imports. True multi-region models, on the other hand, endogenously combine domestic technical coefficient matrices with import matrices from multiple countries or regions into one large coefficient matrix, thus capturing trade supply chains between all trading partners as well as feedback loops50.

After Isard’s (1951) introduction of input-output analysis into regional science, multi-region approaches were first applied to regions in Italy by Chenery (1953, as cited by Polenske 1989) and in the USA by Moses 1955 (as cited by Polenske 1989). Polenske (1976 and 1980) examines the economic interactions and repercussions between the coal mining, freight transport and electricity generation sectors in nine regions of the USA.

49 See for example Wyckoff and Roop (1994). 50 See also (Lenzen et al., 2004) and (Munksgaard et al., 2006a) for a distinction between uni- and multi- directional trade analysis.

SCP001 Page 165 of 196 A multi-sector, multi-region approach is described by Imura and Tiwaree (1994) who adapt an input-output model extended with energy statistics in order to attribute CO2 emissions to trade activities in ten countries in the Asia-Pacific region. The region is treated as if it is a closed economy and trade with the rest of the world is neglected. The model features 20 economic sectors and three types of fossil energy sources.

Imura and Moriguchi (1995) use an international trade matrix in monetary values as an input- output table, assuming that each country or region is represented by one economic sector; only Japanese industries are modelled in sectoral detail. Thus they can derive flows of embedded energy and CO2 emissions between twelve countries and regions, albeit without sectoral breakdown. They show that international trade tends to increase the disparity between North and

South in terms of energy consumption and CO2 emissions. Proops et al (1999) use regional trade flow data from the United Nations’ Statistical Yearbook51 in a multiregional input-output analysis to examine a sustainability criterion for both a closed economy approach and an open economy approach. However, each of the national economies is only represented by one sector and the analysis for the global economy, broken down into 12 regions, assumes identical resource intensities for both domestically produced and imported goods. Interesting is the quantitative comparison between the closed economy or single-region approach with the multi-region approach. The authors demonstrate that for countries with resource-intensive imports, such as the USA and many European nations, the sustainability index decreases when the economies are assessed in a multiregional framework. Similarly, Battjes et al. (1998) test the assumption of identical domestic and foreign factor intensities by examining the differences between energy intensities from a multi-region input-output system and the corresponding single- region systems. Using the consolidated input-output tables of a number of European-Union countries compiled by van der Linden and Oosterhaven (1995), they show that single- and multi- region energy intensities for Germany are equal, but that single-region energy intensities are lower for the Netherlands and higher for Ireland than multi-region energy intensities.

Suh and Huppes (2001) outline a multi-region generalised input-output approach for compiling life-cycle inventories for the industrialised world. Hayami et al. (1999) assess the bilateral trade in greenhouse gases between Japan and Canada. They employ a number of matrices in order to convert between the commodity classifications of different trade statistics, and also between free- on-board (f.o.b.) and cost-insurance-freight (c.i.f.) valuation. An interesting finding in their study is that almost all CO2 embedded in Japanese exports is itself induced by imports, while emissions from Canada’s exports were generated by the respective exporting industries. In their bi-regional analysis of energy and air pollutants in Japan and China, Hayami and Kiji (1997) examine fuel,

CO2, and SOx intensities, and also calculate indices of power and sensitivity of dispersion, in order to establish whether energy- and pollution-intensive industries are also strongly interlinked within the economy. Murata et al. (1998) use a combined Japan-US-EC-Asia input-output table from

1985 and show that for energy consumption, and for emissions of CO2, SOx, and NOx, Japan causes a higher environmental impact due to imports from East Asian countries than vice versa, both in intensity and embodiment terms. Giljum and Hubacek (2001 and 2004) describe a

51 http://unstats.un.org

SCP001 Page 166 of 196 calculation of land appropriation through international trade using a physical input-output model of the EU-15 countries. As no physical tables are available for all countries, they construct a preliminary PIOT by combining data from published national studies with material flow data for the rest of the EU-15. Petri (1976) suggests extending a bilateral trade quantity model with the dual price model. However, such an approach is quite data-intensive.

Atkinson and Hamilton (2002) examine natural resource flows (in value terms) between nine main world regions, and report a net ‘ecological deficit’ for the OECD, with surpluses in most developing regions. They concede however that the high degree of aggregation in their global model is likely to underestimate real resource flows.

The Directorate for Science, Technology and Industry of the OECD develop further the approach described by Wyckoff and Roop (1994). Ahmad and Wyckoff (Ahmad, 2003; Ahmad and

Wyckoff, 2003) present a framework for estimating CO2 emissions embedded in internationally traded goods based on input-output and trade modelling. The goal of this OECD project was to create an indicator that estimates the emissions associated with the total domestic consumption of economies in developed countries (consumer responsibility) as a complement to the more commonly used indicator for emissions from domestic production (producer responsibility). The calculations for 24 countries (responsible for 80% of global CO2 emissions) are based on national input-output tables on a 17 sector level, bilateral trade data for 41 countries/regions and IEA52 data for CO2 emissions from fossil fuel combustion.

The sector specific CO2 emission factors for exporting countries in Ahmad and Wyckoff’s model are based on emissions per monetary unit of exports. For the importing country under investigation separate import matrices for each country or region that exports to this country are established, distinguishing between imports for intermediate and for final demand. Emissions embedded in imports of intermediate goods are allocated to final demand of the importing country by using the national input-output table. For the embedded emissions of services only tentative estimates are included in the analysis.

Following this approach under conservative assumptions shows that estimates of CO2 emissions generated to satisfy domestic consumption in OECD countries in 1995 were 5% or over 0.5 Gt

CO2 higher than emissions related to production. The bulk of these excess emissions can be attributed to a few importing countries, mainly the United States, Japan, Germany, France and

Italy. Based on volume, the US alone account for nearly half of the total global CO2 emissions embedded in imported goods. The largest net outflow of emissions embedded in exports bound for OECD countries in 1995 came from China and to a lesser extent Russia. For individual countries, estimated emissions associated with imports or exports are often above 20% and in some cases over 30% of emissions from domestic production (extremes are Finland and Netherlands with over 40%, and Norway and Sweden with over 50%).

52 International Energy Agency, Paris, France

SCP001 Page 167 of 196

Reference assessed Ahmad, 2003; Ahmad and Wyckoff, 2003 (OECD analysis) Verdict Score 33, satisfactory (top score) Main strengths With 33 out of 38 possible points Ahmad and Wyckoff’s approach scores highest amongst all assessed studies. The study describes an application to the UK and updating the analysis, which is based on official and reliable data, as well as producing a time series would be feasible. Main weaknesses One major drawback is the low sector disaggregation (17 sectors) which limits the potential for applications and interpretations. The only unsatisfactory assessment comes from the fact that the data analysed are from 1995.

Researchers at RIVM, Netherlands (Nijdam et al., 2005) present an analysis of household environmental impacts based on a global input-output model, that differentiates production technology and emissions in the Netherlands and three different world regions. The analysis of Dutch household consumption in 2000 determines nine types of direct and indirect environmental impacts for seven consumption domains which in turn are based on 360 expenditure categories.

For the Netherlands and each of the three world regions environmental load intensities of industrial sectors were calculated separately using IO multiplier analysis. The technological matrices for the three world regions were constructed using IO tables of countries and sub-regions from the international economic Global Trade Analysis Project (GTAP) database53. Three import matrices describing the requirements of imports per region for Dutch production are derived from import statistics and used to calculate the foreign part of the intensities of Dutch production sectors per world region. The results from this study (Nijdam et al., 2005) show that most of the impacts take place abroad, except of greenhouse gases and road traffic noise for which 49% and 9% of the total impact takes place abroad, respectively. A substantial fraction of the impacts is due to imports from non-OECD countries. Most land use was found to take place in developing countries, whereas most emissions occur in industrialised countries.

Reference assessed Nijdam et al., 2005 (Dutch household consumption) Verdict Score 14, unsatisfactory Main strengths Includes not only GHG emissions, but all indicators for all LCA categories; addresses explicitly household consumption. Main weaknesses No adequate information about data sources, data handling, estimation procedures and uncertainty of data is provided; comparability and applicability for UK questionable.

53 GTAP (Global Trade Analysis Project) is a global network of researchers and policy makers conducting quantitative analysis of international policy issues (http://www.gtap.agecon.purdue.edu). Products from GTAP include data, models, and utilities for multi-region, applied general equilibrium analysis of global economic issues. The GTAP project is coordinated by the Center for Global Trade Analysis, Purdue University, USA.

SCP001 Page 168 of 196 Similar to Lenzen et al. (2004, see below), Peters and Hertwich, (2004) develop a consistent theoretical framework for a closed MRIO model to calculate pollution embedded in trade for arbitrary demands in the receiving economy (see also Peters and Hertwich, 2006b). Their approach is based on symmetric IO tables and is in fact identical to Lenzen et al.’s model (2004) if the make-use blocks used in the latter one were collapsed into symmetric matrices (see also Miller and Blair, 1985).

Peters and Hertwich (2004, 2006b) discuss a number of simplifications that lead to reductions in data requirements, without the introduction of large errors. Amongst these are the consideration of uni-directional instead of multi-directional trade, using trade shares to estimate import matrices and grouping similar countries into regions with identical technology. An application of their model to the case of Norway is described in Peters et al. (2004) and Peters and Hertwich (2005 and 2006a). Here, matrices with imports to Norway’s intermediate and final demand from seven countries or world regions are estimated by using the trade shares for each commodity from Norway’s seven major importing partners. The studies find that more than 50% of CO2 emissions from Norwegian consumption are embedded in imports. About half of these emissions occur in developing countries which represent only 10% of Norwegian imports. Assuming imports were produced with Norwegian technology leads to large differences in embedded emissions (Peters and Hertwich, 2005). The embedded CO2 emissions of direct imports to final demand, for example, would be underestimated by a factor of four! These findings clearly confirm the importance of considering regional technology differences when calculating pollution embedded in trade.

Reference assessed Peters and Hertwich, 2004 and 2005 (Norwegian imports) Verdict Score 24, satisfactory Main strengths Thorough description of theoretical framework; robust application; household consumption explicitly addressed. Main weaknesses No mention of uncertainty or margin of error.

As an example for a study that applies IO modelling in order to calculate environmental impacts other than air emissions, Guan and Hubacek (2006) present a paper that evaluates the embodiment of water in inter-regional trade in China. ‘Virtual water’ is the water associated (embedded) with the production of commodities along the whole production chain that are traded between regions or exported to other countries.54 For the assessment of trade flows and effects on water resources, Guan and Hubacek developed an extended regional input-output model for eight hydro-economic regions in China to account for virtual water flows between North and South China. The findings show that the current trade structure in China is not very favourable with regards to water resource allocation and efficiency. North China as a water scarce region virtually exports about 5% of its total available freshwater resources while accepting large amounts of wastewater for other regions’ consumption. By contrast, South China a region with abundant water resources is virtually

54 For or more information on the virtual water metaphor and its policy relevance see Allan (1998) and Wichelns (2004).

SCP001 Page 169 of 196 importing water from other regions while their imports are creating waste water polluting other regions’ hydro-ecosystems.

Reference assessed Guan and Hubacek, 2006 (virtual water use in China) Verdict Score (15) 55, satisfactory Main strengths Addresses the indirect use of water (‘embedded’ or ‘virtual water’); approach in principle transferable to the UK. Main weaknesses Specific data requirements preclude comparisons with other models; no mention of uncertainties, assumptions or limitations.

Multi-region input-output models including feedback-loop analysis A number of multi-region input-output approaches include an analysis of feedback loops in trade (Round, 1985; Sonis et al., 1993 and 1995; Sonis and Hewings, 1998). This concept is based on a decomposition of the Leontief inverse matrix into sub-matrices describing the disjoined interdependence of two sectoral or regional sub-groups in terms of internal and external multipliers, which was introduced by Miyazawa (1966), and then further developed by Cella (1984) and Clements (1990). Reinert and Roland-Holst (2001) utilise a social accounting matrix to examine industrial pollution feedbacks between NAFTA member countries. Their analysis treats air pollutants such as CO, SO2, NO2 and volatile organic compounds, but not CO2. Miller (1969) examines the deviation of single- and multi-region models in his “experimental” studies of Kalamazoo County and the USA, and of the USA and India. He concludes that (monetary) feedback effects are very small, typically affecting multipliers to less than 1%. These results were confirmed in other studies, as summarised by Richardson (1985, p. 631). However, in a study by Greytak (1970), interregional feedbacks appear to be significant, although this result seems to be an outlier when compared to others (see Tab. 4-6, p. 127 in Miller and Blair, 1985). Gillen and Guccione (1980), and later Miller (1985) show that the magnitude of feedbacks depends critically on the norm of the technical coefficient matrices (reflecting the interconnectedness of industries in the respective regional economies), on the level of aggregation and on the proportion of interregional trade in intermediate demand (i.e. the self-sufficiency of the economies). Round (2001) reviews the literature on feedback measures and upper feedback bounds, and also presents a decomposition of a multi-region Leontief inverse into feedback, spill- over and “Leontief” effects. There are no studies examining feedback effects in a generalised input-output framework incorporating factors such as labour, energy, or pollutants.

A detailed multi-region input-output model including CO2 and feedback-loop analysis is described by Lenzen et al. (2004). In order to calculate CO2 multipliers for multi-directional trade between Denmark, Germany, Norway, Sweden and the rest of the world they construct a consistent MRIO system which, as a central element, features domestic make and use matrices as well as use matrices for traded goods and services between all trading partners. The latter ones contain the essential information of how industry j in country s uses commodity i produced in country r in the

55 This verdict is not directly comparable with other studies as a different research question was addressed and specific data requirements exist.

SCP001 Page 170 of 196 intermediate production process. Using this extensive MRIO model, a compound total requirements matrix with the dimensions 1199x1199 is constructed, resulting in total, region- specific multipliers of intermediate demand, trade, energy consumption and CO2 emissions. With this closed model it is possible to include feedback loops and capture direct, indirect, and induced effects of trade.

Reference assessed Lenzen et al., 2004 (CO2 multipliers in MRIO models, see also Munksgaard et al. 2005 and 2006) Verdict Score 29, good (second highest score; only study with no unsatisfactory assessments) Main strengths Reliable and flexible method which allows the highest degree of sector disaggregation and analysis; thorough description of data preparation; robust application. Main weaknesses The rest of the world was modelled on the basis of Australian input- output, energy and CO2 statistics.

By running different scenarios with this model, Lenzen and colleagues (Lenzen et al., 2004) can demonstrate the differences in results when either domestic or foreign production recipes are used for traded commodities. In the case of Denmark, 18.9 Mt of embedded CO2 emissions resulting from a single-region model, which assumes that Danish imports are produced using Danish production structure, turn into 38.4 Mt of imported CO2 emissions when multidirectional trade with specific production recipes for the country/region of origin is considered (see also

Munksgaard et al., 2005 and 2006b). The first model shows an 11 Mt CO2 trade surplus whereas the second one results in a trade balance (0.3 Mt deficit). The example also shows that feedback loops induce changes in multipliers of around +1.5%. Lenzen et al. (2004) come to the conclusion that, in contrast to purely monetary flows, generalised feedback loops can under certain circumstances be significant, because they are often amplified by large physical factor contents

(for example energy, CO2, etc). Lenzen et al. (2004) also describe in detail the practical challenges of their five-region compound model and provide pragmatic assumptions and solutions for issues such as re-classification, currency conversion, valuation and estimation of trade flows. They show that the level of sector aggregation has a significant impact on the results and argue that therefore the most possible detail of disaggregation should be used.

Simulation models Other types of models can be used to calculate environmental impacts embedded in trade, such as global econometric-environmental models. A recent literature review by Uno (2002) identifies 34 simulation models with global coverage that have been developed since 1993, most of them focussing on energy-related questions56.

56 In the early 1990s, the University of Maryland presented a set of dynamic input-output models called INFORUM (“INterindustry FORecasting at the University of Maryland”) that are used to predict economic changes (Almon, 1991). INFORUM models are described as internationally-linkable, dynamic,

SCP001 Page 171 of 196 A number of approaches, based on general equilibrium (GE) modelling, attempt to quantify the amount of carbon leakage as a result of restrictive measures in OECD countries (e.g. Perroni and Rutherford, 1991; Oliveira-Martins et al., 1992; Pezzey, 1992). While Perroni and Rutherford (1991) and Oliveira-Martins et al. (1992) determine carbon leakages of not more than 10% and 16% of the initial emission reduction, respectively, Pezzey (1992) finds a 70% leakage offset of the carbon reduced by unilateral OECD action, thus rendering unilateral action largely ineffective in environmental terms. Nevertheless, Pezzey concedes that these differences “no doubt reflect the very different modelling assumptions [...] about world energy markets, as well as different data on energy supply elasticities”. This is confirmed by Paltsev (2001) who finds that not regional sector aggregation but “fossil fuel supply elasticities and trade substitution elasticities are the crucial determinants” in his static, multi-regional GE model of the carbon leakage resulting from the Kyoto protocol.

In most of the simulation models identified by Uno (2002) economic development is exogenous. Only seven models endogenise the economy and only two have a sectoral disaggregation deep enough to allow the distinction of different groups of goods and services. These are the GINFORS model (Lutz et al., 2005) and the GTAP model (Hertel, 1997; Dimaranan and McDougall, 2005). The core of both models is a multi-sector bilateral trade model.

The econometric model GINFORS (Global INterindustry FORecasting System) described by Lutz et al. (2005) (see also Meyer et al., 2003a and 2003b as well as Giljum et al., 2006) is not primarily used to calculate pollution embodiments of trade, but contains all essential elements to do so in the form of linked economy, energy and environment models with global coverage. A bilateral world trade model, closed on the global level, links national models for 25 commodity groups and one service aggregate, using bilateral trade share matrices. All EU-25 countries, all OECD countries and their major trading partners are explicitly modelled and time series from 1980 to 2002 are provided. The economic part consists of macro models for all countries and input-output models where data is available, which is the case for 25 countries, mostly European.

GINFORS is not a Computable General Equilibrium (CGE) but a macro-economic model (‘econometric input-output model’) that uses behavioural parameters estimated by econometric techniques to make simulations and forecasts of economic developments and their effects on markets and employment as well as global energy, resource and land consumption (‘ecological rucksacks’). To this end, GINFORS uses additional energy-emission models, material input models and land use models (Lutz et al., 2005).

interindustry models which imitate as closely as possible the way the economy behaves. They are intended for both public policy analysis and business forecasting and are different from classical input- output models, pure econometric models, and CGE models. Seven of these national models (US, Canada, Japan, Germany, France, Italy and Belgium) were linked by a mechanism that uses trade flows and prices at the individual sector level (Nyhus, 1991). However, the INFORUM system, though theoretically capable, has not been used to calculate emissions embedded in trade flows.

SCP001 Page 172 of 196 Reference assessed Lutz et al., 2005 (GINFORS model) Verdict Score 29, satisfactory Main strengths The GINFORS model includes all components required to calculate energy, emissions, material flows and land use embedded in trade, it has global and long-term coverage and has been applied in a major project with policy applications (MOSUS); high disaggregation of countries/world regions. Main weaknesses No embedded emissions calculated yet, so no comparison possible. Uncertainty of modelled data is not described.

GINFORS has been used as part of the European MOSUS project57 to simulate sustainability scenarios for Europe’s development until 2020. The MOSUS project has linked total resource use (comprising material flows and land use) to socio-economic indicators, e.g. growth and employment, in a global (multi-national and multi-sectoral) view (Giljum et al., 2006). As a follow-up it is intended to set up a global multi-country input-output model in order to quantify embedded natural resource requirements and to calculate comprehensive material flow indicators such as Total Material Consumption (TMC) (Giljum, 2005). Another simulation model with global coverage has been developed by the Global Trade Analysis Project (GTAP) 58 (Hertel and McDougall, 2003). In contrast to GINFORS which follows evolutionary theory assuming agents to decide under conditions of bounded rationality in non- perfect markets, the GTAP model is a static multi-region, multi-sector applied general equilibrium model. GTAP distinguishes 57 sectors and 87 countries/regions and is thus able to capture some detail of interactions between domestic sectors as well as international trading partners. A project sponsored by the US Environment Protection Agency (GTAP/EPA, 2005) uses GTAP data to develop a land-use and greenhouse gas emissions data base for use in the CGE model. The aim is to fill the gap of links between land use changes and net greenhouse gas emissions from agriculture and forestry, and to assess the costs of climate policies and their spill-over effects via international trade and sectoral interaction.

Kainuma et al. (2000) calculate embedded carbon emissions using a GE model, employing the GTAP database and IEA/OECD energy statistics. They compare their results with embodiments calculated via input-output analysis for an open and a closed economy (that is, including and excluding factor inputs from foreign economies, respectively). The study shows that changing from a closed to an open economy model increases the emissions responsibility of Japan, USA and the European Community, but decreases the emissions responsibility for Australia, Former Soviet Union and Eastern Europe, China and India. However, their particular input-output- and GE-based embodiments are only comparable within limits, since they address slightly different questions (static and dynamic, average and marginal analysis).

57 “Is Europe sustainable? MOdelling opportunities and limits for restructuring Europe towards SUStainability”, see http://www.mosus.net. 58 Center for Global Trade Analysis, Purdue University, USA (http://www.gtap.agecon.purdue.edu).

SCP001 Page 173 of 196 An interesting approach is presented by Chung (2005) who couples Ahmad and Wyckoff’s approach (Ahmad and Wyckoff, 2003) with the global trade CGE model developed by GTAP. In Chung’s GTAP model, the world is divided broadly into two groups, Annex I and non-Annex I, the former further aggregated into 5 regions consisting of US, EU, Japan, Eastern Europe / Former Soviet Union and Rest of Annex I and the latter into Energy Exporting Countries, China, Korea and the Rest of the World. CO2 emissions embedded in international trade for these nine regions of the world have been calculated. The measure BEET (Balance of CO2 Emissions Embedded in

International Trade) is used, defined as the difference between CO2 emissions related to domestic production and those related to domestic demand, or as the difference between emissions embedded in exports (EEX) and emissions embedded in imports (EEM) in international trade. The study tries to incorporate the interrelationship of international trade and emissions embedded therein in an applied general equilibrium model for global trade. CO2 emission data rely on the GTAP-E dataset which includes energy balance and greenhouse gas emissions data59. The CGE model context also enables the author to simulate how carbon tax on energy use will affect BEET for the 9 major regions. In order to compute the emissions aggregates, domestic emission coefficients for each industrial sector are estimated using sectoral emissions statistics of the GTAP-E dataset. Under the assumption that the production of exports and the production for domestic consumption in each industrial sector has the same efficiency, the respective emission coefficients are being applied to a bilateral exports matrix for 9 regions in order to arrive at a matrix of emissions embedded in bilateral exports of each country. Chung’s baseline calculations suggest that the countries/regions with the highest BEET deficit are Japan and the EU, with 7.3% and 3.9% of their domestic emissions, respectively. In other words, Japan and the EU import more embedded emissions than they export and thus carry some responsibility for emissions outside of their territory.

Reference assessed Chung, 2005 (GTAP application) Verdict Score 25, satisfactory Main strengths Makes use of an existing, consistent database with world coverage (GTAP, GTAP-E). Main weaknesses No mention of uncertainties, assumptions or limitations; not clear which year was analysed.

Two other models that have recently been described should be mentioned here as well. Nijkamp et al. (2005) have used the GTAP-E model to simulate the effects of different climate change policies such as carbon taxes, tradable emission permits, joint implementation and clean development mechanisms. However, no results on emissions embedded in international trade have been reported. Duchin, (2005) describes a globally closed input-output model for international trade. Her ‘World Trade Model’ is a linear programming model where the values of endogenous

59 GTAP-E is an extension of the standard GTAP model that adds a module for the substitution effects towards more energy efficient capital and a module of CO2 emissions resulting from the use of emission generating commodities in the production process (see Truong, 1999 and Burniaux and Truong, 2002). A comparison of the GTAP-E model with other CGE models can be found in Kremers et al. (2002).

SCP001 Page 174 of 196 variables – output, exports, imports, factor scarcity rents for each region, and world prices for traded goods – are determined through production assignments for all goods that are made according to comparative advantage. This model has been extended by Strømman and Duchin (2005) to the ‘World Trade Model with Bilateral Trade’ (WTMBT) which takes into account the geographically dependent freight transportation requirements for goods by means of four modes of marine transportation. In an application of the WTMBT, Strømman et al. (2005) examine the relationship between global CO2 emissions and factor costs in order to explore how changes in the global division of labour can contribute to reducing carbon emissions, albeit without explicitly addressing CO2 embodiments of trade.

SCP001 Page 175 of 196 III.3 Detailed Assessment Results

The three tables below show how the different studies were rated according to the assessment criteria.

Boundaries Data

INFO: study Coverage of INFO: number of world Real or modelled Criterion Clearly defined Comparisons considers trade in Year(s) analysed Time series Updatable Reliable indirect impacts regions / countries data …

Single region nonIO models estimation of indirect Harris, 2000 only goods, not services only for goods, not UK trading with Denmark and emissions from goods, not services (environmental accounts) included services Sweden electricity use Assessment 121 O1122

Shui and Harriss, 2006 traded commodities not yes for the US, but ? (not defined!) 2 (USChina trade) (EIOLCA database) defined questionable for China

Assessment O11 221OO

Multi region nonIO models Trade of 18 industrialized countries Muradian et al., 2002 only 11 most polluting only 11 most polluting only 11 most polluting only products from 11 with the ROW and developing 1976 to 1994 (IPPS database) sectors sectors sectors most polluting sectors countries Assessment 1OO O211 Wiedmann et al., 2005 trade data from 2001, CO2 factors from only goods, not services only for goods, not factors include indirect but one off data set still (data from Ahmad and goods, not services 8 + 1 but CO2 factors from Ahmad & Wyckoff included services emissions a major limitation Wyckoff, 2003) ca. 1995 (2003) Assessment 221 11O11

Single region IO models Wyckoff and Roop, 1994 (six OECD countries, manufactered goods 7 (6 OECD countries and ROW) manufactured products) Assessment 211 OO121

Lenzen, 1998 1 (Australia & ROW, single goods and services (Australian analysis) region)

Assessment 222 O1121

Wenzel, 1999 1 (Germany & ROW, single some might be difficult 3 example products 1993 (German products) region) to update

Assessment 111 OO111

Munksgaard et al. 2000 1 (Denmark & ROW, single goods and services 1966-1992 (household consumption) region)

Assessment 212 O2122

Machado et al., 2001 1 (Brazil & ROW, single nonenergy goods 1995 (Brazilian economy) region)

Assessment 222 O1112 SanchezCholiz and Duarte, 1 (Spain & ROW, single 2004 goods and services 1995, 1997 region) (Spanish economy) Assessment 222 O1122

Multi region IO models Ahmad, 2003; Ahmad and 24 IO tables, 41 countries/ goods (tentative Wyckoff, 2003 regions for bilateral trade; estimates for services) (OECD analysis) unidirectional Assessment 222 O2222 Lenzen et al., 2004 (CO2 multipliers in MRIO goods and services 5; multidirectional 1999-2000 models) Assessment 222 11122 8 (Norway plus 7 aggregated base year 2000 (2000 Peters and Hertwich, 2004 exporting regions, based on the goods and services values for data ranging (Norwegian imports) technology of the top 7 importing from 1995 to 2000) countries) Assessment 222 1O121 Nijdam et al., 2005 4 (NL + OECD Europe, (Dutch household goods and services OECD other, nonOECD); ??? consumption) unidirectional Assessment 222 11?OO

Guan and Hubacek, 2006 virtual water North and South China (virtual water use in China)

Assessment 1O2 OO111

Simulation models Lutz et al., 2005 All EU25 countries, all OECD 25 commodity groups (GINFORS model, MOSUS countries and their major trade 1980-2002 and services project) partners. Assessment 222 22221

Chung, 2005 goods and services 9 ? (GTAP application) Assessment 222 ?1111

SCP001 Page 176 of 196 Data (continued) Method robustness

Robustness Documentation of Peer reviewed Uncertainty of INFO: Number of Criterion INFO: data sources (credible & Objectivity assumptions and and width of modelled data economic sectors reliable) limitations understanding

Single region nonIO models

Harris, 2000 Eurostat; several national statistical offices; HM Customs no flaws but serious 11 ? (environmental accounts) and Excise limitations

Assessment 1 O22O

Shui and Harriss, 2006 US Census Bureau; EIOLCA database, Carnegie Mellon ? (not defined!) ? (EIOLCA database) University; disparate data sources

Assessment 1 O?1 2

Multi region nonIO models Industrial Commodity Statistical Yearbook (UN); International Muradian et al., 2002 Yearbook of Industrial Statistics (UN); Commodity Trade no flaws but serious 11 ? (IPPS database) Statistics (UN); Industrial Pollution Projection System (IPPS), limitations World Bank Assessment 1 O222 Wiedmann et al., 2005 85 commodity groups, only for trade data (data from Ahmad and based on 12 aggregated documented, not for Wyckoff, 2003) sectors for CO2 factors CO2 factors Assessment 1 121O

Single region IO models Wyckoff and Roop, 1994 (six OECD countries, 33 OECD IO tables and bilateral trade data, IEA energy data manufactured products) Assessment 1 1?22

Lenzen, 1998 >50 Australian statistics ? (Australian analysis)

Assessment O 2?12

Wenzel, 1999 German national statistics for IO and energy data, single 58 ? (German products) references for emission factors and other data

Assessment O O?1O Danish IO tables and energyflow matrices from Statistics Munksgaard et al. 2000 117 production sectors, Denmark; renewable energy data from MoEE and Danish ? (household consumption) 66 commodities Energy Agency and CO2 emission factors from European CORINAIR database. Assessment 1 2?12

Machado et al., 2001 19 commodities Brazilian govt' statistics, IEA conversion factors (Brazilian economy) and 14 industries

Assessment O 12O2 SanchezCholiz and Duarte, 2004 18 Spanish IO 1995, ‘‘Satellite Atmospheric Accounts’’ 1997 (Spanish economy) Assessment O 2?O2

Multi region IO models Ahmad, 2003; Ahmad and OECD IO tables, OECD Bilateral Trade Data, IEA sector disaggregation Wyckoff, 2003 17 energy/CO2 data low (OECD analysis) Assessment 2 1221 Lenzen et al., 2004 different for each (CO2 multipliers in MRIO country/region, ranging National IO tables and CO2 data models) from 39 to 229 Assessment 2 2222

Peters and Hertwich, 2004 Statistics Norway, Eurostat, OECD, many disparate data 49 (Norwegian imports) sources, mostly from governmental statistics

Assessment O 1221 Nijdam et al., 2005 30 for world regions, (Dutch household disparate data sources, incl. VROM and GTAP ? 105 for NL consumption) Assessment O 1?12

Guan and Hubacek, 2006 State Statistical Bureau of China, Ministry of Hydrology, 40 (virtual water use in China) hydrological model CHARM

Assessment O 12O2

Simulation models Lutz et al., 2005 (1) OECD, (2) the International Monetary Fund (IMF), (3) (GINFORS model, MOSUS 25 Eurostat, (4) the COMTRADE data banks of the UN and (5) project) the International Energy Agency (IEA). Assessment O 2211

Chung, 2005 57 GTAP data for trade and energy (GTAP application) Assessment O 22O1

SCP001 Page 177 of 196 Consistency Results Method can be With other standardised to Consistent with With standard Applicability to INFO: Indicators (env. impacts Comparability Criterion established provide fully Verdict UN SEEA classifications the UK embodied in trade) with other results methods/studies comparable results Single region nonIO models

Harris, 2000 CO2 (environmental accounts)

Assessment 1O12 11 21

Shui and Harriss, 2006 1CO2 (EIOLCA database)

Assessment O1?O OO 12

Multi region nonIO models

Muradian et al., 2002 SO2, NO2, CO, VOC, TSP, fPM (IPPS database)

Assessment OOO1 1 1 15 Wiedmann et al., 2005 (data from Ahmad and ?? CO2 Wyckoff, 2003) Assessment 1O22 12 22

Single region IO models Wyckoff and Roop, 1994 (six OECD countries, CO2 manufactured products) Assessment 2222 12 25

Lenzen, 1998 GHGs (Australian analysis)

Assessment 1211 12 24

Wenzel, 1999 total energy requirements and CO2 (German products) emissions

Assessment 1O11 11 12

Munksgaard et al. 2000 CO2 (household consumption)

Assessment 1211 12 26

Machado et al., 2001 energy and CO2 (Brazilian economy)

Assessment 1111 11 22 SanchezCholiz and Duarte, 2004 CO2 (Spanish economy) Assessment 2211 12 25

Multi region IO models Ahmad, 2003; Ahmad and Wyckoff, 2003 CO2 (OECD analysis) Assessment 2212 22 33 Lenzen et al., 2004 (CO2 multipliers in MRIO CO2 models) Assessment 2121 11 31

Peters and Hertwich, 2004 CO2, SO2, NOx (Norwegian imports)

Assessment 1211 11 24 land use, GHG emissions, acidification, Nijdam et al., 2005 eutrophication, summer smog, fish (Dutch household extraction, freshwater use, road traffic consumption) noise, pesticide use Assessment 1 1OO OO 14

Guan and Hubacek, 2006 (virtual) water (virtual water use in China)

Assessment O1O1 11 15

Simulation models Lutz et al., 2005 includes all components required to no embedded emissions (GINFORS model, MOSUS calculate energy, emissions, material calculated project) flows and land use embedded in trade Assessment 2121 N/A 2 29

Chung, 2005 CO2 (GTAP application) Assessment 2211 22 25

SCP001 Page 178 of 196 III.4 Technical Specification

III.4.1 General framework (ideal model)60 This section provides a theoretical exposition of the ‘ideal case’ of a MRIO framework for the calculation of an embedded emissions indicator, designed to meet all requirements mentioned in the introduction. We illustrate how such a model should ideally be constructed, regardless of practical constraints on data or resources. We explicitly identify the key practical problems that are likely to arise and what the most appropriate solutions may be. A practical way forward for the actual implementation is given in Sections III.4.2 and 6.3.4 (main report).

Theoretical framework For the purpose of simplicity, the following exposition (derived from McGregor et al, 2004b, and Miller and Blair, 1985) is given in terms of a 2-region world. However, it is straightforward to extend to the multiple region case (see Allan et al, 2004b). The key equation determining the vector of output q in the single region IO framework is given by the following equation (see Leontief, 1970, Miller and Blair, 1985): [1] ()−= −1 cAIq where q is an N×1 vector of gross outputs qn n = 1,…,N of each economic sector n, c is an N×1 vector of final demands cn n = 1,…,N for the output of each sector n. A is the direct requirements (or input-output coefficients) matrix, describing the amount of intermediate demand of output form sector n used by sector j. ( −= AIL )−1 is the N×N Leontief inverse with elements {n,j} describing the amount of output generated in each sector n per unit of final demand for the output of sector j.

Total pollution (emissions) generation (or resource use)61 in production is determined as [2] q = Ωqf

q q where f is a K×1 vector, with element fk (k = 1,…,K) being the total emission of pollutant k generated by all production activities in the economy. Ω is a K×N matrix where element ϖ kn, is the average emission of k per unit of gross output in sector n.

Then the standard IO attribution (Leontief, 1970, Miller and Blair, 1985) can be employed so the equation [1] is extended to [3] f c Ω()−= −1cAI

c c where f is a K×1 vector, with element fk being the total emission of pollutant k directly or indirectly required to satisfy final demand c in the economy.

If final demanders also directly emit pollutants, [3] would be extended for final demand as

60 This section is based on personal communications from Dr. Karen Turner, University of Strathclyde and Dr. Manfred Lenzen, University of Sydney, February 2006. 61 Compare McGregor et al. (2004a).

SCP001 Page 179 of 196 [3a] f ∗ Ωqc ()−1cAI +−= Ωcc where we distinguish the KxN matrix of emission coefficients for the N production sectors, now q c relabelled Ω , from a K×Z matrix, Ω , where each Kx1 column within has elements ϖk as the average direct emission of k per unit of expenditure by final demander. 62 For simplicity we abstract from this extension in the current exposition but, as shown in [3] and [3a], it is straightforward to introduce this element where appropriate.

Note that, in the closed economy example, it is the case that fq = fc, so that all emissions in production can be attributed to final consumption demand for the outputs of that production.

If we take equation [1] as region 1 in a 2-region world and separate the element c (final demand) into final demand in region 1 of commodities produced in region 1 (c11) and region 2 (c12), and final demand in region 2 of goods and services produced in region 1 (c21) and region 2 (c22), we have

−1 ⎛ qq 1211 ⎞ ⎛ 11 −− AAI 12 ⎞ ⎛ cc 1211 ⎞ [4] ⎜ ⎟ = ⎜ ⎟ ⎜ ⎟ ⎝ qq 2221 ⎠ ⎝ 21 −− AIA 22 ⎠ ⎝ cc 2221 ⎠

rs rs where elements Anj of the submatrices A are the N×J show the transactions between sector i in producing region r and using sector j in region s, per unit of output of sector j in region s . The partitioned matrix (I – A)-1 is the inter-regional Leontief inverse, breaking down the gross output multiplier for each sector in each region into gross outputs that are induced by domestic and by foreign final demand. In other words, by having partitioned the A-matrix for each region into local and imported intermediate consumption, and the c vector for each region into domestic and traded final demand, we can determine the level of inter-regional spillovers in terms of how activity in one region drives activity in the other.

Of course, the factor we are interested in here is the emission of greenhouse gases and other pollutants. Just as we extended the basis economic framework in equation [3] for the single region case, we simply introduce a (K×N) matrix of coefficients Ωq, showing the direct emission intensity of output in each production sector n for each region:

cc q −1 ⎛ 11 ff 12 ⎞ ⎛Ω 1 0 ⎞⎛ −− AAI ⎞ ⎛ cc ⎞ ⎜ ⎟ = ⎜ ⎟⎜ 11 12 ⎟ ⎜ 1211 ⎟ ⎜ cc ⎟ ⎜ q ⎟⎜ ⎟ ⎜ ⎟ ⎝ 21 ff 22 ⎠ ⎝ 0 Ω 2 ⎠⎝ 21 −− AIA 22 ⎠ ⎝ cc 2221 ⎠ [5] q q q q ⎛ Ω 1 cL + Ω 1 cL Ω 1 cL + Ω 1 cL ⎞ = ⎜ 1111 2112 1211 1212 ⎟ ⎜ q q q q ⎟ ⎝Ω 2 cL 1121 + Ω 2 cL 2122 Ω 2 cL 1221 + Ω 2 cL 2222 ⎠

c where f11 is a Kx1 vector telling us the amount of total emissions that stem from production c activities in region 1 to support region 1 final demand, while f21 tells us the amount of total emissions in region 2 production to support region 1 final demand. The sum of these, in a 2-region world, will give us the total emission for region 1 final demand:63

ccc [6] += fff 21111

62 Emissions occurring directly in households are caused by the combustion of household and car fuels. 63 As mentioned above, the direct emissions caused by final consumers is omitted here for simplicity.

SCP001 Page 180 of 196 And the total emissions of region 2 are equal to

ccc [7] += fff 12222

Similarly if we extend to the n-region case, this will simply involve summing down a column with an additional 2 entries for each additional region. For example f1 would become

ccc c [8] 21111 ... +++= ffff n1

c The indicator F for emissions embedded in imports to region (country) 1 ( F1 ) is the sum of emissions from production in all other regions 2,…,N that support the final demand of region 1.

N c c [9] 1 = ∑fF i1 i=2

III.4.2 How to deal with data problems64

Data requirements Analytical MRIO models bring up an array of practical problems, resulting in a large labour intensity of the task. There are three basic problems that have so far prevented the application of a full inter-regional framework of the type described above: data availability, data reconciliation and computability. Turner (2003) discusses the problem of the availability of physical environmental data reported at an activity/sectoral level that is consistent with that used for economic accounting. However, these problems are exacerbated when the emission and/or resource use content of imports is required. In order to estimate embedded emissions in the MRIO system in Eq. [5] information is required on65:

• a (monetary) IO table for each country from which imports are drawn (to determine the

elements of the inter-regional trade component, −As,r , of the partitioned inter-regional −1 Leontief inverse, [I − A] , in order to determine multiplier effects in the exporting country, s)66,

• the imports used as intermediate inputs for each national industry sector, broken down by

commodity and country of origin (to derive the elements of As,r ),

• the direct imports to final consumption in the national economy r, broken down by commodity

and country of origin (to derive the elements of cs,r for each external region, s, from which imports are drawn),

64 This section is based on Lenzen et al. (2004) and personal communications from Dr. Manfred Lenzen, University of Sydney, February 2006. 65 In addition to national IO, energy and CO2 data. 66 As described by Lenzen et al. (2004) these tables don’t have to be symmetrical and they don’t have to have the same sector disaggregation. Domestic make and use matrices as well as use matrices for traded goods and services between all trading partners can be used as well to derive a compound total requirements matrix.

SCP001 Page 181 of 196 • energy use matrices by fuel type and industrial sector for each country s from which imports are drawn,

• country-specific CO2 emission factors, i.e. the amount of CO2 emitted per unit of fuel used in country s,

• alternatively to energy use matrices and CO2 emission factors: CO2 emission matrices by industrial sector for each country from which imports are drawn. The IO tables from each of the exporting countries, s, would themselves need to have the following characteristics:

• a sector mapping (i.e. a matrix that maps, or re-classifies sector n in the exporting country, s, into sectors n in each of the importing nations, r),

• a comparable set of input-output coefficients for emissions (i.e. for each sector i there must be

a coefficient ϖ kn within the matrix Ωq showing the average direct emission intensity of producing one unit of output).

• equivalent IO data to track direct and indirect imports from all other regions that the exporting region, s, trades with.

Data availability Input-output tables are available for all developed and some developing countries and can be estimated or approximated for minor trading regions or where national tables are not available. Although the sector aggregation varies from country to country, the principal economic accounting framework is a standardised process (United Nations, 1999, 2003). Some data sources provide IO tables in a consistent format for a number of countries and or world regions, e.g. OECD (Ahmad, 2002) or the Global Trade Analysis Project (GTAP; version 6) which provides data for 87 world regions in 57 sector detail (Dimaranan and McDougall, 2005). The economic structure and production technology of countries without IO data can be approximated by a country or region with similar technology where data is available (see e.g. Ahmad and Wyckoff, 2003 and Lenzen et al., 2004). Another approach is presented by Peters and Hertwich (2006a, 2006b), who identify the most significant trading partners first and then allocate the minor trading partners to one of the major trading partners to make larger aggregated regions with fixed technology.

Detailed trade data are required for any model that deals with impacts embedded in traded commodities. Naturally, the more countries and regions are featured in the model, the higher are the data requirements. Trade data is available from several sources, but often data are missing or suppressed and data from different sources often do not match well. This requires additional processing, balancing and cross-checking for consistency (see below).

SCP001 Page 182 of 196 In the case of CO2, direct emissions or energy use data are required per economic sector for all countries/regions. This type of data is available from environmental accounts for many developed countries, but additional work might be required for some developing countries or regions.

In the last few years data availability has become better and more comprehensive due to improvements in input-output databases (Ahmad, 2002; Dimaranan and McDougall, 2005) and trade data and models (Eurostat, 2003; Pain et al., 2005) and environmental accounts (United Nations, 2003). Therefore, it can be expected that more comprehensive and robust modelling approaches will be developed in the near future.

In any case, constructing the database for a MRIO model requires a sophisticated method of data handling. The specific challenges are detailed in the following sections.

Cross-classifying sectors between countries Input-output classifications differ between countries, because their different economic structures are governed by particular commodity groups: agricultural, manufacturing and service sectors have different importance, and some inputs and outputs (such as coffee, district heat, uranium) might even only exist in a few countries. All off-diagonal flow matrices therefore have to be cross- classified with respect to the usage direction (industries) into the national classification of the receiving economy, resulting in rectangular matrices. This procedure releases from unnecessarily aggregating domestic and foreign sectors – which would result in the lowest common denominator and a considerable loss of detail – and keeps sector disaggregation to the most possible extent. This is accomplished by concordance matrices and data templates that break down one sector total into several matching sectors of the trading partner using appropriate proxies. Cross-classification is usually accomplished by scrutinising handbooks detailing the definition of national industry sectors.

Currency Conversion Currency conversion is necessary in models examining monetary flows in order to make the transaction entries in the monetary matrices and vectors (A*, y*, x*) comparable between countries. Some authors found it convenient to choose either an important trade currency (US$, £, €, ¥), or the currency of the main country appraised in the study as the common model currency. It is also possible to apply a mixed-units approach, in which the national production and final demand data including exports (rows) are kept in national currency units whereas the rest of the world is modelled in different units (see Lenzen et al., 2004). Diagonal (domestic production) blocks Arr are then in national currency units, while off-diagonal (trade) blocks Ars are in mixed units. Currency conversion rates can be obtained from OECD trade data.

Valuation Input-output tables come in different formats. In the use matrix, trade can be treated in two ways: (1) it can be incorporated into the domestic transactions matrix, treating imports as competitive,

SCP001 Page 183 of 196 and (2) it can be separated from domestic transactions, generating a separate non-competitive imports matrix. The latter is an essential requirement for truly multi-regional frameworks. If only a use matrix of type (1) exists for a country, imports flows can only be estimated from total imports by commodity, assuming that the usage pattern across domestic industries is identical to that of domestically produced commodities. For many developed economies, however, there is separate information on domestic and imported flows available.

Furthermore, entries in trade statistics or input-output tables will differ depending on whether transactions are valued at the point of producers shipping their products, or at the point of purchasers receiving them. The different valuations are referred to as basic prices and purchasers’ prices. While in general, domestic use matrices are provided in both valuation systems by national statistical bureaux, exports are often only available in f.o.b. (free-on-board) valuation (corresponding to basic prices plus domestic commodity taxes and commercial margins), and imports only in c.i.f. (cost-insurance-freight) valuation. The latter includes insurance and transport cost. Purchasers’ prices then consist of the c.i.f. tariffs plus domestic import duties and trade margins (see Hayami et al., 1999). Matrices valued in basic prices show generally a higher stability over time, because they are not subjected to potentially drastic changes in taxes and margins, and are therefore mostly used for analytical purposes (Furukawa, 1986).

An environmental MRIO framework would mainly be concerned with tracing physical and not monetary flows, expressed in basic prices. In the absence of information on margins and tax flows, and on imports flows in both f.o.b. and c.i.f. terms, a way to estimate imports matrices valued in basic prices is to (1) reduce entries in an existing c.i.f. matrix homogeneously using aggregated f.o.b./c.i.f. ratios, as well as tax rate and trade margins of the exporting country, for a whole commodity group, before applying the trade coefficients, and (2) to apply a RAS technique for balancing the resulting multi-region input-output table (Boomsma et al., 1991, pp. 22-24; see also Round, 1978a). These procedures were applied in a compilation of an inter-country input-output table for the European Community, and a subsequent analysis of spillover effects by van der Linden & Oosterhaven (1995).

Estimating missing data Even if compatible IO tables were available for any pair of trading nations, r and s, inter-regional trade data to determine the Ars matrices (where r ≠ s) are generally not available and have to be estimated.67 There is a range of international trade statistics that specify trade volumes in both f.o.b. and c.i.f. valuation. However, these statistics only detail the amounts of commodities traded rs between countries (a vector {ui } of commodities i exported by country r and imported by country rs s, see summary for goods in Tab. 6), but not their usage by industries (a flow matrix {Uij }). In other words, it is in general not possible to find information on the spatial origin of every intermediate and final import, disaggregated according to the consuming sector in the country of destination (see also Boomsma et al., 1991, pp. 7-8). This is mainly because of the considerable

67 For example, see Allan et al (2004b) who detail the estimation of inter-regional trade matrices for a 4- region model of the UK.

SCP001 Page 184 of 196 cost, time and resources that are associated with conducting international industry surveys (Round, 1978a,b).

One solution to the estimation problem for off-diagonal trade flow matrices is to use trade coefficients (a non-survey approach): u rs [10] c rs = i with c rs = 1 , i u rs ∑r i ∑r i describing the percentage of imports of commodity i into country s that come from country r. These trade coefficients can then be applied to an entire row of the national imports matrices s s {Mij } and imported final demand vectors {yi } in order to yield breakdown according to country of origin:

rs rs s rs rs s ij = i McU ij and i = i ycy i .

This procedure assumes that the trade coefficients are identical for all entries along a row of the imports matrix, i.e. for all using domestic industries.

Using trade coefficients as well as OECD trade data (Organisation for Economic Co-operation and Development, 2001), Hayami et al. (1999) give a detailed recipe for converting exports ξr from r (in r’s input-output classification) into imports μs to s (in s’s input-output classification) in a six- step process.

As a consequence of using the trade coefficients, the off-diagonal elements Ars (r ≠ s) have incomplete, extrapolated and therefore uncertain information, while the diagonal elements Arr are the original intra-country direct requirements matrices containing full, original information (Boomsma et al., 1991, p.9). In addition, the trade statistics mentioned above bear a range of uncertainties (Furukawa, 1986), which are due to (1) time lags between shipping of exports and receipt of imports, (2) differences in commodity classification (see 2.6.1), (3) reporting errors, (4) losses due to accidents in transits, and (5) discrepancies of origin and destination country due to commodity re-export. Factors such as these add to inconsistencies in multi-region input-output frameworks estimated using the pro rata techniques outlined above (Boomsma et al., 1991, pp. 10- 17). These inconsistencies require balancing techniques to be applied.

Balancing conflicting data Typically, many of the physical and monetary data to be assembled into a MRIO are conflicting, due to sampling and/or rounding errors, or due to differences in definitions or base years. Moreover, data often refer to different years. Strictly speaking, a temporal misalignment of surveys calls for a simultaneous balancing procedure of all data over all periods, imposing certain conditions of temporal stability (compare Tarancon and Del Rio, 2005).

Conflicting data causes conventional RAS balancing techniques (Gretton, 1979) to oscillate, thus preventing convergence. The more serious the data discrepancies, the more imperfect the final RAS result. Such non-convergence can only be dealt with using optimisation techniques. These

SCP001 Page 185 of 196 techniques have been the subject of recent issues in the expert literature (Economic Systems Research Vol. 16, No. 2, June 2004), and shall not be dealt with further here.

Technical issues of interpolating time series data Some technical issues of producing time series data shall be discussed here further.

First, if there are missing years for financial data or if the base year for countries/regions is different then adjustments for inflation are required to make the data consistent for a given base year. (Peters and Hertwich, 2006) describe the issues involved: “The easiest approach is to use the Consumer Price Index (CPI) in each country to adjust for inflation. However, the CPI is likely to introduce other errors. The CPI is an aggregated index, while price changes are likely to be different in each of the IO sectors. Further, the CPI also varies depending on the base year used and the method of indexing applied. These issues are difficult to resolve and the errors will be greater for a large CPI and when there is a big difference in base years.”

Second, if no analytical IO tables are available for certain years (or certain countries), separate supply and use tables (SUT) can be used instead. The availability is generally better for these types of tables. The latest analytical input-output table for the United Kingdom – which are being produced every five years – is currently from the year 1995. Supply and use tables in contrast are available on an annual basis. The SUT framework represents a complete picture of the economy showing all inputs (domestically produced goods and imports) and all outputs (domestic final consumption and exports) in monetary terms. Table III.2 provides an overview of the supply and use table framework in the UK. A supply table shows the commodities supplied by (domestic) industries for a particular year. Principal products are recorded on the principal diagonal, while secondary products are shown as the off-diagonal elements of the matrix. The industrial dimension of the combined use matrix shows, for each industry, the total costs incurred in the production process as intermediate consumption, including the costs for imported intermediate products and services.

Note that this particular system is closed with respect to imports and exports, i.e. imports are included in the combined supply matrix V, the combined use matrix U and – in the form of direct imports – the final demand matrix Ycom. Exports are included in the final demand matrix Ycom. Hence, the terms ‘commodities’ and ‘industries’ in Error! Reference source not found. refer to both domestic and foreign markets. Total industry and commodity outputs include imports, which in this approach are treated as competitive and – with respect to imports for intermediate demand – as endogenous. The principle of this approach can be applied to the multi-region case with separate tables for imports (see Lenzen et al., 2004).

Combining the supply and use table involves the following procedure.

The product dimension of the use matrix shows intermediate consumption and final demand by product and is valued at purchasers’ prices. Both intermediate and final demand estimates include

SCP001 Page 186 of 196 goods and services both domestically produced and imported. The officially published use tables are only available in a mixed price system (ONS, 2003) rendering it unsuitable for immediate application. In order to make the use table consistent with the supply table, the intermediate flow matrix needs to be transformed from purchasers’ into basic prices. In particular, this requires the exclusion of direct taxes and re-distribution of trade margins.

Commodities Industries Final Demand Total Output Commodities com q U ×nm Y ×om Industries x V ×mn Value Added W Total Input q x ∑ with: V = combined matrix for the supply of commodities (m) by industries (n); including imports U = combined matrix for the use of commodities (m) by industries (n) = intermediate flows; including imports Ycom = combined matrix for the final demands (o) for commodities (m), including direct imports and exports. q = commodity output vector, including imports x = industry output vector, including imports W = value added / primary input matrix

Table III.2 - Overview of a monetary supply and use table framework

Similar to the A matrix of the standard input-output model (Leontief and Ford, 1970) a technical coefficient matrix B with generic elements bij can then be derived from the use matrix (Miller and Blair, 1985).

⎡u ij ⎤ [11] []bB ij == ⎢ ⎥ ⎣⎢x j ⎦⎥ where: uij = use of commodity i by industry j and xj = total output of industry j including imports.

Each element bij represents the amount of commodity i required to produce one unit of the output of industry j. Therefore, the input-output systems can be written as: [12] += yBxq com where: q = commodity output vector (including imports); B = technical coefficient matrix; x = industry output vector and ycom = vector of the final demand for commodities. To derive the direct and indirect requirement matrix (generally known as the ‘Leontief Inverse’ in the standard input- output model), information on primary and secondary production needs to be added to the framework. For this, a matrix D can be defined whose individual coefficients dji are often referred to as commodity output proportions.

⎡v ji ⎤ [13] []dD ji == ⎢ ⎥ ⎣q i ⎦

SCP001 Page 187 of 196 where: vji = supply of commodity i by industry j and qi = total (domestic + imported) supply of commodity i.

Matrix D can be used to ‘weight’ the technical coefficient matrix B and assign all secondary products to the industry where they have been originally produced. This implies that we treat all secondary products as by-products being manufactured with the same technology as the principal product of this industry (industry based technology assumption). Alternatively, secondary products could be considered as subsidiary and assigned to the industry where they constitute the principal product.

A symmetric (industry-by-industry) input-output framework can then be constructed in the following way: [14] −1 ])[( ⋅⋅−= yDDBIx com where: x = industry output vector; I = identity matrix; D = industry-based technology coefficient matrix; B = technical coefficient matrix and ycom = vector of the total final demand for commodities. The bracketed term [(I-DB)-1D] represents the direct and indirect requirement matrix (the ‘Leontief Inverse’) of the SUT framework.

III.4.3 Details of extending the model Generalised input-output models can be extended to include other (environmental) impact data and indicators. Mathematically, equation [3] and following from above (section III.4.1) apply: [15] f c Ω()−= −1cAI where Ω is a K×N matrix with element ϖ kn, as the average factor k per unit of gross output in c c sector n and f is a K×1 vector, with element fk being the total amount of factor k required to satisfy final demand c in the economy.

Emissions of pollutants The Environmental Accounts of the United Kingdom, provided by the Office for National Statistics, report on a number of atmospheric pollutants by industrial sector. These include greenhouse gases other than CO2 [methane (CH4), nitrous oxide (N2O), hydrofluorocarbons

(HFCs), perfluorocarbons (PFCs) and sulphur hexafluoride (SF6)], acid rain precursors [ammonia

(NH3), nitrogen oxides (NOx), sulphur dioxide (SO2)], other air pollutants [benzene, 1,3-butadiene, carbon monoxide (CO), non-methane volatile organic compounds (NMVOCs) and particulate matter (PM10)] and heavy metals [Arsenic (As), Cadmium (Cd), Chromium (Cr), Copper (Cu), Mercury (Hg), Nickel (Ni), Selenium (Se), Vanadium (V), Zinc (Zn)]. Detailed accounts like these – particularly in the fine classification of 76 or 93 sectors – can directly be used in input-output analysis.

For each trading partner in a MRIO model, this information needs to be collected or generated. It can be expected that most developed countries with environmental accounts hold similar data of

SCP001 Page 188 of 196 atmospheric pollutants by industrial sector.68 Therefore, extending the MRIO model to include the emissions of pollutants is straight forward, solely depending on the data availability.

Water use ‘Virtual water’ is the water associated (embedded) with the production of commodities along the whole production chain.69 These commodities can be consumed domestically or exported to other countries. Like other environmental factors of production (e.g. land use), fresh water is a scarce and finite resource and understanding the causes for its consumption opens up the possibility of tailor-made environmental policies in both, water-poor exporting countries and countries importing large amounts of virtual water.

For the assessment of trade flows and effects on water resources, Guan and Hubacek (2006) developed an extended regional input-output model for eight hydro-economic regions in China to account for virtual water flows between North and South China (see section on ‘Multi-region input-output models’ above). Similarly, Velazquez (2006) presents an IO model of sectoral water consumption and applies it to Andalusia, a region situated in the South of Spain which is characterised by water shortage. An input-output analysis of Australian water usage, analysing the impacts of virtual water consumption is described by Lenzen (2004). The study demonstrates that it is uncomplicated to extend (multi-region) IO analysis with the use of water as a factor of production. Again, the availability of data for water use by economic sectors is the determining factor of feasibility.

Velazquez (2006) write (p. 227): “In an attempt to extend the NAMEA to water accounts, the European Union carried out a pilot project (European Communities, 2002) consisting in the elaboration of water accounts according to the methodology of the SEEA and the NAMEA, in the 15 countries of the Union. The project consisted in extending the matrix of national accounts with new rows and columns so as to represent the flows of the substances (water and polluting agents) in physical units, which were analysed jointly with economic transactions”. This indicates that, at least for a number of European countries, national accounts of sectoral water use exist that can be used to extend the database of a MRIO model. Lenzen (2004) describes how water accounts can be integrated into input-output tables. The study by Guan and Hubacek (2006) further shows that it is possible to use other data sources to generate the required input (the authors use hydrological data). It was not possible in the scope of the present project to analyse further the data availability and quality of water use in national statistics. For a comparison, the official account of water consumption by sector in the UK is depicted in the table below.

68 The System of Integrated Environmental and Economic Accounts (SEEA) was developed by the United Nations et al. (1993), together with other international organisms, in order to integrate international environmental accounts. A step further in these efforts is the National Accounting Matrix including Environmental Accounts (NAMEA), an environmental accounting framework developed by Statistics Netherlands at the end of the 1980s. It consists of a conventional national accounting matrix extended with environmental accounts in physical units, following SEEA indications (see e.g. (de Haan, 1999;Keuning et al., 1999;Vaze, 1999) and http://www.mnp.nl/mnc/x-en-0-b4.html). NAMEAs present all information in a compact and consistent format (United Nations, 2003). 69 For or more information on the virtual water metaphor and its policy relevance see (Allan, 1998) and (Wichelns, 2004).

SCP001 Page 189 of 196

Table III.3 - Consumption of water resources by industrial sector in the UK, 1997/98 (ONS, 2005, p.39)

Another good example of national water accounts is published by the Australian Bureau of Statistics (2000 and 2004). The motivation to produce water accounts at a detailed industry level in Australia were to better understand environmental impacts of different uses of water, the realisation of the paucity of water use data for Australia, and hence a lack of systematic information to guide decision makers. In the ABS (2000) water account, readily available water resources data from various government and non-government organisations was aggregated into a number of tables. Following the UN System on Environmental and Economic Accounting (SEEA) and the System of National Accounts, the physical water accounts are arranged in the form of a use and a supply matrix, similar to the make-use concept in input-output tables. This arrangement enables linking water data to economic data. Supply and use tables are produced for all Australian states and for the whole of Australia. Four types of use are distinguished: mains water, self- supplied (surface and groundwater), effluent reuse, and regulated discharge. All water is assumed to be extracted from the environment (surface water or groundwater).

SCP001 Page 190 of 196

Waste and recycling Waste can be classified into two major categories depending on the way it is generated:70 (1) production waste: materials and products generated as ‘undesired’, i.e. by-products in the production process or wastewater, and

(2) consumption waste: materials and products originally produced as goods for final consumption but turned to waste with the passing of time, such as old paper or discarded consumer durables.

Category (1) waste can be directly attributed to industrial sectors as demonstrated in Table III.4 and can thus be modelled as a factor of production in IO models. Waste of production by 123 sectors is held in a database by the UK Environment Agency71. It can be expected that similar data are being collated by government agencies in other (developed) countries, following guidelines such as the System of Environmental and Economic Accounting (SEEA) (United Nations, 2003). Very detailed accounts have been produced, for example, in Germany (Statistisches Bundesamt, 2001) and Denmark (Pedersen, 2004).

However, the classification of waste types (inert, organic, mineral, etc.) and the sector disaggregation might very substantially and therefore it is likely that the inclusion of waste generation into a MRIO model would require a considerable amount of data preparation work.

Category (2) wastes are not part of the production process and therefore will be described in an IO model as a factor of final consumption. However, depending on the final destination of waste from household and other final consumers70, some of this waste will be recycled and will therefore re- enter the production process72. According to the SEEA handbook (United Nations, 2003, pp.99) there are mainly two types of economic activities that may result in the absorption of residual flows within the economic system, recycling and waste (water) collection and treatment. Economic activities specifically involved in the recycling of material flows are “recycling industries” (ISIC 37) and “wholesale trade in waste and scrap” (ISIC 5149). The main activity of recycling industries consists of the mechanical or chemical transformation of materials in order to make these usable again as industrial inputs. The activity of wholesale trade in waste and scrap is restricted to waste storage, sorting etc. and is one where goods are sold in the same condition in which they are acquired without undergoing any physical transformation other than sorting, and packaging.

70 Waste can also be classified according to its four principal final destinations (United Nations, 2003, p.114): waste discharged into the environment, waste for recycling, waste for treatment and waste for landfill. 71 Environment Agency, 2006. Personal communication. 72 Of course this is true for production wastes (category 1) as well.

SCP001 Page 191 of 196

Table III.4 - Total waste arisings in the United Kingdom, 2002/03 (ONS, 2005, p.35)

The fact that waste flows and recycling procedures can be modelled with input-output analysis has been demonstrated in a number of studies. Examples include a static inter-industry model for analyzing economic and environmental effects of the recycling of waste, including a numerical example for alternative paper recycling scenarios based on the Dutch NAMEA data (Nakamura, 1999); a detailed input-output analysis of waste management in Japan (Nakamura and Kondo, 2002); modelling the relationship between economic activity and waste generation, treatment and disposal in the Scottish economy (Allan et al., 2004a,b); the treatment of waste data in physical input-output analysis (Dietzenbacher, 2005; Dietzenbacher et al., 2005) and waste allocation in a national input-output model (McDonald et al., 2006).

Materials and substances A methodology to extend the UK Environmental Accounts with material flow data from has been described in Wiedmann and Barrett (2005). This approach can be applied to any country with similar physical account data and therefore is suited for use in an extended multi-region input- output model. This has already been demonstrated with an allocation of natural resource use in the European MOSUS project (see www.mosus.net, Giljum, 2005; Lutz et al., 2005; and Giljum et al., 2006a).

SCP001 Page 192 of 196 The crucial point of modelling the flow of individual substances (e.g. Aluminium or Copper) or very specific materials (e.g. PVC plastic) with a (multi-region) input-output model is sector disaggregation. Normally, national input-output accounts combine all factories that produce similar commodities in one sector. In the UK, for instance, the IO tables identify • ‘metal ores extraction’ but not “bauxite extraction” • ‘non-ferrous metals’ but not “aluminium” • ‘other metal products’ but not “aluminium products” • ‘plastics & synthetic resins etc’ but not “PVC polymer” • ‘plastic products’ but not “PVC products” As described under ‘Data Handling’ above, the proposed MRIO model is kept flexible in terms of sector (dis-)aggregation. This allows dividing any sector into two or more sub-sectors and thus explicitly modelling the substance or material of interest. In the following, the example of aluminium is chosen to describe the basic procedure.

1. Those sectors mainly involved in the mining, producing and manufacturing of the substance need to be disaggregated, i.e. split in two sectors, one containing the factories that mainly produce the substance and its products and the rest.

2. In input-output tables one additional industry and one additional commodity is included. Financial data are needed for the supply and the use of the substance by other industries (e.g. how much aluminium products are being produced and how much are being bought by other sectors). Environmental and other impact data are needed for the additional industry (e.g. specific emissions of aluminium smelters). 3. Not necessarily all sectors involved need to be disaggregated. This is particularly important in a multi-regional context where it would require too much effort to disaggregate all sectors in all countries/regions potentially involved. Instead, a pre-analysis is undertaken, using the existing trade database, to establish where bauxite or raw aluminium is imported from.

4. Such an analysis could show, for example, that no bauxite is mined in or imported to the UK and that no raw aluminium is produced in the UK but that 95% of it is imported from only two countries or regions, e.g. South America (Brasil) and Australia. In this case, the sectors ‘metal ores extraction’ and ‘non-ferrous metals’ only need to be disaggregated in those two countries/regions, but not in the UK. This might be sufficient if only the impacts of bauxite mining or aluminium smelting are to be investigated.

5. If, for example, specific impacts of aluminium product manufacturing (e.g. can manufacturing) are the object of interest then the ‘other metal products’ sector need to be disaggregated as well, in the domestic economy as well in those economies that export aluminium products to the UK. Again, a trade data analysis can be used to identify those countries where the UK mainly imports from. Additional financial as well as environmental data are needed to perform a detailed analysis of substance flows. This is the case with any model employed, e.g. traditional life cycle analysis

SCP001 Page 193 of 196 (LCA). In fact, process-based LCA data, e.g. emission factors can be used in the input-output model.

Life Cycle Assessment Indicators Many Life Cycle Assessment (LCA) models already use input-output analysis in hybrid approaches to calculate comprehensive upstream impacts of the production of certain commodities (important examples for methodological and empirical studies are: Briceno et al., 2005; Cicas et al., 2005; Hendrickson et al., 1998; Joshi, 1999; Lenzen, 2002; Matthews and Small, 2000; Nielson and Weidema, 2001; Suh and Huppes, 2002; Suh, 2004a,b; Suh et al., 2004). IO models are particularly suited for life cycle assessments as they automatically take into account the impacts accumulated along complete supply chains. Generally, the use phase of products is not included in IO analyses but modifications to accommodate use phase modelling have been described (e.g. in Joshi, 1999). Embedded emissions data generated by the IO analysis can be multiplied post-analysis with specific factors for LCA impact categories such as global warming, acidification, eutrophication, tropospheric ozone creation (summer smog), stratospheric ozone depletion, human toxicity, ecotoxicity etc. (two prominent LCA impact factor data sets are the Dutch CML midpoint set and PRé’s endpoint Eco-indicator99 set). The same problem as for single materials and substances applies, i.e. IO analysis can only calculate embedded impacts for groups of commodities that are represented in the input-output tables. If single products are to be assessed then a sector disaggregation needs to be performed as described above.

International transportation Financial information on the demand for transportation can be found (United Nations, 1999; United Nations, 2003) … • … in the rows of transport industries in input-output tables showing the intermediate demand from all economic sectors as well as the final demand for services from those industries. The UK IO classification, for example, distinguishes five transport sectors: railway transport, other land transport, water transport, air transport and ancillary transport services. (Note that these are different from the distributing sectors motor vehicle distribution, wholesale distribution, retail distribution and postal and courier services, that only use transport services. Note also that there is no distinction between passenger and freight transport). • … in the additional row of transport margins that each sector pays for. Often transport margins are combined with trade margins and taxes/subsidies which makes it more difficult to quantify them. • … in the difference between export and import prices for a commodity. Exports are usually reported in f.o.b. (free-on-board) valuation (corresponding to basic prices plus domestic commodity taxes and commercial margins) whereas imports are reported in c.i.f. (cost- insurance-freight) valuation. The latter includes insurance and transportation costs.

SCP001 Page 194 of 196 Environmental impacts such as greenhouse gas emissions from transport service sectors are usually reported in national Environmental Accounts. The handbook of ‘Integrated Environmental and Economic Accounting 2003’ explains the accounting procedure (United Nations, 2003, pp. 99 and 149): “3.126. Connecting residual flow data to national accounts data will usually result in significant modification of the scope of the original emissions data. One particular point of concern in this respect is the delimitation problem with respect to international transport and tourism […]. Regular emission inventories usually cover emissions within the geographic territory of a country and do not correspond exactly to the coverage of the economy of resident units due to the existence of international transport. In order for the environmental consequences of these activities to be taken into account in the SEEA, it is necessary to allow for flows of emissions by non-resident units in the national environment and emissions by resident units outside the national environment. This means that in the SEEA, the output generated by international transport services such as airline and shipping activities is attributed to the country where the operator is resident.

3.127. This is not the case in the IPCC (Intergovernmental Panel on Climate Change) conventions on global warming emissions which specify that pollution from international transport is not attributed to individual countries. A greenhouse gas emission account based on the SEEA will deviate in this respect from the methodology laid down by the IPCC. Equally, emission inventories will usually record emissions that broadly correspond to the (national) geographical territory, including pollution from cars registered in foreign countries.

3.128. It is quite likely that different emission estimates may be compiled for different types of uses in one country. Sometimes emissions will be adjusted for incidental changes such as differences in the average annual temperatures. In other cases, emission figures estimated according to a geographical criterion will be used alongside figures estimated according to the resident criterion. A simple way to avoid misunderstandings is to publish these figures together with straightforward bridge tables which clearly explain these differences. […] Such tables are often published with the national accounts to explain differences in various income measures such as between domestic product and national income.

4.85. […] the IPCC has developed guidelines on the basis of which the emissions of greenhouse gases and so-called indirect greenhouse gases can be calculated and reported for each individual country. According to the IPCC guidelines, international bunkers for civil aviation and marine transport are reported separately from national totals. However, in the agreements on reductions for countries or regions as laid down in the Kyoto protocol, residual emissions from international transport have not been taken into consideration.”

In principle, the emissions of transport sectors reported by national statistics constitute a direct impact or production factor of an industry and can be re-allocated to final demand in any country or region which is included in a MRIO model. It is possible, for example, to quantify the CO2 emissions of road transport in the USA caused by the final demand for textiles in the UK. A limitation is the fact, mentioned above, that in most cases passenger and freight transport are aggregated in one sector which results in average multipliers.

SCP001 Page 195 of 196 It becomes more complicated and uncertain however, if one is looking at transport across national borders. Emissions often occur outside of national territories, e.g. in international air and sea space. As described above, the allocation of consumption of fuels for international aviation and marine transport – and consequently the allocation of CO2 and other emissions – is based on the amount taken from international fuel bunkers. It is not possible, at least not in practice, to exactly attribute these fuels to activities of specific countries.

Transportation costs for commodities as expressed in transport margins and freight costs for exported/imported goods provides valuable information but have several limitations as well: they do not necessarily distinguish between different modes of transport, prices might change considerably over one reporting period and they might be aggregated with other costs. Strømman and Duchin (2005) explicitly address freight transport in their world trade model which takes into account the geographically dependent freight transportation requirements for goods by means of four modes of marine transportation. However, in contrast to a descriptive analytical MRIO model, the intention of their model is different in that it tries to determine regional production and trade flows as well as world prices and scarcity rents, based on comparative advantage and the minimization of factor use. As exogenous variables, the model uses interregional distances and the weight of internationally transported goods.

In the scope of this project it was not possible to review and assess specific studies on embedded transport impacts, let alone provide a specification for a detailed methodology. Further research is required in this area to identify the best possible method that can be employed or combined with a MRIO model.

SCP001 Page 196 of 196