a methodology for csr reporting: assuring a representative diversity of indicators across...

17
A methodology for CSR reporting: assuring a representative diversity of indicators across stakeholders, scales, sites and performance issues Martin O’Connor a,1 , Joachim H. Spangenberg b, * a C3ED, Centre d’Economie et d’Ethique pour l’Environnement et le De ´veloppement, Universite ´ de Versailles Saint-Quentin-en-Yvelines, 47 boulevard Vauban, Guyancourt 78047 cedex, France b Helmholtz Centre for Environmental Research e UFZ, Vorsterstr. 97e99, D 51103 Cologne, Germany Received 19 July 2006; received in revised form 21 November 2006; accepted 17 August 2007 Available online 23 October 2007 Abstract This paper presents a framework called the CSR Deliberation Matrix for the structuring of CSR issue identification, stakeholder dialogues, indicator selection and reporting, with an overarching goal to achieve an appropriate balance between sensitivity to individual situations and the benefits of ‘‘generic’’ indicators applicable to a large spectrum of reporting contexts. We suggest guidelines to (1) define the full spectrum of sustainability concerns and of relevant stakeholder dialogue contexts; (2) mobilise a relevant ‘‘data bank’’ which provides a profile of candidate CSR indicators; (3) obtain a parsimonious selection of indicators in a site-level CSR reporting process through a stakeholder dialogue; and (4) obtain a ‘‘representative diversity’’ of indicators at the interface of site-level and higher-level CSR reporting contexts. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Corporate Social Responsibility (CSR) indicators; Deliberation; Governance; Representative diversity; Site-specificity; Stakeholder dialogue 1. Introduction The decade since the mid-1990s has been marked by a new societal demand for measuring the performance of the busi- ness sector relative to a spectrum of sustainable development goals (e.g., De Marchi [1]; GRI [2]; WBCSD [3]). This emerg- ing profile of Corporate Social Responsibility (CSR) places distinct requirements on company management, notably the need to address a Triple Bottom Line of economic, social and environmental performance. There is also an emphasis on new forms of social dialogue as a contribution to company reporting, strategy definition and decision-making (De Marchi et al. [4]; Faucheux, Nicola ı [5]; Le Dars [6]; Faucheux, Hue [7]; EEC, Fishburn Hedges [8]). In this paper, we report on the development of a rigorous ‘‘bottom-up/top-down’’ process for CSR evaluation and com- munication that takes into account the differences between plants, activities, countries and relevant stakeholder groups and bridges the gaps between site-level and higher levels of performance reporting. We motivate our methodological sug- gestions by close reference to findings of a study carried out dur- ing 2002e2004 for the European Aluminium Association (EAA). This work led to proposals to the EAA for an informa- tion management framework and for a set of guidelines that permits the efficient identification of a CSR indicator system responding to a range of communication needs at site or sector-wide level. 2 Considered as a case study, it highlighted the importance of four distinct and complementary sources of * Corresponding author. Tel.: þ49 221 2168 95; fax: þ49 221 2168 94. E-mail addresses: [email protected] (M. O’Connor), joachim. [email protected] (J.H. Spangenberg). 1 Tel.: þ33 1 39 25 53 75; fax: þ33 1 39 25 53 00. 2 The work was carried out by the C3ED supported by the European Alu- minium Association through the EAA ‘‘Aluminium for Future Generations’’ Programme, in two phases for which the principal reports are: Faucheux et al. [6] and O’Connor et al. [7]. We refer to these as, respectively, the ‘‘Phase One’’ and ‘‘Phase Two’’reports. 0959-6526/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jclepro.2007.08.005 Available online at www.sciencedirect.com Journal of Cleaner Production 16 (2008) 1399e1415 www.elsevier.com/locate/jclepro

Upload: martin-oconnor

Post on 26-Jun-2016

213 views

Category:

Documents


1 download

TRANSCRIPT

  • asal

    ach

    le Danc

    UF

    rm 2

    e 23

    The decade since the mid-1990s has been marked by a new munication that takes into account the differences between

    * Corresponding author. Tel.: 49 221 2168 95; fax: 49 221 2168 94.E-mail addresses: [email protected] (M. OConnor), joachim.

    [email protected] (J.H. Spangenberg).1 Tel.: 33 1 39 25 53 75; fax: 33 1 39 25 53 00.

    2 The work was carried out by the C3ED supported by the European Alu-

    minium Association through the EAA Aluminium for Future Generations

    Programme, in two phases for which the principal reports are: Faucheux

    et al. [6] and OConnor et al. [7]. We refer to these as, respectively, the Phase

    One and Phase Two reports.

    Available online at www.sciencedirect.com

    Journal of Cleaner Production 16 (2008) 1399e1415www.elsevier.com/locate/jcleprosocietal demand for measuring the performance of the busi-ness sector relative to a spectrum of sustainable developmentgoals (e.g., De Marchi [1]; GRI [2]; WBCSD [3]). This emerg-ing profile of Corporate Social Responsibility (CSR) placesdistinct requirements on company management, notably theneed to address a Triple Bottom Line of economic, socialand environmental performance. There is also an emphasison new forms of social dialogue as a contribution to companyreporting, strategy definition and decision-making (De Marchiet al. [4]; Faucheux, Nicola [5]; Le Dars [6]; Faucheux, Hue[7]; EEC, Fishburn Hedges [8]).

    plants, activities, countries and relevant stakeholder groupsand bridges the gaps between site-level and higher levels ofperformance reporting. We motivate our methodological sug-gestions by close reference to findings of a study carried out dur-ing 2002e2004 for the European Aluminium Association(EAA). This work led to proposals to the EAA for an informa-tion management framework and for a set of guidelines thatpermits the efficient identification of a CSR indicator systemresponding to a range of communication needs at site orsector-wide level.2 Considered as a case study, it highlightedthe importance of four distinct and complementary sources of1. Introduction In this paper, we report on the development of a rigorousbottom-up/top-down process for CSR evaluation and com-Abstract

    This paper presents a framework called the CSR Deliberation Matrix for the structuring of CSR issue identification, stakeholder dialogues,indicator selection and reporting, with an overarching goal to achieve an appropriate balance between sensitivity to individual situations and thebenefits of generic indicators applicable to a large spectrum of reporting contexts.

    We suggest guidelines to (1) define the full spectrum of sustainability concerns and of relevant stakeholder dialogue contexts; (2) mobilisea relevant data bank which provides a profile of candidate CSR indicators; (3) obtain a parsimonious selection of indicators in a site-level CSRreporting process through a stakeholder dialogue; and (4) obtain a representative diversity of indicators at the interface of site-level andhigher-level CSR reporting contexts. 2007 Elsevier Ltd. All rights reserved.

    Keywords: Corporate Social Responsibility (CSR) indicators; Deliberation; Governance; Representative diversity; Site-specificity; Stakeholder dialogueA methodology for CSR reporting:indicators across stakeholders, sc

    Martin OConnor a,1, Joa C3ED, Centre dEconomie et dEthique pour lEnvironnement et

    47 boulevard Vauban, Guyb Helmholtz Centre for Environmental Research e

    Received 19 July 2006; received in revised fo

    Available onlin0959-6526/$ - see front matter 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.jclepro.2007.08.005suring a representative diversity ofes, sites and performance issues

    im H. Spangenberg b,*

    eveloppement, Universite de Versailles Saint-Quentin-en-Yvelines,ourt 78047 cedex, France

    Z, Vorsterstr. 97e99, D 51103 Cologne, Germany

    1 November 2006; accepted 17 August 2007

    October 2007

  • information and expertise for obtaining candidate indicatorswhich can be considered for deployment in a CSR reportingprocess. These sources are:

    - Identification directly through a stakeholder consultationprocess;

    - Appraisal of indicator concepts provided by sector associa-tions, international agencies, etc.;

    - Looking at information sets the company uses for thepurposes other than CSR reporting;

    - Assessment of the indicator concepts identified or deployedat other sites.

    Exploitation of each of these sources involves interaction be-tween two or more groups of stakeholders. This also raises thequestion of the generic versus site specific character of in-dicators, which has to be resolved with reference to the severaldifferent purposes of evaluation, their distinctive communica-tion contexts and their respective organisational scales. Wesummarise in Box 1 some of the challenges that typically arise.

    The purpose of this paper is to present a set of methodolog-ical guidelines for CSR reporting work at site-level and for the

    The aluminium sector, like all major manufacturing sectors,involves the coordination of production, sales and transporta-

    1400 M. OConnor, J.H. Spangenberg / Journal of Cway that site-level activity can effectively be interfaced withCSR reporting perspectives at higher levels. The overallframework that we suggest has four main components:

    - Step one: identification of CSR performance and communi-cation goals: define the full spectrum of sustainability con-cerns and of relevant stakeholder dialogue contexts (seeSection 2).

    - Step two: exploitation of an Indicator Kiosk: create ormobilise a relevant data bank which makes an inventoryand provides a profile of candidate indicators (Section 3).

    Box 1. Research questions

    - What procedure for selection of indicators pro-vides for a satisfactory appraisal of CSR perfor-mance at a site-level?

    - On what basis might it be decided if an indica-tor proposed at a given site, might have mean-ingful application at another site (and, byextension, might permit some comparabilitybetween sites)?

    - How do the history of a site, the geography, thenational and company culture, and the prevail-ing regulatory environment bear on prioritiesfor CSR?

    - What priority should be given to selection of in-dicators that are generic not just in the senseof being widely applicable across sites but alsomeaningful at higher organisational scales,e.g., corporate, sector, or territorial regulatory

    and reporting contexts?tion activities across large distances. Social, technical, environ-mental and financial conditions of operations for productionand processing plants can differ widely from place to place.Indicator systems in this context are a type of multi-criteriamanagement support tool.

    Work in 2002 at three French sites (Phase One, Faucheuxet al. [9]) generated nearly 100 indicator proposals emergingfrom stakeholders experiences and views of social responsi-bilities. The 2003 work (Phase Two at the Italy, Germanyand UK pilot sites, OConnor et al. [10]) augmented the varietyof individual suggestions and permitted further comparativeappraisal from site to site. Outlines of the stakeholder dialogueprocess and the spectrum of suggestions by stakeholders ofissues for CSR reporting are provided elsewhere (Faucheux,Nicola [11,12]; Faucheux et al. [13]).

    We present in tabular form in the appendix, as our referencedata, the full list of indicator concepts suggested.

    - In Table A1 we list the indicator concepts emerging fromthe pilot site discussions. In this table the EAAs generic in-dicator concepts are also integrated, as derived by a mem-bership consultation in the 2003 EAA survey of Europeanaluminium firms. In the cases that do not have a closeequivalent in suggestions made at site-level, the EAA indi-- Step three: construction of a CSR evaluation matrix:exploit a selection of the candidate indicators in asite-level CSR reporting process engaging stakeholderdialogue with a full spectrum of target stakeholder groups(Section 4).

    - Step four: a balanced multi-scale interface: harmonise oroptimise the site-level CSR evaluation and reportingprocess in relation to higher-level coordination require-ments and, as required, for comparisons from site to site(Section 5).

    In Section 6 we give a short discussion of the question ofmultiple scales, suggesting how, pragmatically, it can be ac-commodated by classes of meta-information within the Indica-tor Kiosk and through a judicious identification of stakeholdercategories. After having established the main methodologicalrecommendations, in Sections 7 and 8 we return to the centralquestion of getting the balance right between generic andsite-specific considerations, across the spectrum of CSR perfor-mance issues and across the spectrum of stakeholders. Progres-sively through the paper, we present guidelines for effectiveCSR reporting procedures. These are numbered sequentiallywithin each section, e.g., 4.2 for the second key point withinSection 4.

    2. Discourse-derived CSR information andcommunication goals

    leaner Production 16 (2008) 1399e1415cators are provided, identifiable by being written in italicswith (EAA) as a suffix.

  • - In Table A2 we present a schematic overview of the distri-bution of suggestions from all stakeholder groups, for eachissue, from site to site.

    [5,11,12]). These considerations can be stated formally as:

    M. OConnor, J.H. Spangenberg / Journal of CThe comparative presentation in Table A2 will be an impor-tant reference point throughout this paper. It facilitates identi-fication of issues that are common across all or most sites andgroups, and that can, by induction, be considered as possiblegeneric issues. But there are many important issues thatare not common to all sites or to all stakeholders. We mustconsider individual industrial sites, and we must consideralso the transferability, or not, of indicators suggested at onesite, to other sites. We must furthermore consider the useful-ness of indicators at different organisational scales, in movingfrom a site management and communication context toa higher-order coordination perspective. The challenge thatwe take up is to bring all these concerns together.

    We suggest, as the cornerstone for this reconciliation, theuse of a standard set of CSR performance issues to structureinformation management, stakeholder dialogue and communi-cations.3 We suggest, moreover, that this standard set needsto be established, and thereafter validated periodically, througha dialogue amongst a cross-section of stakeholder groups.

    Table 1 presents the standard set of CSR indicator cate-gories established by the C3ED and the EAA in the courseof the reference study.

    In putting a bit of order into the diverse CSR suggestions, wehave followed some strands of the international academic andinstitutional literature referring to four dimensions of sustain-ability: financial/economic, social, environmental and politi-cal/institutional (cf. OConnor [15]; Spangenberg [16,17]). Sowe suggest:

    Guideline 2.1: the selection and deployment of indicators insite-level, company or sector-wide CSR evaluation pro-cesses, should be made with a discursively establishedand standardised list of CSR performance issues, with refer-ence to the four dimensions of sustainability (financial/economic, social, environmental and political/institutional).

    This convention has already been adhered to in the organi-sation of the EAA/C3ED set of categories as a basis for clas-sification of issues and indicators. For CSR reporting, thisstandard set of categories has several important roles to play:

    - It works as a bridge between bottom-up and top-downperspectives, allowing stakeholders at site-level (includingcompany management) to see how their particular concernsare examples of categories of social responsibility addressedby the international community, and vice versa;

    - It helps to build a common understanding within and be-tween stakeholder groups, about CSR reporting objectives;3 For a theoretical account of this procedure, see also OConnor and Frame

    [14].Guideline 3.1: an integrated framework for CSR reportingneeds to be based on three main principles:

    Recognition of site specificities: what are the social, geo-graphical, technological (etc.) factors that can have a bear-ing on the range of sites at which a proposed CSR indicatorcan meaningfully be applied?

    Stakeholder diversity: CSR reporting must include proce-dures for stakeholder dialogues that build up a sharedunderstanding of the different stakeholders concerns,permitting an appropriate balance of site-specific as wellas generic indicators.

    Full spectrum of CSR performance issues: a commonground for stakeholder dialogues and for CSR reportingat site and industry levels is assured through use of a stand-ardised set of CSR indicator categories based on sustain-ability considerations.

    Many questions arise about how best to develop and main-tain a record of the contextual information of indicators. Thisis a knowledge management problem calling for an inventory(or a catalogue) of the various information sets and candidateindicators made available to those engaged in CSR reporting.For each information category (viz. a potential indicator), it- It helps to achieve a consensus about appropriate indicatorsin each category of CSR reporting.

    Our methodology is distinctive compared with many otherapproaches to CSR or sustainability performance assessmentof business, in that it does not start with a pre-defined (typi-cally top-down) CSR or sustainability goal that is simply tobe operationalised by a set of indicators for managementuse. It considers indicator development as a deeply socialdecision-making process for which a diversity of viewpointsmust be brought together in order to furnish a comprehensiverepresentation of the direct and indirect impacts of and ona company. For this reason, the resolution by a diverse audi-ence of the question what is CSR? or what is sustainabil-ity is the core element of our methodology. This is why wespeak of discourse-derived CSR information e obtainedthrough reference to existing discourses which, schematically,might be termed merchant discourse with customers and sup-pliers including credit and insurance institutions, industrial re-lations discourse including staff, management, and e whereappropriate e works council and trade union representatives,and corporate citizen discourse between company, authoritiesand civil society.

    3. Profiling the candidate indicators for CSR reporting

    Our site studies developed a profiling system for candi-date indicators proposed at site-level, based on informationabout the context of each suggestion (Faucheux, Nicola

    1401leaner Production 16 (2008) 1399e1415provides a descriptive profile relating to the several contextsof possible deployment of an indicator.

  • e four dimensions of Sustainability

    mental Institutional

    ce u

    ce u

    ns

    t us

    of Cleaner Production 16 (2008) 1399e1415Guideline 3.2: it is desirable to develop a standardised indi-cator database, here referred to as an Indicator Kiosk,which can provide for all users a common framework forthe characterisation of CSR candidate indicators.

    The Indicator Kiosk is an interactive data managementframework e essentially a meta-information system e thathelps to assure two complementary aspects of knowledge qual-ity in the context of CSR reporting: (1) the scientific and tech-nical considerations of data accessibility and measurement,etc.; and (2) the user-oriented considerations of pertinence forframing a decision problem.

    Within the Kiosk, the detailed elements of meta-information should be organised in a hierarchical way, fol-lowing a principle of progressive disclosure of information.For each indicator (and thus, for the system as a whole)three levels of meta-information need to be established,which are as follows (see OConnor [16]).

    - At Level One, the user can see the overall structure ofthe Profile of the meta-information offered about anindicator.

    - At Level Two, a standard format is offered for the character-isation of individual indicators. This standard formatincludes what the UNCSD calls a methodology sheet,plus provision, in several places, for judgements and com-ments to be offered by system users (who are the stake-holders in the reporting or evaluation process).

    - Finally, the user may be referred to elements of supplemen-tary information that are perhaps rather long, perhaps quitespecialised, or not in a standardised format. This constitutesLevel Three of the system.

    Overall, the CSR Indicator Kiosk is envisaged as a toolpermitting a dialogue about indicators as well as an indi-cator based dialogue. It does this by presenting a profile ofeach indicator or information category that, in part, depends

    Table 1

    The EAA/C3ED standard indicator categories for CSR reporting in terms of th

    Economic Social Environ

    Competitiveness Working conditions/health and safety Resour

    Pay and Benefits Employee opportunities and relations Resour

    Revenues and payments Internal communications Emissio

    Production (physical) Community relationships Produc

    Source: EAA [20].

    1402 M. OConnor, J.H. Spangenberg / Journalon the contributions made by stakeholders (as users of the Ki-osk) in their processes of deliberation (see OConnor [19,20]for further elaboration on this point). Provision can be madefor characterising whether and how each indicator is utilisedin the (comparative) characterisation of sites. Furthermore,since indicators are (by hypothesis) selected with referenceto one or more specific CSR performance issues, the Kiosksindicator profiles should highlight the insight that each indica-tor is thought to offer on the relevant issues. Finally, informa-tion can be presented about the categories of stakeholderssuggesting an indicator and/or its deployment. All of theseconsiderations are conveyed within the structure proposedfor the Kiosks Part B.4

    The EAAs own preliminary CSR indicator catalogue estab-lished in the context of their mid-2003 data-gathering exercise(EAA [21]) has been one input into our Kiosk design. We showin Table 2 the comparison between the main components of theEAAs framework for indicator profiles, and the features of ourproposal for a CSR Indicator Kiosk. The significant differences(apart from ordering) are the Kiosks provision for character-isation of (1) indicator pertinence on a site-by-site basis, and(2) indicator pertinence from the standpoint of different stake-holder groups e types of meta-information that, for theirsector-wide priorities, were not part of the EAA profiling.

    In sum, the following set of guidelines is offered about thescientific profile and the organisation of information about theapplication of an indicator:

    Guideline 3.3: the CSR Indicator Kiosk profile for an indi-cator should start with a profile of the scientific status andquality of the indicator, e.g., sources of the information,units of measure, considerations of uncertainty, underlyinghypotheses for measurement and interpretation.Guideline 3.4: since indicators are, by hypothesis, selectedwith reference to one or more specific CSR performanceissues, the Kiosks indicator profiles should highlight theinsight that each indicator is thought to offer on the relevantissues.Guideline 3.5: the issue of site-specificity for site-levelCSR evaluation and for transferability and/or comparisonsfrom site to site can be addressed in terms of the fourdimensions of sustainability.Guideline 3.6: information should be retained about thecategories of stakeholder suggesting an indicator and/orits deployment, and the site(s) at which the suggestionsare made.Guideline 3.7: a characterisation should be included of theindicators meaningfulness or robustness in the face of

    se (National/European) Environmental management system

    se e global (International exchange) Company CSR strategy/policy

    and impacts Supply chain relationships

    e (life cycle)changes of scale, notably the possibilities of aggregationfor changes in organisational scale along physical (territorial,material and spatial), economic, and political axes.

    The general framework of an Indicator Kiosk is, potentially,a powerful tool in support of stakeholder dialogues, CSR

    4 For a discussion and full design illustration of the Indicator Kiosk frame-

    work for on-line knowledge management and stakeholder dialogue, with ref-

    erence to environmental governance processes, see OConnor [18] who used

    there the term Indicator Dialogue Box.

  • 4. A matrix framework for stakeholder dialogues

    CSR performance can and evidently will be consideredfrom a variety of different points of view. This leads to a sec-ond key design concept for organising CSR reporting, theconcept of a CSR Evaluation Matrix. In a multi-criteria,multi-stakeholder perspective, an evaluation of CSR perfor-mance for a selected site results in an array of judgements.We illustrate with the matrix structure of Fig. 1 where each

    Table 2

    Comparison of EAA and C3ED components of a CSR Indicator Dialogue

    Box framework

    EAA April 2003 exercise

    structure of explanatory

    document

    Components of the C3EDs

    proposed Generic CSR

    Indicator Kiosk

    1. Indicator

    1.1. Name xA.0 Name, acronym

    1403M. OConnor, J.H. Spangenberg / Journal of Cleaner Production 16 (2008) 1399e14151.2. Definition xA.0 Short definition1.3. Unit of measurement xA.1 Scientific profile1.4. Level of aggregation xA.2 Scale of the description

    xB.1 Pertinence at what scale(s)?1.5. Category xB.3 Pertinence for What? (with

    reference to the standard set of

    CSR performance indicator categories)

    2. Links to sustainable development and other indicators

    2.1. Purpose xA.1 Scientific profile InterpretationxB.3 Pertinence for What?reporting, and also strategic analysis including forward-look-ing studies (performance targets, investment scenarios, etc.).5

    In the rest of this paper we offer further guidelines for thedifferent aspects of indicator profiling, including the classifica-tion of CSR performance issues, the typology of stakeholdersand a method of characterising site specificities. This allows usto address the central issue of getting the right mix ofindicators through a process of stakeholder dialogue.

    2.2. Relevance to sustainable

    development

    xB.3 Pertinence for What?

    2.3. Links to other indicators xA.1 Scientific profilexA.2 Scale of the descriptionxA.5 Independent uses of pre-existingInformation

    xB.1 Pertinence at what scale(s)?3. Methodological description

    3.1. Measurement methods xA.1 Scientific profilexA.3 Knowledge quality assessment

    3.2. Aggregation method xA.2 Scale of the descriptionxB.1 Pertinence at what scales?

    3.3. Status of the methodology xA.3 Knowledge quality assessment4. Assessment of data

    4.1. Data needed xA.1 Scientific profile Type ofInformation

    x1.2 Scale of the description4.2. Data availability xA.4 Information source and status

    xA.5 Uses of pre-existing information4.3. Data sources xA.4 Information source and status

    xA.5 Pre-existing informationSource: EAA [20] Source: See OConnor et al. [10],

    Appendix 3.

    5 If the purposes of analysis include a forward-looking or what if? dimen-

    sion (e.g., future performance targets, scenarios for comparative evaluation of

    investment options), the Kiosk could also include information about evolution

    through time of those indicators used to characterise possible futures. Provi-

    sion is then made for characterising whether and how each indicator is en-

    gaged for exploration of what if? situations. This forward-looking aspect,

    although evidently important in company strategy at all levels, goes beyond

    the scope of our study and is not further discussed in the present paper.cell within the array corresponds to a judgement to be madeby a specific category of stakeholder for each category ofperformance issue.

    In Fig. 1, we have left aside (for the moment) the question ofsite-to-site comparisons (see Section 5). The stakeholder cate-gories are shown as successive rows of the array and the CSRperformance issues are shown as successive columns of the array.Consistent with our Guideline 2.1, a classification framework forthe CSR performance issues is presumed in terms of the fourdimensions of sustainability. This leaves the question of thespecification of stakeholder categories. Here we follow the rea-soning and experience of our EAA study design.

    Guideline 4.1:the various CSR stakeholder dialogue pro-cesses can usefully be structured with reference to threemain categories of stakeholders:

    Internal stakeholders (including employees, companymanagement and non-staff shareholders, all having directeconomic interests in the company);

    The external stakeholders as traditionally identifiedbusiness partners (suppliers, customers, banks, etc., allhaving direct commercial importance to the company);

    The broader external stakeholders as discourse partners(NGOs, associations, partner companies, local authorities,all having an interest in, or claims about business perfor-mance, and therefore having an indirect significance forcommercial success).

    By separating out the second and third categories, an im-portant distinction is made between traditional externalstakeholders and the extended or broader stakeholder set,helping to see the different needs of (a) those stakeholderswho are of interest to the company as opposed to those whohave an interest in the company (including civil society atlarge). In this way, a high level of relevance for the targetgroups is guaranteed, and the potential for resonanceenhanced.6

    6 The first global CSR stakeholder survey by ECC [8] has found that the

    most frequent readers of CSR reports are not the target stakeholder groups

    but, rather, consultants and competitors. Most CSR reports, in their structure

    and content, lend themselves to this mismatch between target readership

    and actual readership. The sorts of information and analysis that companies

    will tend to report from their in-house perspectives, is likely to be easily ac-

    cessible to competitors, whereas it will not necessarily be so for external stake-

    holders (nor, necessarily, pointed in the directions of what they would like to

    know). But more importantly still, in a lot of reports neither the relevant stake-holders nor their specific interest profile are clearly defined in the companies

    communication efforts.

  • Coordination

    institutions

    ociormssue

    brilliehol

    ion

    1404 M. OConnor, J.H. Spangenberg / Journal of CAs a function of need, there are various possibilities fordisaggregating each of the three categories.

    - First of all, company management, and non-staff share-holders, may be considered as specific subgroups of stake-holders having direct interests in the company.7 Theperspectives of shareholders would be interesting because,by definition, this raises the question of the expectationsplaced on a company as an investment, including the contro-versial question of socially responsible investment (SRI).

    - Second, the two categories of external stakeholders covera great diversity of people and organisations. Fig. 2 (fromthe C3ED Phase One study, Faucheux et al. [9]), presentsthe spectrum of traditional stakeholders (internal and ex-ternal, dotted outline boxes) and the more recently consid-ered wider external stakeholders (solid outline boxes).

    Internal (Company) Stakeholders

    External traditional Stakeholders

    External extended Stakeholders

    Stakeholders

    x Issues

    EconomicPerformance

    Issues

    SPerf

    I

    Key : We use indicative symbols (good) (not logic of the Matric as judgements by a stak

    Fig. 1. The CSR evaluatOur characterisation of stakeholder dialogue processes asengaging internal and external stakeholders partly over-laps with the considerations of communicating from plant siteup to higher levels. Looking at the schematic typology inFig. 2, it can be seen that Governments and ProfessionalAssociations are found amongst the traditional externalstakeholders, whereas Territorial Communities are includedamongst the wider external stakeholders. We suggest:

    Guideline 4.2: for the wider CSR reporting context, it isuseful to consider coordinating authorities (e.g., manu-facturing sector associations, and regulatory authoritiesand territorial agencies) as constituting a distinct stake-holder type.

    7 In the course of our work on pilot sites, a spectrum of the management per-

    sonnel at each pilot site was interviewed mostly informally. These interviews

    and informal conversations with management at each of the sites ensured the

    integration of their opinions and suggestions. By contrast, we did not solicit

    non-staff shareholders in the site studies.This leads to the matrix framework shown in Fig. 1 for theCSR evaluation matrix for a single site. Just as much for theother types of stakeholders, the inputs and judgements ofcoordinating agencies will be based on their specific perspec-tives and concerns which complement those expressed bystakeholders at site-level.

    The role of this matrix framework, including its provision fortop-down contributions by coordinating authorities (the toprow) as well as bottom-up contributions by others, is to pro-vide a supporting structure for a transparent multi-stakeholderdialogue process. The interfacing of stakeholders contribu-tions is what will permit a balanced selection of indicatorsjudged pertinent by the spectrum of stakeholders involved.

    Guideline 4.3: within the framework of the CSR evaluationmatrix, a site-level CSR assessment should be obtained asthe outcome of a stakeholder dialogue process.

    alances

    EnvironmentalPerformance

    Issues

    InstitutionalPerformance

    Issues

    ant) (really bad), simply to highlight the der class for each performance issue.

    matrix for a single site.

    leaner Production 16 (2008) 1399e1415Any process of indicator selection for CSR reporting willhave several steps. These may proceed from initial identifica-tion of candidate indicators to a final reduction of the num-ber of indicators to a small manageable cross-section ofinformation. In the real world, the steps are not always under-taken in a completely linear way, and they may involve variouscombinations of individuals and organisations. However, it isuseful for the development of guidelines to adopt a formalpresentation of a possible logic of the process.

    A stakeholder dialogue process for making a site-level CSRassessment can have, for its goal, to select indicators allowingthe target stakeholders to arrive at a judgement for each cell ofthe CSR evaluation matrix. We refer to this as filling in thecells of the CSR Evaluation Matrix. For example, represen-tatives from each category of stakeholder may work togetherto select, with reference to each CSR performance issue,a basket of indicators from amongst the candidate indica-tors. In case the participants find the list of candidate indica-tors insufficient to cover their specific concerns, new indicatorsuggestions will emerge which can be used to complement the

  • keh

    M. OConnor, J.H. Spangenberg / Journal of Cexisting list in an iterative process. Based on the resulting se-lection of indicators, an overall judgement is to be put forward,by each category of stakeholders, for each of the performanceissues being considered.8

    The dialogue process can do much more than merelyachieve a selection of indicators and a signalling of the perfor-mance assessment associated with the chosen indicators. Itcreates an opportunity for exchange and debate between stake-holders who will learn about what matters to the others and

    Fig. 2. The variety of stawhy. Therefore, the deliberative process can potentially beimportant for partnership building and for the perceived legit-imacy e as well as the information quality e of the outcomes.

    5. Building the dialogues across sites

    The Indicator Kiosk and the CSR evaluation matrix aretools that bring together a full spectrum of candidate indicatorsand that highlight the stakeholder dialogue/communicationcontexts. Let us now consider site-specificity and the questionof transferability of an indicator concept from site to site, ina formal way. Site-specificity can be appraised as a responseto the question: does the indicator concept apply to all alumin-ium sector operations/sites or only to some? The simplest

    8 For example, as a visual convention one might use symbols (as in Fig. 1),

    or one might imagine a judgement signalled by the colour of the cell in the

    Matrix, with the convention of traffic lights (red for bad, orange for poor, green

    for good). The result is a multi-colour array showing the pattern of stakehol-

    der performance issue judgements. Various issues arise concerning the wayto go from a selected basket of indicators to an overall judgement or score

    for the cell (and, by extension, for an entire row, signalling the stakeholders

    overall judgement). In principle each judgement can be arrived at informally,

    or through some formal process of assigning a value and relative weight to

    each indicator selected in the basket. These are typical questions for any mul-

    ti-criteria assessment framework, but we do not go into them here.answers could be yes or no. But there will often bedegrees of more or less.

    Guideline 5.1: each indicator retained within the IndicatorKiosk as a candidate for possible deployment, can be char-acterised for its degree of site-specificity along the fourdimensions of sustainability:

    Along the economic dimension (type of industrialactivity);

    olders in CSR reporting.

    1405leaner Production 16 (2008) 1399e1415Along the institutional dimension (regulatory, legal con-text, ownership considerations);

    Along the physical/environmental dimension (geograph-ical location, climate, etc.);

    Along the social dimension (workforce relations, com-munity concerns and expectations).

    Along each axis, the same question is posed: Does the in-dicator concept apply to all aluminium sector operations/sitesor only to some? If there is a no answer along any of thefour dimensions, then the indicator has a significant degreeof site-specificity.

    There are cases where, even if it is possible to apply onespecific indicator across the board, it is not necessarily usefulor desirable. An indicator may be applicable from a technicalpoint of view, but of only minor relevance in the context ofother sites (e.g., fluorine emissions are important for smeltersbut not for recycling plants). Or, an indicator it may be rele-vant for several sites, but with a quite different meaningfrom one site to another (e.g., gender balance in the work-force). It is therefore important to assess whether or not an in-dicator transfer is justifiable beyond its technical feasibility. If,after consideration of the four aspects of site-specificity, anindicator is considered to have broadly comparable interpreta-tions (although not necessarily similar consequences, as these

  • are determined by the respective legal and cultural frame-work), then the indicator can be considered generic.

    For industry-wide CSR reporting procedures, an empiricalprocess of site-to-site comparison has to be worked up intoa methodical procedure. Therefore, our complete frameworkfor CSR stakeholder dialogues, indicator selection and report-ing is obtained by taking the site CSR Evaluation Matrix(as presented in Fig. 1) and envisaging a methodical compar-ison of CSR evaluation profiles between sites. This yields thethree-dimensional framework shown in Fig. 3 that we call theCSR deliberation matrix.

    This framework will be the core of our subsequent discus-sions and recommendations.

    6. The question of scales

    We must also consider, from the starting point of a site, the

    1406 M. OConnor, J.H. Spangenberg / Journal of Cquestion of aggregation to higher scales. Here, we once againfollow the strands of sustainability literature that emphasisethe need to take account of four different aspects of system or-ganisatione the inter-sector structuring of economic activities,the physicalespatialeecological organisation of natural sys-tems (biological and environmental infrastructures), the mean-ingful structures of community identity and cultural belonging(the heart of the social dimension), and governance/institu-tional arrangements. In Fig. 4, we evoke this multi-dimensionalmulti-scale context in a bottom-up asymmetric way, high-lighting the challenges for the discursive actor to be positionedwithin the larger scales.

    Guideline 6.1: the Indicator Kiosk profile should specify atwhat organisational scale(s) the indicator is considered tobe pertinent and across what changes of organisationalscale the indicator continues to have pertinence. Four dif-ferent aspects of organisation should be noted: the eco-nomic, the physical/spatial, the social-cultural, and theinstitutional-regulatory dimensions.

    The CSR Performance Issues

    (or evaluation categories)

    Th

    e C

    ate

    go

    rie

    s o

    f

    Sta

    ke

    ho

    ld

    ers

    The Sites Fig. 3. The CSR deliberation matrix structure for cross-site comparisons.For each information category being considered, we can askwhether an indicator defined at a site-level, is transposable eby aggregation or other well-defined transformation e forapplication at a higher scale or economic, physical, socialor regulatory organisations. If the answer is yes, then theindicator concept has the corresponding sort(s) of genericityacross changes of scale.

    The guidelines 5.1 and 6.1, which bring the considerationsof pertinence at which site(s) and pertinence at which scale(s),are complementary to the indicator profiling considerations al-ready discussed with reference to the CSR evaluation matrix inSection 4, and the 3-D CSR deliberation matrix in Section 5,namely, pertinence for which stakeholders and pertinence forwhich CSR performance issues? In the language of the Indica-tor Kiosk structure proposed in Section 3, we now have thecomplete set of components for characterisation by contextof use and user.

    However, the question arises, do we need to introducea fourth dimension into the deliberation matrix in order tohave a dialogue across scales? There are some fields ofsustainability assessment (such as integrated analysis in sup-port of management of coastal resources, climate change,territorial freshwater, etc.) where this is almost certainly desir-able. We suggest, however, that in the CSR context, the multi-scale challenge can be met by an astute use of stakeholderdialogues.

    Stakeholders will naturally propose or prefer indicators thatrespond to their particular concerns and at the scales of theirconcerns. We have already recognised this when, as alreadyproposed in guideline 4.2, we consider the coordination insti-tutions as constituting a distinct stakeholder type, with theirown specificities. More specifically, we postulate that,

    Guideline 6.2: distinct coordination interests and, hence,stakeholder subgroups, are likely to be associated withthe challenges of governance of each of the four systemtypes mentioned above: the economic (e.g., sectoral), theenvironmental (e.g., biophysical systems and cycles), thepolitical (e.g., indicators for cross-industry benchmarking,regulatory policy and monitoring) and the social (e.g., theprestige of a particular community).Guideline 6.3: the appropriate balance in selection and/ordeployment of indicators applicable to high scales of sys-tem aggregation or representation, relative to other legiti-mate stakeholder concerns including local interest (andhence perhaps site-specificity), should be sought through de-liberation within the overall CSR stakeholder dialogue pro-cess, through the incorporation of appropriate sub-classes ofgovernance stakeholders for each of the four types ofsystem organisation.

    The problem of identifying generic indicators is one partof a multi-facetted question of the pertinence of an indicatorconcept as we shift from one site to another, from one scaleto another, from one organisational context to another. From

    leaner Production 16 (2008) 1399e1415a scientific standpoint, only very specific categories of informa-tion (such as mass, or energy content, etc.) can be aggregated

  • collective identity, etc.) emerges in quite specific management

    nce or Politicalposes of national and agencies) Aggregation across a Physical-

    spatial Dimension (e.g., territorial environmental

    considerations, watersheds, globalatmosphere, etc)

    The Site or Local Scale (Information relating to peoplesobservations and interests asmembers of communities in

    specific localities)

    anges in organisational scale.

    1407of Cleaner Production 16 (2008) 1399e1415and governance contexts but is not a preoccupation of every-one. Indicators having a cross-site industry-wide applicabilityand/or amenable to aggregation at company, national, or sectorscales, are features of interest only to a certain number of stake-holders (e.g., coordination institutions and policy scientists andsystems modellers).

    7. Getting balance in indicator selection: the EAA case

    We return now to the question at the heart of our paper: pro-cedures for getting the indicator balance right, across all theseacross sites and across physical scales, without loss of qualityor pertinence. From a sociological standpoint, the concernwith cross-site comparability and with aggregation from site-level to higher organisation levels (company, sector, territory,

    Aggregation for GovernaCoordination pur

    (e.g., reporting requirementsinternational regulatory

    Aggregation across an EconomicDimension

    (e.g., aggregate company statistics,national or international sectoral

    statistics)

    Fig. 4. Four aspects of ch

    M. OConnor, J.H. Spangenberg / Journaldifferent dimensions and scales. As our starting point, consideragain the EAA/C3ED standard set of CSR indicator categoriesas already shown in Table 1. For each standard CSR category,there can be a great variety of candidate indicators. TableA1 shows the spectrum of candidate indicators for each CSRcategory, and Table A2 the distribution of the suggestionsacross the six pilot sites and from the EAA.

    With reference to these tabular presentations, we may read-ily identify the Hits and Misses in the confrontation ofbottom-up and top-down perspectives. This examination illus-trates directly the operation of an interface between site-and company-specific aspects of CSR reporting, with concernsand frameworks that are generic to the European aluminiumsector.

    By Hits we refer to CSR indicator concepts or issues thatappear as widespread concerns across all sites and that havebeen suggested also by the EAA (see Box 2). It seems safe toconclude, on an empirical basis, that those indicator conceptsthat are commonly identified across all pilot sites and by theEAA can be considered as robust generic CSR reportingcategories.The Misses are, correspondingly, of two types. On theone hand, there are indicator concepts or issues that appearas widespread concerns across all sites but are not suggestedby the EAA. On the other hand, there are candidate indicatorsthat are suggested by the EAA but are not reflected in sugges-tions documented at the pilot sites.

    Making judgements about the pertinence of candidate indi-cators is less straightforward for the Misses than for theHits. In the course of our study, we reviewed in detailthe Misses identified from a bottom-up perspective,that is, suggestions for CSR reporting themes that arestrongly present at pilot sites but are not matched by EAAgeneric proposals (see Box 3). Four CSR indicator cate-gories are particularly noteworthy: Employee Opportunitiesand Relations and Community Relationships (in the so-cial dimension column of Table 1), Emissions and ImpactsBox 2. The CSR indicator Hits (common acrosssites and EAA)

    - Economic performance (code EC): the level ofwages (EC1-05), meaning both absolute andrelative wage levels.

    - Social performance (code S): the documenta-tion of accidents (S1-02) and occupationalhealth programmes (S1-01), of training oppor-tunities (S2-04), and of relations with the localcommunity (S4-01, S4-04).

    - Environmental performance (code EV): green-house gas emissions (EV3-05), energy and wa-ter resources use (in EV1), fluorine emissions(EV3-03, EV3-04) for smelter sites.

    - Institutional aspects (code IN): plant certifica-tion (IN1-05) and sustainability mission state-ment (IN2-08).

  • Box 3. Misses (suggestions from sites that arenot matched by EAA categories)

    1408 M. OConnor, J.H. Spangenberg / Journal of C(environment dimension) and Environmental ManagementSystem (institutional).9 On the basis of these various obser-vations (see the Phase Two report, OConnor et al. [10]),we made the following suggestions to the EAA for the defi-nition of generic reporting categories (see Table 3):

    - The categories of pay and benefits (EC2), internal commu-nications (S3), stakeholder dialogue more generally (withinS4 and IN3), noise (EV3-06) and site-related local trafficflows (EV3-07) should be considered by the EAA for adop-tion as generic CSR reporting categories.

    - Attention should be given to the adoption as reportingcategories, of toxic emissions control and local waste man-agement, whose information content will be site- and pro-cess-specific, as complements to categories such as wateruse, energy use and greenhouse gas emissions whose con-tent is comparable from site to site.

    - Economic performance (EC): site-level stake-holders attach high importance to pay andworker benefits (EC2-01, EC2-02, also S2-03).

    - Social performance (code S): stakeholders atsite-level highlight internal communications(S3-01, S3-02, S3-03) as well as benefits from in-ternal-external dialogue (IN3-01, IN3-02, IN3-03).

    - Environmental performance (EV): stake-holders at all sites highlight noise control(EV3-06) and local traffic flows including trucks(EV3-07). They also highlight process-specificemissions, health and waste management is-sues, e.g., toxic residues from paint.- Institutional aspects of CSR can usefully be signalled bydocumentation of formal actions such as plant certification(IN1-05) and sustainability mission statement (IN2-08), butattention should be given to communication of what thesemean in terms of visible actions and programmes at siteas well as company levels.

    In these guidelines, we placed the emphasis more on theCSR reporting categories to be retained than on the individualindicator(s) that might be used. This is because site-to-sitecomparisons cannot always be based on using identical oper-ational indicators, but they can be achieved in a meaningfulway by assuring that the same set of discursively establishedCSR issues or categories are addressed. For example, per-tinent indicators relating to the socialeinstitutional aspects ofinternal communications and stakeholder dialogue, and to the

    9 In the Phase Two full report (OConnor, et al. [10]) we discuss these

    individual cases in order to bring out the significance of the bottom-up/top--

    down dialogue process for seeking the appropriate balance of scales and gen-

    erality for indicator selection and interpretation.physicaletechnological aspects of noise, traffic flows, toxicemissions control and waste management will necessarilyvary from site to site. Insisting on fully standardised measure-ments would, overall, reduce the pertinence and quality of theinformation (maybe even in some cases producing nonsense).Nonetheless, these CSR reporting categories are important andrelate to tangible aspects of company social responsibilities ata local scale.

    Let us now consider briefly the Misses in the other direc-tion, that is, EAA candidate indicators that were not picked upat site-level. In Table 3 we list these misses, along with a briefdescription adapted from the EAA questionnaire of the in-tended coverage and purpose of these indicator categories.Then, in the right hand column of the table, we give a briefcomment about their absence from the site-level suggestions.

    Mostly these Misses relate to quantitative aspects of pro-duction activity such as production output levels, and to inter-national dimensions of materials supply. There are evidentlygood reasons, at the industry reporting level, for attention tothese information categories. Several different points arisefrom the Table 3 presentation. First of all, there are a numberof apparent misses between the EAA and site stakeholderssuggestions that are due to different ways of framing mattersof waste management. The indicators proposed by the EAAof spent pot lines (our code EV 3-28) and bauxite residue(EV 3-27) can be considered to fall under the general categoryvolume of treated wastes (EV 3-01) identified by stake-holders at all sites. The emissions of Benz(a)Pyrene (EV3-26) which are of significance in the primary manufacturingbut not for all plants in the aluminium industry, are an exampleof the more general category of toxic wastes emissions.

    There are also several aspects of Table 3 that highlight thespecificity of the EAA as a stakeholder having sector-wide co-ordination responsibilities at the interface with national and in-ternational regulatory authorities. This is the case notably forquantitative economic performance data such as taxes paid(EC 1-07), value added (EC 3-03), total revenue (EC 3-04)and total production (EC 4-01). Such financial and physicaldata is standard for almost all companies. Nonetheless, whiletotal revenue and total production can be broadly comparablefrom site to site, the figures for taxes paid and value added de-pend significantly on the accounting, fiscal and broader regu-latory framework within which the plant operates and so aredifficult to interpret for an outside observer and are compa-rable from site to site only within limits. Also, these data donot necessarily say much about burning issues of economicperformance, for example issues of the competitiveness ofEuropean industry in the new conditions of globalisation(including, e.g., labour costs and regulatory obligations, accessto raw materials).

    Finally, the EAAs list of CSR indicators is strongly influ-enced by an underlying Life Cycle Analysis (LCA) frame-work. The latter is powerful for obtaining insights intoprocess-related aspects of production and for giving an inte-grated picture of industry activities from raw materials exploi-

    leaner Production 16 (2008) 1399e1415tation (EV 2-01 bauxite availability and EV 2-02 minerehabilitation) to recycling (EV 4-02 end of life phase).

  • any.

    of CTable 3

    EAA candidate indicators not matched by pilot site suggestions

    Indicator

    code

    EAA label Short description

    (definition from EAA source)

    EC 1-07 Taxes paid Total amount of tax paid by plant/comp

    Includes direct taxes, property tax,

    energy tax and environmental taxes

    M. OConnor, J.H. Spangenberg / JournalSeveral of the EAAs indicators not picked up at site-level re-fer to the use of aluminium in society, including recycling.This is important for situating the aluminium sector in relationto other energy and resource intensive sectors (and so is an as-pect that could be taken up in future stakeholder dialoguework). So the EAAs system perspective has brought into focussome key high level and relatively generic indicators. At thesame time, the LCA framework can usefully be complementedby qualitative considerations about the significance to ex-tended community stakeholders of raw materials use, landuse and rehabilitation, and emissions management. It wastherefore recommended to the EAA that thought should be

    EC 3-03 Value added Economic value added is the difference

    between the total revenue for a company

    or plant, and the cost of bought in

    materials, services (excluding wages)

    and components

    EC 3-04 Total revenue Total revenue for company in year selected

    EC 4-01 Total production Total production in tonnes per year of

    alumina, primary, recycled aluminium,

    rolled products, extruded products and foil

    EV 2-01 Bauxite availability The duration of presently registered

    bauxite reserves according to present

    consumption rates

    EV 2-02 Mine rehabilitation Fraction of mined area (bauxite) which

    is rehabilitated back to its original state

    EV 3-26 Benz(a)Pyrene

    emissions

    Average emission of B(a)P from paste

    plants, anode plants and Soderberg

    primary smelters

    EV 3-27 Bauxite residue

    deposited

    The quantity of bauxite residue deposited

    on specially designated landfill site after

    separation and sand removal at alumina plan

    EV 3-28 SPL deposited Quantity of spent potlinings from electroly

    potrooms deposited after removal of materia

    for reuse and recycling. This includes both

    the carbon and the refractory parts.

    EV 4-01 Use Phase The weight of aluminium used in specific

    products, initially limited to cars, but could

    be expanded to other product sectors

    EV 4-02 End of life phase Data for aluminium recycling rates by

    product sectors, and descriptive statements

    on benefits of aluminium recycling

    EV 4-03 Life cycle aspects The relevance of the use phase in LCA

    studies for aluminium products

    EC 1-06 Aluminium use

    per capita

    Use of aluminium in Europe for all

    purposes in relation to populationRemarks in relation to perceptions at site-level

    The main focus of discussions was on local CSR performance

    with emphasis on community and environmental aspects and on

    expectations of and relations between stakeholders

    Quantitative economic performance data such as envisaged here,

    contributes to the CSR profile of any firm. However, some thought

    could be given to ways in which this sort of data on firms

    contributions to society can be made interesting and accessible

    1409leaner Production 16 (2008) 1399e1415given to ways in which physical production and financialdata can be translated and made interesting and accessibleto a wide public as indicators of firms contributions to society.For example, attention could be given to the significance thatstakeholders (as consumers and citizens) attach to the imageand uses of aluminium in society and its life cycle.

    8. Methodology for a representative diversity ofindicators

    A sample of six pilot sites is not a sufficient basis formaking definitive affirmations that this indicator is generic

    to a wide public

    Management and some workers expressed a clear knowledge of the

    up-stream environmental impacts of aluminium smelting. But the

    main focus of discussions was on local CSR performance

    This is a specific waste category of major significance in the primary

    manufacturing, but not for all plants in the aluminium industry

    t

    The main focus of discussions was on local CSR performance.

    However, attention to the ecological footprints of manufacturing

    activity and consumption is an increasing feature of environmental

    management

    sis

    ls

    This item is implicitly covered in the more sweeping category

    Volume of treated wastes (EV 3-01)

    These candidate indicators refer to the use of aluminium in society,

    including recycling. The discussion groups at the pilot sites did not

    enquire into participants perceptions (as consumers, as citizens)

    of the role of aluminium in society and its life cycle. However

    this is clearly important for situating the aluminium sector in

    relation to other energy and resource intensive sectors, and so

    is an aspect to take up in any future stakeholder dialogue work

  • And so we come to the recurrent question: how many indi-cators? In the Phase One EAA study (Faucheux et al. [9]), itwas remarked:

    Stakeholders indeed conveyed the view that, for reallearning, understanding and engagement to take place, it

    of Cleaner Production 16 (2008) 1399e1415and that one is not. For industry-wide procedures, the em-pirical process of interfacing has to be complemented by thedefinition of methodical steps. Therefore, our complete 3-DCSR deliberation matrix framework for CSR stakeholder dia-logues, indicator selection and reporting, as shown in Fig. 3were obtained by taking the site CSR Evaluation Matrix(as presented in Fig. 1) and envisaging a methodical compar-ison of CSR evaluation profiles between sites. It is within thisframework that we now address the question of procedures forgetting the indicator balance right, across all the differentdimensions and scales.

    Guideline 8.1: indicator selection for CSR evaluationreporting should be approached as a task of achievingand maintaining a dynamic balance between counter-vailing stakeholder concerns (including local and highlevel governance perspectives), across the spectrum ofperformance issues, and of generic versus site-specificcharacteristics.Guideline 8.2: the indicator mix for any site-level or sector-level CSR reporting process should respect a principle ofrepresentative diversity within the overall frameworkof the CSR deliberation matrix.

    The term representative diversity works at several levelsof the overall procedure. First of all, the notion of a represen-tative diversity is intrinsic in the structure of the CSR deliber-ation matrix itself.

    - A site-level CSR assessment is achieved, as we have out-lined in Section 4, by filling in cells of the deliberation ma-trix for the evaluation slice corresponding to the site inquestion.

    - A cross-site CSR comparative assessment, for selected cat-egories of performance issues and from the points of viewof selected stakeholder representatives, is achieved bycomparing the slices, which requires having filled thecells with data for each slice (site) of the three-dimensionaldeliberation matrix.

    In both these contexts, the array of cells is considered to beirreducible, meaning that users are not allowed to sandwichseveral stakeholder types together, nor to collapse several per-formance issues into a single index problem, nor to collapseseveral sites together.

    The CSR deliberation matrix framework (cf. Fig. 3) thusputs the emphasis on spanning the field of CSR consider-ations. The elaboration of the three dimensions of the CSRdeliberation matrix e multiple stakeholders, multiple issues,multiple sites e should assure that there is, visible to all con-cerned, within the evaluation of CSR performance, a repre-sentative diversity along the three deliberation matrix axes:across sites, across stakeholder types, and across performanceissues. Then, within a CSR indicator system or reportingprocess, the question may be posed of a balance in the num-

    1410 M. OConnor, J.H. Spangenberg / Journalber of indicators associated with each performance issue, witheach stakeholder type, for each site.is necessary to work with a well-balanced range of indica-tors, each one clearly appropriate to the specific aspect ofsocial responsibility being addressed. But not too many!Stakeholders in this study proposed limiting the indicatorset (social and environmental) to not more than about 30.Their view was that, beyond this number, the system wouldbe in danger of losing transparency and hence having nomeaning or relevance.

    To this pragmatic wisdom, we add the affirmation that theusefulness and manageability of a set of indicators depends asmuch on the structure and the procedure for mobilising the in-formation, as it does on the total number of indicators. Withthis in mind, we adopt as a rule-of-thumb:

    Guideline 8.3: evaluation involves comparisons but (formost people) loses quality if the number of items being com-pared at the same level exceeds a small number.10 Manage-ability can be enhanced through the judicious use of usehierarchical and crosscutting classifications in the designingof CSR evaluation frameworks and reporting procedures.

    The indicator dialogue box and the CSR deliberation matrixare, indeed, structuring frameworks making use of hierarchicaland crosscutting classifications that, by design, ensure that, atany moment, comparisons are being made between a smallnumber of items e a small number of indicators, a smallnumber of deliberation matrix cells, a small number of indi-vidual performance issues within each of the four sustainabil-ity dimension groups, and so on. This rule-of-thumb is ourstarting point for guidelines in response to the question,how many indicators should be associated with each site,with each performance issue (or group of performance issues),with each stakeholder type? Within the framework that we areproposing, the question has two main aspects.

    (1) At the level of the CSR deliberation matrix, the questionis, how many indicators can be put in the basket corre-sponding to a cell (for a selected issue, assessed bya specified stakeholder type, for a given site)?

    (2) At the level of the indicator dialogue box, there is thebackground question, how many candidate indicatorsshould be retained in the Indicator Kiosk, in reflection ofsite, stakeholder and performance issue diversity?

    10 We are mindful of work in psychology that suggests that seven is the max-

    imum number of ideas or objects that a typical person can entertain simulta-

    neously. So for us, a small number is about five and, if the number ofelements will significantly exceed five, then ways should be sought to intro-

    duce internal structure.

  • (head office, regulatory agency, sector association, etc.). A

    based on selection from local suggestions and the adaptationof pertinent indicator concepts from other sites including sec-

    1411of Cleaner Production 16 (2008) 1399e1415Conclusions about how many indicators will turn out to beabout right can be drawn only once experience has beengained with the proposed procedures in real situations. Inthe meantime, and as a reference point for applied work, wecan usefully consider the sorts of numbers that emerge duringan application of the CSR deliberation matrix framework.

    - In the present version of system design, there are 15 CSRperformance issue categories (grouped under the four di-mensions of sustainability) and four main stakeholder types(counting the coordination institutions). This means15 4 60 cells in the CSR evaluation matrix for a site.

    - Let us suppose that each stakeholder group is permitted, ina first phase of appraisal activity, to propose/select fromthe catalogue available in the Indicator Kiosk, up to threeindicators per performance issue. Then the maximumnumber of distinct information categories mobilised (ifthere is no redundancy) would, in principle, be 3 60180 (maximum).

    - In practice, this figure of 180 is likely to be higher than thenumber of indicators mobilised at a site, for two reasons.On the one hand, the same indicator may be mobilised(by one or more stakeholders) for more than one perfor-mance issue. On the other hand, the same indicator maybe selected (with application to one or more performanceissues) by two or three stakeholder categories.

    A workable reporting system would depend on reduction ofthe number of indicators well below this figure of 180. This iswhat has to be done through seeking out indicators that arerobust and generic across stakeholder groups. For exam-ple this might be achieved, at the site-level, by a second phaseof appraisal during which a group of stakeholder engages indeliberation to obtain a representative diversity within theconstraint of, say, a maximum of three indicators per perfor-mance issue. With 15 performance issues, this would yield15 3 45 indicators deployed for the CSR reporting at thesite-level. Probably, some indicators will be used across twoor more performance issues. So, we arrive through this proce-dure, not too far from the rule-of-thumb proposed by stake-holders, of about 30.

    Finally, let us suppose that prior stakeholder concentra-tion at site-level achieves a consensus for no more than threeindicators per CSR performance issue, including indicatorssuggested by the relevant higher-level coordination agencies.The principle of representative diversity could lead us to sug-gest that, within each basket of three indicators per perfor-mance issue, there should be not less than one (and notmore than two) site-specific indicator(s), and there shouldbe not less than one (and not more than two) generic (indus-try-wide) indicator(s).

    In sum, within the CSR deliberation matrix framework, weenvisage dialogues taking place, on the one hand, betweensites (e.g., for comparison, or for learning where one site isbenefiting from the others prior experience for the implemen-

    M. OConnor, J.H. Spangenberg / Journaltation of steps in the CSR evaluation and reporting) and, on theother hand, between a site and a higher-level reporting contexttor-wide indicator requirements. In this regard, we have putparamount emphasis on two points: (i) the standardisingthrough a discursive process of CSR performance categoriesand (ii) adherence to the principle of a representative diversityof indicators across stakeholder classes, issues and sites.

    Although our emphasis has been methodological, we do notforget that only limited resources can be devoted by compa-nies and regulatory agencies to CSR indicator system deploy-ment procedures, and these resource constraints must be takeninto account at site, company and sector levels. It is importantthat, as far as possible, new procedures are synergistic with ex-isting management and reporting activities. In this regard,some of the benefits of our proposed framework and proce-dures include:

    - A framework is provided for assessing and mobilisingexisting categories of information. Indeed a First Bestsituation can be imagined where a system of CSR reportingis based very largely on existing information categoriesavailable within the plant and in known external sources(local community, regulatory authorities, national or sec-tor-wide statistics).paramount consideration in this deliberation process is, asstated in Guideline 8.2 above, to maintain the representativediversity of indicators that signal the specificities of individ-ual stakeholder groups and performance issues. We might alsoexpress this as a principle of equitable stakeholder visibility.As in other forms of industrial bargaining or multi-stakeholdernegotiations, compromises can sometimes be agreed if honouris preserved.

    As already mentioned, firm conclusions about how manyindicators will be judged necessary or sufficient, must awaitexperience with applications of the proposed procedures.11

    The pilot site work of the EAA does, nonetheless, providesome preliminary corroboration. The double dialogue perspec-tive (site/site and site/regulatory authority) was, in effect, thecontext of the Phase Two stakeholder dialogues takingplace at the Grevenbroich pilot site in Germany. There, froma large list of nearly 250 indicator suggestions, team and stake-holders proposed a selection of 32 indicators, covering allreporting categories, for the companys future use in reporting,including to regulatory authorities (for details see Spangen-berg [22]). Of course, this result was obtained in the particularindustrial and regulatory context of the Grevenbroich site.

    9. Conclusions

    In this paper, a framework has been put forward for devel-oping operational CSR indicator systems on the site-level,11 Several studies are currently underway that, among other things, are

    putting the framework outlined in this paper to the test.

  • - A framework is provided for assessing if there is a bal-anced coverage of the full spectrum of performance issues,with adequate consideration to the diversity of stakeholderconcerns. As needs are identified, priority can be given to fill-ing notable gaps with selected new information via on-site

    agree about a representative diversity of indicators servingthis governance purpose.

    In conclusion, by our approach, stakeholders are consideredand treated as (physical or moral) persons aware of their ownobjectives and concerns and having autonomous standing rela-tive to the firm or corporate entity in question. Although theCSR performance issues and the individual indicators thatmay be suggested through discursive process will be of varying

    1412 M. OConnor, J.H. Spangenberg / Journal of Cleaner Production 16 (2008) 1399e1415measurement, company research or stakeholder dialogueprocesses.

    - The CSR evaluation framework and stakeholder dialogueprocedures are more robust than the individual informationcomponents. The overall framework remains powerful fororienting CSR work even if there are changes in detailedinformation sets (e.g., changes in the indicators availableor chosen to be mobilised for a cell in the deliberation ma-trix). This is an important design feature. Although continu-ity in information categories and, where possible, data sets,is an important consideration for indicator selection and de-ployment, there are situations where such continuity is im-possible to maintain or where new information categoriesmust be introduced.12

    - The multi-stakeholder dialogue framework is directlyadaptable for cross-sector comparisons. In methodologicalterms, a comparison between sectors is analogous tothe comparison between sites.

    The overall procedure of CSR indicator selection and re-porting, if built as a multi-stakeholder dialogue or reconcili-ation procedure, can be effective for the communication ofmeaning and purpose with a relatively small number of indica-tors and therefore low implementation costs. A shared struc-ture of CSR reporting, permitting identification of issues thatare common across all or most sites (and that can be consideredas generic issues) is, in our view, more important than gettingcomplete industry-wide agreement on the specific indicators tobe deployed.

    By allowing stakeholders (including company management)with local and often site specific preoccupations, to see howtheir particular concerns fall within the categories of social re-sponsibility and issues addressed by the wider internationalcommunity, the local significance of proposed indicators isclearly in view of the local stakeholders and, at the sametime, they can appreciate the role of the indicator systems asa contribution to wider societal objectives of economic, envi-ronmental and social dimensions of sustainability. Moreover,when it is possible to identify reasons why a particular indicatoror exploitation of indicator information is advocatedewhetherin a local situation or for a generic reporting procedure for com-panies and sites across Europe e it is easier for stakeholders tosee the primary role of CSR indicators as signalling the gover-nance framework as a wholee that is, a framework of attention,observation, communication and accountability e and then to

    12 As circumstances change, with new technologies, modifications to produc-

    tion, sales and personnel profiles (etc.), the individual indicators will necessar-

    ily have to be reassessed and sometimes changed. A benefit of the new

    multimedia technologies for information management is that variations inthe information sets (e.g., from year to year), whatever their causes, can easily

    be recorded and communicated.scope and of widely varying quality regarding data availability,controllability and possibility for governance, they are affirmedin the deliberative framework as of equal standing, in the sensethat they represent comparable levels of interest of the differentstakeholder groups which have suggested them.13 The diversityof candidate indicators resulting from the stakeholder work-shops, organised pragmatically in terms of a set of CSR perfor-mance issues, gives rise to a sort of patchwork CSR goalwhich is an amalgam of generic sustainability concerns andspecific issues of the firm and its stakeholders. This patch-work character, while associated with a rather uneven levelof detail and governability in a purely technical sense, is botha product of and a necessary condition for stakeholderresonance e inside and outside of the perimeter of companyoperations e which, due to links with acceptability, confidenceand legitimacy, is a key consideration for success in this kind ofsetting.

    Acknowledgements

    The work reported in this paper was funded by and carriedout with the support of the European Aluminium Association(EAA), but the study was in no sense an audit or evaluationof the individual plants and companies concerned, nor of theEAA. Our thanks are due to all the stakeholders, includingcompany management, employees and members of the localcommunities, for their invaluable contributions. We also ac-knowledge the contributions of the staff of the EAA and thestudys steering committee members, particularly to LauraBerneri (Communications Manager, EAA) and Nicolas Bouv-ier (Communication and Institutions. Thanks to our colleagueswithin the organisations making up the study team, includingSylvie Faucheux, Christelle Hue, Lisa Booth, Marie-FrancoiseGuyonnaud and Stephanie Chatenay (C3ED), Anke Valentin(SERI), Bruna De Marchi, Luigi Pellizzoni and Maura DelZotto (ISIG, Italy). Responsibility for any errors and for thearguments and recommendations put forward remain exclu-sively with the authors.

    13 It is perhaps obvious to some readers that the emphasis on stakeholders

    autonomy in defining the themes they wish to be informed about also safe-

    guards against attempts of greenwash or window dressing, as choice of in-

    dicators for reporting is not exclusively in the hands of the management. In

    cases where the stakeholders came forward with themes, but not indicators,

    the project team has helped with indicator suggestion based upon the GRI

    or WBCSD indicator sets. However, the crucial condition is still the accep-

    tance by the stakeholders who can be supported but not replaced by scientific

    expertise. In this setting, pre-defined indicator sets can undoubtedly be a sourceof inspiration, but are not suitable as ready to wear reporting or evaluation

    tools.

  • Appendix.Table A1. Summary list of stakeholders indicator suggestions

    Economic performance (code EC)

    Competitiveness

    EC 1-01: innovation investment

    EC 1-02: dependence on customers, suppliers, markets, ex. rates

    EC 1-03: average energy cost

    EC 1-04: risk of failure of customers, suppliers, machinery, labour

    EC 1-05: level of wages in the aluminium companies

    EC 1-06: aluminium use per capita (EAA)

    EC 1-07: taxes paid (EAA)

    Pay and benefitsEC 2-01: fringe benefits, economic and others offered to employees

    EC 2-02: equity between employees for wages and profit shares

    Revenues and payments

    EC 3-01: profitability

    EC 3-02: productivity

    EC 3-03: value added (EAA)

    EC 3-04: total revenue (EAA)

    Production (physical)

    EC 4-01: total production (EAA)

    Social performance (code S)

    Working conditions/health and safety

    S 1-01: occupational and lifestyle health programmes

    S 1-02: records of accidents

    S 1-03: means developed within the company to prevent exclusion

    S 1-04: turnover and absenteeism rates

    S 1-05: time/output lost in strikes

    S 1-06: difficulty to recruit adequate people

    S 1-07: ratio of workforce to yearly output tonnage.

    Labour productivity

    Employee opportunities and relationsS 2-01: gender balance

    S 2-02: equity between employees/anti-union bias

    S 2-03: equity of wages between firms in the same region

    S 2-04: training programmes for the employees

    S 2-05: workers would like to have contacts with the other sites

    S 2-06: problems related to cultural differences

    S 2-07: responsibility of the firm towards the employees

    S 2-08: professional development. Lack of career succession planning

    S 2-09: staff number/job security

    Internal communications

    S 3-01: diffusion of information for employees

    S 3-02: information is not communicated in the local language

    S 3-03: dialogue with the management

    Community relationshipsS 4-01: contribution of the firm to the community

    S 4-02: employee involvement into the local community

    S 4-03: employment of local population

    S 4-04: communication/dialogue with the local actors

    S 4-05: turnover of management

    S 4-06: information exchange among aluminium companies

    S 4-07: origins of workers

    S 4-08: company involvement outside the region

    S 4-09: number of mergers/acquisitions

    Environmental performance (code EV)

    Resource use (National/European)

    EV 1-01: energy consumption

    EV 1-02: water consumption

    EV 1-03: energy production from renewable sources

    EV 1-04: pure water consumption

    Table A1 (continued)

    Resource use e global (international exchange)

    EV 2-01: bauxite availability (EAA)

    EV 2-02: mine rehabilitation (EAA)

    Emissions and impacts

    EV 3-01: volume of treated wastes

    EV 3-02: means devoted to handle wastes

    EV 3-03: fluorine emissions

    EV 3-04: fluoride emissions harming local population

    EV 3-05: carbon dioxide emissions

    EV 3-06: noise levels

    EV 3-07: road traffic, number of trucks per day

    EV 3-08: dust emissions

    EV 3-09: local community complaints about bad smells

    EV 3-10: NO2 emissions

    EV 3-11: level of SO2 emissions in the area

    EV 3-12: waste water emissions

    EV 3-13: reduction of heavy metal emissions

    EV 3-14: disposal of solvents contained in paints

    EV 3-15: level of dioxin emissions

    EV 3-16: PCBs releases

    EV 3-17: subsidies paid to the affected populations

    EV 3-18: type of fuel used

    EV 3-19: storage of hazardous substances

    EV 3-20: accidents linked to handling of hazardous substances

    EV 3-21: expenditure in insurance

    EV 3-22: land use

    EV 3-23: recultivation of area used

    EV 3-24: recycling quota

    EV 3-25: material consumption wood

    EV 3-26: benz(a)pyrene (BaP) emissions (EAA)

    EV 3-27: bauxite residue deposited (EAA)

    EV 3-28: spent pot line (SPL) deposited (EAA)

    Product use (life cycle)

    EV 4-01: use phase (EAA)

    EV 4-02: end of life phase (EAA)

    EV 4-03: life cycle aspects (EAA)

    Institutional aspects (code IN)

    Environmental management system

    IN 1-01: objectives anticipating the regulations

    IN 1-02: means devoted to handle environmental problems

    IN 1-03: environmental impact of the firm. Historical liability

    IN 1-04: infractions to regulations

    IN 1-05: plant certification

    IN 1-06: road-rail transport/adequate rail infrastructure

    IN 1-07: incentives for compensating initiatives and innovations

    IN 1-08: link between innovation and incident reduction

    Company CSR strategy/policy

    IN 1-01: Voluntary objectives fixed by the site or

    the company

    IN 2-02: formalization of corporate social responsibility

    IN 2-03: observance to statutory obligations

    IN 2-04: tools used for measuring performances

    IN 2-05: number of hierarchical levels

    IN 2-06: installations must be designed to increase safety

    IN 2-07: investment in environment and H&S

    IN 2-08: sustainability mission statement (EAA)

    Supply chain relationships

    IN 3-01: dialogue with suppliers and members of the supply chain

    IN 3-02: equitable treatment of employees of the firm

    and of suppliers

    IN 3-03: selection criteria for choosing suppliers S 4-03

    1413M. OConnor, J.H. Spangenberg / Journal of Cleaner Production 16 (2008) 1399e1415

  • Table A2

    The distribution of stakeholders indicator suggestions across sites

    Pech

    (F)

    Alcoa

    (F)

    CFF

    (F)

    Grev

    (Deut)

    Metra

    (Italy)

    Alcan

    (UK)

    EAA

    EC 1-01 - CC

    EC 1-02 -

    EC 1-03 -

    EC 1-04 -

    EC 1-05 - - -

    EC 1-06 CCEC 1-07 CC

    EC 2-01 - - - -

    EC 2-02 - -

    EC 3-01 -

    EC 3-02 -

    EC 3-03 CC

    EC 3-04 CCEC 4-01 CC

    S 1-01 - - - - CC

    S 1-02 - - - - - - CCS 1-03 - CC

    S 1-04 - - -

    S 1-05 - - -

    S 1-06 -

    S 1-07 -

    S 2-01 - - - - -

    S 2-02 - -

    S 2-03 - - - - -

    S 2-04 - - - - - - CC

    S 2-05 - - - -

    S 2-06 -

    S 2-07 - -

    S 2-08 - - -

    S 2-09 - CC

    S 3-01 - - - - - -

    S 3-02 - -

    S 3-03 - - - - - -

    S 4-01 - - - - - CC

    S 4-02 - - - -

    S 4-03 - - - - -

    S 4-04 - - - - - - CC

    S 4-05 - -

    S 4-06 -

    S 4-07 -

    S 4-08 -

    EV 1-01 - - - CCEV 1-02 - - -

    EV 1-03 - - CC

    EV 1-04 - - CC

    EV 2-01 CCEV 2-02 CC

    EV 3-01 - - - - - -

    EV 3-02 - -

    EV 3-03 - -

    EV 3-04 - - - CC

    EV 3-05 - - - - - CC

    EV 3-06 - - - - -

    EV 3-07 - - - - - -

    EV 3-08 - - -

    EV 3-09 - -

    EV 3-10 - -

    EV 3-11 -

    EV 3-12 - - - CC

    EV 3-13 - -

    EV 3-14 -

    EV 3-15 -

    1414 M. OConnor, J.H. Spangenberg / Journal of CTable A2 (continued )

    Pech

    (F)

    Alcoa

    (F)

    CFF

    (F)

    Grev

    (Deut)

    Metra

    (Italy)

    Alcan

    (UK)

    EAA

    EV 3-16 -

    EV 3-17 -

    EV 3-18 -

    EV 3-19 -

    EV 3-20 -

    EV 3-21 -

    EV 3-22 -

    EV 3-23 -

    EV 3-24 -

    EV 3-25 -

    EV 3-26 CCEV 3-27 CC

    EV 3-28 CC

    EV 4-01 CC

    EV 4-02 CCEV 4-03 - CC

    leaner Production 16 (2008) 1399e1415References

    [1] De Marchi B. Seveso: from pollution to regulation. Int J Environ Pollut

    1997;7(4):526e37.[2] GRI (Global Reporting Initiative). Sustainability report guidelines.

    Boston: GRI; 2002.

    [3] WBCSD (World Business Council for Sustainable Development). Corpo-

    rate social responsibility: making good business sense. Geneva: WBCSD;

    2000.

    [4] De Marchi B, Funtowicz S, Guimar~aes Pereira A. From the right to be

    informed to the right to participate: responding to the evolution of

    European legislation with ICT. Int J Environ Pollut 2001;15(1):1e21.[5] Faucheux S, Nicola I. From sustainable development to corporate social

    responsibility: an application to European aluminium sector. Int J Sustain

    Dev 2003;6(2):155e69.[6] Le Dars A. Pour une Gestion Durable des Dechets Nucleaires. Paris:

    Presses Universitaires de France; 2004.

    [7] Faucheux S, Hue C. From irreversibility to participation: towards a partic-

    ipatory foresight for the governance of collective environmental risks.

    J Hazard Mat 2001;86:223e43.

    [8] ECC Kohtes Klewes GmbH, Fishburn Hedges. Global stakeholder report

    2003. Shared values? Bonn/London: ECC/Fishburn Hedges; 2003.

    IN 1-01 - - -

    IN 1-02 - - - - -

    IN 1-03 - - - -

    IN 1-04 -

    IN 1-05 - - - - CC

    IN 1-06 - - -

    IN 1-07 - - - - -

    IN 1-08 -

    IN 2-01 - - - -

    IN 2-02 - -

    IN 2-03 - - - -

    IN 2-04 - -

    IN 2-05 -

    IN 2-06 -

    IN 2-07 -

    IN 2-08 CC

    IN 3-01 - - -

    IN 3-02 - - -

    IN 3-03 - - - -

    - One or more indicator suggestions were received from stakeholders at the

    site.

    CC The indicator concepts used by the EAA in their 2003 survey of alumin-

    ium firms.

  • [9] Faucheux S, Hue C, Nicola I, OConnor M. Integration of the socialdimension of sustainable development in enterprise strategies within

    the aluminium industry, full final research report (Phase one). Guyan-

    court: Universite de Versailles St-Quentin-en-Yvelines; 2002.

    [10] OConnor M, Booth L, De Marchi B, Hue C, Spangenberg JH,

    Valentin A. Implementation of a system of indicators for social respon-

    sibility reporting, full final research report (Phase Two). Guyancourt:

    Universite de Versailles St-Quentin-en-Yvelines; 2004.

    [11] Faucheux S, Nicola I. La Responsabilite Societale dans la Construction

    dIndicateurs: lExperience de lIndustrie Europeenne de lAluminium.

    Natures Sciences Societes 2004;12:30e41.

    [12] Faucheux S, Nicola I. Quels Indicateurs pour la ResponsabiliteSociale des Entreprises? Une application a` lindustrie europeenne

    de laluminium. Annales des Mines: Gerer & Comprendre 2004;76:

    42e54.

    [13] Faucheux S, Hue C, OConnor M. A bottom-up/top-down methodology

    for indicators of corporate social performance in the European alumin-

    ium industry, Cahiers du C3ED. Guyancourt: Universite de Versailles

    St-Quentin-en-Yvelines; 2005.

    [14] OConnor M, Frame B. Signs are taken for wonders: some thoughts on

    reconciling the diversity of accounting frameworks, valuation metrics,

    stakeholder positionings and knowledge quality considerations in assess-

    ments for sustainability, Cahiers du C3ED. Guyancourt: Universite de

    Versailles St-Quentin-en-Yvelines; 2007.

    [15] OConnor M. Social costs and sustainability. In: Bromley DH, Paavola J,

    editors. Economics, ethics and environmental policy: contested choices.

    Oxford: Blackwell Publishing; 2002. p. 181e202.

    [16] Spangenberg JH. Investing in sustainable development. Int.J Sustain

    Dev 2001;4(2):184e201.[17] Spangenberg JH. A systemic approach to institutional sustainability. To-

    wards institutional sustainability performance indicators. In: IIG Institut

    Internacional de Governabilitat, editor. Governance for sustainable devel-

    opment. Barcelona: IIG; 2002. p. 87e106.

    [18] OConnor M. The KerBabel indicator dialogue box: generic design

    specifications for the Indicator Dialogue Box e Version 3, Rapport

    de Recherche du C3ED, Guyancourt: Universite de Versailles St-Quen-

    tin-en-Yvelines; 2004.

    [19] OConnor M. The four spheres fraremework for sustainability. Eco Com

    2006;3(4):285e92.

    [20] OConnor M. Indicators and deliberation: knowledge quality, social

    choices and environmental learning through multi-mediated multi-stake-

    holder dialogues. Int J Sustain Dev 2008;11, in press.

    [21] EAA (European Aluminium Association). Sustainable development in-

    dicators: list of indicators to be used for the first EAA data gathering

    exercise and definitions (Revision 4, 3.12.2003). Brussels: EAA; 2003.

    [22] Spangenberg JH. Discourse derived indicators: 2nd order governance to

    solve 2nd order problems of corporate responsibility. Corp Soc Respon-

    sib Environ Mgmt, in press, doi:10.1002/csi 137.

    1415M. OConnor, J.H. Spangenberg / Journal of Cleaner Production 16 (2008) 1399e1415

    A methodology for CSR reporting: assuring a representative diversity of indicators across stakeholders, scales, sites and performance issuesIntroductionDiscourse-derived CSR information and communication goalsProfiling the candidate indicators for CSR reportingA matrix framework for stakeholder dialoguesBuilding the dialogues across sitesThe question of scalesGetting balance in indicator selection: the EAA caseMethodology for a representative diversity of indicatorsConclusionsAcknowledgementsReferences