Download - Beattie 2007

Accounting Forum 31 (2007) 129163

Lifting the lid on the use of content analysis toinvestigate intellectual capital disclosures

Vivien Beattie a, Sarah Jane Thomson b,a Department of Accounting and Finance, University of Glasgow, 65-73 Southpark Avenue,

Glasgow G12 8LE, UKb Department of Accountancy and Finance, School of Management and Languages,

Heriot-Watt University, Edinburgh EH14 4AS, UK

Abstract

This methods paper highlights specific issues that arise in using content analysis to investigate intellectualcapital (IC) disclosures. The use of content analysis in the IC context is debated through an analysis of priorstudies and the use of an illustrative example (Next plcs 2004 annual report). It is concluded that the depthand breadth of the IC concept and the lack of common definitive language make it difficult to establish theextent and nature of disclosure currently provided. The range of choices available to researchers in termsof analysing and measuring IC disclosures further hinders interpretation and comparability. Transparencyin the choices made is required. Shared meanings could be developed and the IC concept better understoodthrough increased transparency in the categorisation of IC information, which in turn could further assist inthe interpretation and comparison of findings across studies. 2007 Elsevier Ltd. All rights reserved.

Keywords: Intellectual capital; Corporate disclosure; Methods; Content analysis

1. Introduction

Content analysis has become a widely used method of analysis in financial accounting research(Beattie, 2005). In recent years, several papers in accounting journals have identified and discussedsignificant issues regarding the use of content analysis to investigate accounting disclosures. Onestrand of this literature takes corporate social reporting (CSR) as its context (i.e. Hackston &Milne, 1996; Milne & Adler, 1999; Unerman, 2000). More recently, the topic area of intellectualcapital (IC) disclosures has been explored (Abeysekera, 2006; Guthrie, Petty, Yongvanich, &Ricceri, 2004). The present paper contributes to the latter area of enquiry.

Corresponding author. Tel.: +44 131 451 3559; fax: +44 131 451 3296.E-mail addresses: [email protected] (V. Beattie), [email protected] (S.J. Thomson).

0155-9982/$ see front matter 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.accfor.2007.02.001

130 V. Beattie, S.J. Thomson / Accounting Forum 31 (2007) 129163

IC is the term attributed to intangible assets which create company value (Mouritsen, Larsen,& Bukh, 2001). It is, at least in part,1 reflected in the difference between market and book values,as the value and impact of intangibles are inadequately reflected in the traditional accountingframework (Cordon, 1998). To highlight the potential significance of IC, studies have reportedmarket-to-book multiples in excess of unity. For example, Gu and Lev (2004) report that theS&P 500s average market-to-book ratio was 4.5 in September 2003 indicating for every US$ 4.5of market value, only US$ 1 appears on the balance sheet. Beattie and Thomson (2005) foundthe mean market-to-book value for the UK FTSE 100 companies to be 2.52 based on data foryear-end 2002/2003. In light of this evidence, a method for reporting IC information to externalstakeholders appears to be required.

The term IC is now widely used among regulators, professional bodies and academics. Manyattempts have been made at formal definition. However, according to Guthrie, Petty, & Johanson(2001), intellectual capital frequently is poorly defined or is not defined at all. Zambon (2005)has stated that a generally agreed taxonomy is needed. Despite this apparent stumbling block,considerable efforts have been made to develop models for IC reporting (e.g. DATI, 2000, 2002;DMSTI, 2003; Edvinsson & Malone, 1997; Lev, 2001; Sveiby, 1997). Suggestions have been madeto extend the balance sheet to integrate IC, or to create complementary balance sheets (Rylander,Jacobsen, & Roos, 2000). Recently, a focused narrative-based approach to IC reporting has beenproposed (DATI, 2000, 2002). However, the opportunity to report IC in narrative format alreadyexists within corporate annual reports.

Corporate annual report narratives may provide the opportunity for IC reporting, but what aboutthe incentive to do so? Voluntary disclosure of IC information can be explained in terms of theoriessuch as positive accounting theory (PAT), legitimacy theory and stakeholder theory (Deegan,2000; Deegan & Gordon, 1996). If company managers interests are aligned with shareholders,IC information will be disclosed if it brings benefits to the company (PAT). IC reporting providescompanies with the opportunity to take advantage of increased transparency to capital markets,establishing trustworthiness with stakeholders and to employ a valuable marketing tool (Vander Meer-Kooistra and Zijlstra, 2001). Disclosure of IC information could be self perpetuatingin terms of maintaining and enhancing IC value given that intangible asset creation occursthrough enhanced reputation and disclosure influences the external perception of reputation(Toms, 2002, p. 258). However, reluctance to report IC information may arise from fear of bothloss of competitive advantage and litigation.2 Companies may disclose IC information to appearlegitimate in the eyes of society and avoid the imposition of costs arising from non-legitimacy. Thedisclosure choices of comparable companies may shape legitimacy. IC disclosure may respondto the demands of the stakeholders most critical to the companys ongoing survival (managerialbranch of the stakeholder theory).

These theories are mutually consistent, IC disclosure being explained in terms of a cost-benefittrade-off. The ethical branch of the stakeholder theory appears to offer an alternative explanation.Companies recognise that different stakeholders have a right to IC information and so disclosure isresponsibility-driven. However, executing responsibilities in terms of disclosure is not necessarily

1 The difference between market and book values can result from other factors such as the undervaluation of tangibleand financial assets recognised in the balance sheet, intangible liabilities that are not captured in the balance sheet andmarket prices that do not accurately capture intrinsic value (Garca-Ayuso, 2003).

2 Elliot and Jacobson (1994) argue that increased informative disclosure actually decreases litigation costs as a resultof fewer allegations of insufficient disclosure. It also decreases litigation costs arising from allegations of misleadingdisclosures through smaller claims, better defences and fewer law suits.

V. Beattie, S.J. Thomson / Accounting Forum 31 (2007) 129163 131

incongruent with increasing firm value. Another proposition (suggested by Miller, 1977, in thecontext of capital structure decisions) is neutral mutation. Companies fall into disclosure patternsor habits which have no material effect on firm value.

Given these theoretical explanations for disclosing (not disclosing) IC information, what iscorporate practice? The disclosure of IC information in annual reports is beginning to be investi-gated using content analysis (e.g. Bozzolan, Favotto, & Ricceri, 2003; Brennan, 2001; Guthrie &Petty, 2000). (Other accounting-related documents that have been studied are IPO prospectuses,e.g. Bukh et al., 2003, presentations to analysts, e.g. Garca-Meca, Parra, Larran, & Martnez,2005, and analyst reports, e.g. Arvidsson, 2003.) This type of investigation could potentially servetwo purposes. First, to measure the extent to which different categories of IC information are dis-closed. Second, IC reporting in practice provides valuable examples of attempts to understand andcapture the IC concept (Van der Meer-Kooistra and Zijlstra, 2001). Practical experiences wouldassist in the development of a generally agreed taxonomy of IC terms, as called for by Zambon(2005).

To date, content analysis appears to have been mainly used with the aim of quantifying thenumber of IC disclosures, typically in relation to 22-25 categories of IC information. The observedlevel of IC disclosure has consistently been described as low, and this has been attributed tothe lack of an established IC reporting framework and the general lack of a proactive stanceby companies in attempting to measure and externally report IC information (Guthrie & Petty,2000). However, the lack of an established IC reporting framework hinders not only the companiesdisclosing information. The depth and breadth of the IC concept evident in the academic literature,and the subjectivity involved in constructing an operational IC definition, could also be said tohinder researchers aiming to quantify IC disclosures. In this context, it is essential that the precisedetails of the content analysis method used are transparent, to allow findings to be interpretedand to make comparisons (or not) across studies. Transparency is important because the contentanalysis method used to investigate disclosures is reflective of the researchers conception ofreality (Gray, Kouhy, & Lavers, 1995) what the researchers perceive constitutes IC rather thanany potential objective reality which exists in relation to the IC concept. Despite this importance,a general lack of transparency in the content analysis methods used in the IC disclosure studiesto date is apparent. Increased transparency in relation to the IC information found and how itis categorised would also clarify researchers understanding of the IC concept and assist in thedevelopment of shared meanings.

This need for transparency and the development of shared meanings has already been recog-nised by researchers in the CSR context. As noted in Gray et al. (1995, p. 85), the use of contentanalysis either demands, or at a minimum implies strongly, that the categories of analysis arederived by reference to shared meanings and that the data collection and analysis must be replica-ble. In their construction of a research database of social and environmental reporting, categoriesof disclosures and examples of types of information relating to the categories is provided. In doingso, the result is at least transparent and replicable, even if it fails to meet an ideal of a fixed andperfect definition (p. 82).

The aim of the present paper is to highlight specific issues that arise in using content analysisto investigate the extent of IC disclosures. The use of content analysis in this context is debatedthrough an analysis of prior studies and the use of an illustrative example (Next plcs 2004 annualreport). The present study responds to Abeysekeras (2006) suggestions that coding frameworksused to analyse annual reports need to be critically analysed, and the real problems of comparabilitybetween IC disclosure studies need to be addressed. He calls for the operational issues arisingfrom the use of content analysis research methods to investigate IC disclosures to be resolved.


The present paper highlights and illustrates these issues. In doing so, it aims to contribute to thedebate and help build a more secure foundation for future work.

The remainder of this paper is structured as follows. Section 2 considers the problems asso-ciated with defining the IC concept. Section 3 offers a review of extant content-analytic studiesof IC disclosure in corporate annual reports. Section 4 documents the use of Next plcs 2004annual report to illustrate the use of content analysis to investigate IC disclosures. Summary andconclusions are offered in Section 5.

2. Dening the IC concept

Many writers observe that there is no consensus on a precise definition of IC (e.g. Marr,Schiuma, & Neely, 2004, p. 314). Furthermore, the terms intangibles and IC are frequently usedinterchangeably.3 For example, Rylander et al. (2000, p. 716), Meritum (2002), Lev (2001) and Levand Zambon (2003) all explicitly state that both terms are used synonymously. Other studies offerdefinitions of intangibles that coincide with common definitions of IC. For example, FASB (2001,p. vi) state that [i]ntangibles include not only those resulting from research and development butalso human resources, customer relationships, innovations and others. Similarly, Gu and Lev(2004, p. 1) define intangible assets as R&D, software, brand enhancement, employee training,and the development of unique organizational designs and processes (organizational capital). Ithas been argued that the precise terms used are associated with different disciplines, intangiblesbeing an accounting term and IC being a term used in the management/human resource field(Chaminade & Roberts, 2003, pp. 736737). Traditional accounting definitions of intangibles arenarrow, including items such as intellectual property and patents that meet the criteria for balancesheet recognition.

Despite the lack of an agreed definition of IC, it is argued (see, for example, Lev & Zambon,2003, p. 603) that a broad consensus exists that IC comprises three major categories: humancapital, structural capital and relational capital (the latter two elements exist at the organisa-tional rather than the individual level). The definitions presented in the Meritum Report (2002),an influential report, are shown in Table 1. It is also increasingly accepted that synergies existin operating these categories of IC together, creating a fourth IC element, termed connectiv-ity capital (see Habersam and Piber, 2003, who empirically identify connectivity capital as alinking pin).

Although many authors adopt this set of three IC categories, there is evidence of synonymousterms being used. For example, employees and employee competence are sometimes used inplace of human capital. Some writers use the terms internal capital or organisational capitalto refer to structural capital. Relational capital includes relationships with customers and othergroups external to the firm, also referred to as external structures. In consequence, relationalcapital is also referred to as customer capital or external capital. The words structure(s) andcapital appear to be interchangeable in some situations.

Most writers perceive the IC construct as a hierarchy of nested concepts, with high-level cat-egories such as human, structural and relational capital having multiple lower-level categories.A total of 128 lower level IC terms, identified from a detailed content analysis of the IC litera-ture, are shown in Table 2 (Beattie & Thomson, 2004). Four observations can be made regardingthis table. First, although many terms might be judged synonymous, this list suggests that IC is

3Knowledge assets is the term often used in place of IC by economists (Lev, 2001).


Table 1Classification and definition of intellectual capital

Category of IC

Human The knowledge that employees take with them when they leave the firm. Includes the knowledge,skills, experiences and abilities of people. Some of this knowledge is unique to the individual,some may be generic

Structural The knowledge that stays within the firm at the end of the working day. Comprises theorganisational routines, procedures, systems, cultures, databases, etc. Some may be legallyprotected and become Intellectual Property Rights, legally owned by the firm under separate title

Relational All resources linked to the external relationships of the firm, with customers, suppliers or R&Dpartners. Comprises that part of human and structural capital involved with the companysrelations with stakeholders (investors, creditors, customers, suppliers, etc.) plus the perceptionsthat they hold about the company

Source: Meritum (2002, p. 63).

a broad concept. Second, there also appears to be a boundary problem between the high-levelcategories, in that some lower-level categories are placed under different high-level categoriesby different writers. For example, brands and distribution channels frequently appear underrelational/external/customer capital (e.g. Bozzolan et al., 2003; Brennan, 2001; Guthrie & Petty,2000), but were placed by Rodgers (2003) under structural/internal/organisational capital). Thismatters if these high-level categories have functions. Third, a distinction needs to be made (butoften is not) between lower level IC categories and indicators (i.e. measures) relating to them. Forexample, staff profile is a lower level category, whereas number of employees is an indicator.Fourth, the meaning to be attached to some lower level categories (e.g. innovation) cannot beestablished without knowledge of context. Further, there is the potential for confusion arisingfrom a variation in the terms used to describe this hierarchy of nested concepts. For example,some writers refer to human, structural and relational capital as components of IC rather thancategories (e.g. Van der Meer-Kooistra and Zijlstra, 2001). Some writers refer to both com-ponents and categories (e.g. Guthrie, Petty, & Ricceri, 2006). Lower-level categories are alsoreferred to as IC items (e.g. Abeysekera & Guthrie, 2005), attributes (e.g. Guthrie & Petty,2000), or variables (e.g. Guthrie et al., 2006). For the remainder of this paper, in the IC context,the term category refers to human, structural and relational capital. The term sub-category is usedto refer to the different categories of information within each of the high level human, structuraland relational capital categories.

The Meritum Report goes on to distinguish between resources (static) and activities (dynamic).The resource concept is the stock or value of a given intangible at a point in time. The activityconcept implies an allocation of resources aimed at creating or acquiring new intangibles, increas-ing the value of an existing one or evaluating and monitoring the results of intangible activities.Rylander et al. (2000, p. 737) describe flows as the transformations between and within stocksof human, structural, physical and financial capital.

As the IC literature has developed, and practitioners and researchers have struggled to find waysof managing and reporting on IC, there appears to have been an increasing amount of discussion ofrelated concepts such as strategy, value drivers, critical success factors and value creation (Bukh,2003; Rylander et al., 2000). The boundary around the IC construct is not clear (Mouritsen, 2003)and the link between the corporate value creation process and strategy suggests situation-specificIC components.


Table 2128 lower level IC categories identified from prior literature

Human capital Structural capital Relational Capital

Absence Achieving mechanism culture Basic marketing capabilityAdaptability Administrative processes BrandsAttitudes Brands Business collaborationsCapability/abilities Communication systems Client profileCommitment Competitive and market channels CollaborationCommunicative abilities Copyrights Commercial powerCompetence Corporate/organisational culture Competitive intelligenceComputer literacy Cultural diversity CompetitorsCreativity Culture ConnectivityDevelopment Customer support Customer knowledgeEducation Customer-centered Customer loyaltyEmployee expertise Databases Customer namesEmployee flexibility Distribution channels Customer reputationEmployee knowledge Documentation services Customer satisfactionEmployee productivity Financial relations CustomersEmployee satisfaction Infrastructure DiffusionEmployee value Innovation Distribution channelsEmployees Intellectual property Environmental activitiesEntrepreneurial spirit Intellectual resources Favourable contractsEquality Knowledge centre Financial contractsExpert networks Knowledge-based infrastructure Franchising agreementsExpert teams Laboratories ImageFriendliness Management philosophy IntensityFurther personal/professional training Management processes Knowledge/acquaintance with communityHuman assets Operation process Knowledge/acquaintance with governmentHuman resources Organisational flexibility Knowledge/acquaintance with suppliersHuman value Organisational learning Licensing agreementsIdentification Organisational routines Links with suppliersInnovation Organisational structure Market intensityInnovative capacity Patents Negotiating capacity with financial entitiesJuristic competence Procedures NetworkingKnow-how (employees) Process capability New strategic customersLearning capacity Quality improvements ReputationLoyalty to organisation Quality management Research collaborationsMotivation Research projects StakeholdersPerceptions Specialised software/IT Supplier knowledgePersonal/professional experience Systems (information/network)Personal ability TrademarksPersonnelRecruitmentReflect experiences (previous)SensitivitySkill (employees)Social competenceStaff (employee) profileStaff turnoverStructural knowledgeTaking responsibilityTeamwork capacityTolerance for ambiguityUp-to-date competenceVocational qualificationsWork-related competenciesWork-related knowledge

Source: Beattie and Thomson (2004).


3. Content-analytic studies of IC disclosures in annual reports

Despite the impediment of no precise IC definition, researchers have used content analysisto examine corporate annual reports with the aim of identifying IC disclosures. Key features ofthe methods employed in the main studies to date are summarised in Table 3. All but one studyfrom those identified conducted manual searches of small samples of corporate annual reportsfor a number of sub-categories of IC information. Typically 22-25 terms across the three mainIC categories are searched for; Abeysekera and Guthrie (2005) used 45 terms clustered into 16sub-categories across the three main IC categories; Bontis (2003) used a list of 38 terms withoutclassifying these into high-level categories. The IC sub-categories used are shown in Table 4.

From an analysis of the methods reported (or not reported) in these studies, the present paperidentifies six specific issues in using content analysis to investigate IC disclosures in corporateannual reports. Many of these issues arise initially from a lack of transparency in the methods used.The generic issues that arise in content analysis are well covered in texts such as Krippendorff(1980), Boyatzis (1998) and Neuendorf (2002).

3.1. Concept boundary problems and coding reliability

According to Boyatzis (1998), content analysis requires a description of how to know whena particular category occurs, any qualifications or exclusions, and examples of categorised infor-mation. There is a pervasive lack of explanation in previous studies of the detailed coding rulesused to allocate information to IC categories. This results, at least in part, from the boundaryproblems identified above in relation to the IC construct as a whole, its main categories and the sub-categories. This lack of explanation problematises the interpretation of findings from these studies.It is impossible for the reader to judge which IC disclosures have been captured. For example,Guthrie and Petty (2000), Brennan (2001) and April, Bosma, and Deglon (2003) use an IC frame-work containing six human capital IC sub-categories: employee know-how, employee education,vocational qualifications, work-related knowledge, work-related competency and entrepreneurialspirit. It is not known for sure whether, for example, employee profile information (turnover,average age, etc.) is captured in this framework. This information may not be considered IC dis-closure by the researchers under taking these studies, but as it is considered to be IC informationby some researchers in the field, knowledge of its inclusion (or not) is necessary for findings to beinterpreted and comparisons made across studies. Bozzolan et al. (2003) circumvent this situationby removing vocational qualifications and entrepreneurial spirit in favour of a sub-categoryemployees. Guthrie et al. (2004) have also subsequently modified the list of IC sub-categories toinclude this catch-all employee sub-category. It seems likely that this modified framework takesa broader view of the IC concept and capture IC disclosures not previously counted, explainingthe increased disclosure levels reported by Bozzolan et al. (2003). However, replacing one listof IC sub-categories with another without providing an adequate description of the content ofeach sub-category does not further our understanding of the employee information found. Giventhe subjectivity involved in defining IC, knowledge of the nature of information included and/orexcluded would appear necessary for others to judge from their perspective the extent to which ICdisclosures have been captured, and whether they perceive some disclosures counted as IC may,in fact, not relate to IC at all.

The problem of allocating IC to categories without providing adequate explanation is illustratedin the study by Abeysekera and Guthrie (2005). One of the sub-categories of human capital waslabelled equity issues. In the absence of further explanation, a logical interpretation of this

136V.Beattie

,S.J.Thom

son/Accounting

Foru

m31(2007)129163

Table 3Summary of methods used by content-analytic studies of IC disclosures in corporate annual reports

Study Country Sample Coding methods Reliability checks

Guthrie and Petty (2000) Australia 20 listed companies (inc. 19 largestbased on market capitalisation) asat December 1998

24 IC terms across 3 categories: humancapital, internal capital and externalcapital; incidence of terms recorded;voluntary disclosures only

One researcher coded the report; asecond researcher independentlyconfirmed the coding

Brennan (2001) Ireland 11 knowledge-based (i.e.technology and people-orientated)listed companies as at May 1999.1997-99 annual reports


No mention

Bontis (2003) Canada 10,000 listed companies 38 IC terms identified from IC literature;Electronic word search; incidence of termsrecorded

No mention

Bozzolan et al. (2003) Italy 30 non-financial companies listedon Italian stock exchange, stratifiedbased on industry and sales.2001 annual reports

22 IC terms across 3 categories: humancapital, internal capital and externalcapital; sentences scored and counted:0 = no IC information, 1 = qualitative ICinformation, 2 = quantitative ICinformation; disclosure index computed atoverall and category level; repetitionsexcluded.

Used two coders; explanatory notes onthe content of each IC category withexamples prepared before analysis; coderdifferences established over 5 annualreports; Krippendorff alpha used tomeasure reliability; annual reports codedat two points in time to measure stability

April et al. (2003) South Africa 20 largest listed companies as atMarch 2001 (inc. 7 miningcompanies)


No mention

Bozzolan, ORegan, andRicceri (2004)

Italy, Ireland, UK 30 non-financial listed companiesin each country. 2001 annualreports (partial matching on size &sector)

As Bozzolan et al. (2003); repetitionsexcluded

As Bozzolan et al. (2003)

Abeysekera and Guthrie(2005)

Sri Lanka Top 30 listed on Colombo stockexchange, based on marketcapitalisation for y/e Dec.1998 and 1999 annual reports

45 terms clustered into 17 sub-categoriesacross 3 categories: human capital, internalcapital and external capital; sentencescounted and scored: 1 = IC liability,0 = no IC information, 1 = IC asset

No mention

V.Beattie

,S.J.Thom

son/Accounting

Foru

m31(2007)129163

137

Table 4IC categories and sub-categories used in prior content analytic studies

Guthrie and Petty (2000), Brennan(2001) and April et al. (2003)

Bozzolan et al. (2003) and Bozzolanet al. (2004)

Bontis (2003) Abeysekera and Guthrie (2005)

Human capital (employee competence) Human capital (employee competence) Business knowledge Human capitalKnow-howEducationVocational qualificationWork-related knowledgeWork-related competenciesEntrepreneurial spirit

Structural (internal)CopyrightsPatentsTrademarksManagement philosophyCorporate cultureManagement processesInformation systemsNetworking systemsFinancial relations

Know-howEducationEmployeesWork-related knowledgeWork-related competencies

Structural (internal)CopyrightsPatentsTrademarksCorporate cultureManagement processesInformation systemsNetworking systemsResearch projects

Relational (external/customer)Brands

Company reputationCompetitive intelligenceCorporate learningCorporate universityCultural diversityCustomer capitalCustomer knowledgeEconomic value addedEmployee expertiseEmployee know-howEmployee knowledgeEmployee productivityEmployee skillEmployee valueExpert networksExpert teams

Training and development (know-how, vocationalqualifications, career development and tainingprograms)Entrepreneurial skillsEquity issues (race, gender, religion and disabilityissues)Employee safetyEmployee relations (union activity, employeesthanked, employees featured in annual report,employee involvement with the community)Employee welfare (employee and executivecompensation plans, employee benefits, andemployee share and option ownership plans)Employee-related measurements (value-addedstatements, employee numbers, professionalexperience, education levels, expert seniority, ageof employees)

Structural (internal)Processes (management and technological)Systems (information and networking)Philosophy and cultureIC propertyFinancial relations

Relational (external/customer)Brand building (brands, customer satisfaction andquality standards)Corporate image building (company names andfavourable contracts)

138V.Beattie

,S.J.Thom

son/Accounting

Foru

m31(2007)129163

Table 4 (Continued )Guthrie and Petty (2000), Brennan(2001) and April et al. (2003)

Bozzolan et al. (2003) and Bozzolanet al. (2004)

Bontis (2003) Abeysekera and Guthrie (2005)

Relational (external/customer) Customers Human assets Business partnering (business collaboration,Brands Customer loyalty Human capital licensing agreements, franchising agreements)

Distribution channelsMarket Share

Customers Distribution channels Human valueCustomer loyalty Business collaborations ICCompany names Research collaborations Information systemsDistribution channels Financial contacts Intellectual assetsBusiness collaborations Licensing agreements Intellectual capitalLicensing agreements Franchising agreements Intellectual materialFavourable contracts Intellectual propertyFranchising agreements Intellectual resources

KMKnowledge assetsKnowledge managementKnowledge stockManagement qualityOrganisational cultureOrganisational learningRelational capitalStructural capitalSupplier knowledge


IC sub-category might be that it contained information pertaining to issues of share capital toemployees (employee share option schemes). Fortunately, Abeysekera and Guthrie (2005) doexplain that the equity issues sub-category in their study captures equity issues relating to race,gender, religion and disabilitynothing to do with employee share option schemes but ratherwhat some researchers might term equality issues. This example highlights the potential forambiguity and misunderstanding. In the absence of explanation and transparency, interpretationsof the findings across studies are potentially meaningless.

In a valuable development upon prior studies, Brennan (2001) reproduced annual report extractsof IC disclosures, thus enriching the readers understanding of the nature of IC disclosures capturedby the study. Unfortunately, it is not clear in many instances which IC sub-category the disclo-sure had been allocated to. For example, Brennan reproduces an extract in which one company(SupaRule plc) names five of its major customers. Relational capital includes the IC sub-categorycompany names, a logical sub-category for placing this disclosure (albeit company names canrefer to customers, competitors, suppliers, etc.). However, Brennan reports no incidences of dis-closure under the IC sub-category company names across all companies in her sample, leavingthe reader unsure as to which sub-category this disclosure was allocated to. The explicit linkageof annual report extracts of IC disclosures to IC sub-categories, along with comprehensive codingrules, could substantially reduce such problems.

Abeysekera and Guthrie (2005), in the clustering of 45 IC terms into 16 IC sub-categories,provide a degree of useful explanation in relation to the content of their IC categories. However,through clustering, the composition of each sub-category is lost. Abeysekera and Guthrie explainthat clustering was necessary because the data in the coding framework was too descriptiveto bring analytical rigor to data interpretation (p. 156). However, there may be an alternativejustification for clustering a larger number of IC terms into fewer sub-categories. As noted byMilne and Adler (1999), as the number of content categories in a scheme increases, the potentialfor coding errors increases. They suggest that increased reliability is traded against increasedunderstanding of disclosures, and reliability is essential to permit replicable and valid inferencesto be drawn from data derived from content analysis (Milne & Adler, 1999, p. 238). However,the use of too few coding categories increases the likelihood of random agreement in codingdecisions which may result in an overestimation of the measures used to assess reliability (Milne& Adler, 1999). Consideration of the potential impact of the number of IC sub-categories usedon these issues is not apparent in prior IC studies.

There are two reliability issues when using content analysis to investigate disclosures (seeMilne and Adler, 1999, for an in-depth discussion). First, the coding instrument needs to bereliable in terms of well-specified decision categories and decision rules, to facilitate consis-tent coding decisions across time and researchers. Second, consistent coding decisions needto be demonstrated through the use of multiple coders, multiple time periods, and by report-ing and addressing any inconsistencies. Reliability issues do not appear to be addressed in themajority of IC disclosure studies to date (see Table 3). Coding instrument reliability is onlyacknowledged by Bozzolan et al. (2003), who state that explanatory notes on the content ofeach category item and examples of sentences were prepared and discussed before analysis.Without being privy to these notes and examples, reliability is still difficult to judge. However,Bozzolan et al. (2003) do demonstrate consistent coding decisions through the use of mul-tiple coders, coding at alternative time periods and by reporting inconsistencies. Apart fromGuthrie and Petty (2000), who mention that a second researcher independently confirmed thecoding of a first researcher, the consistency of coding decisions is not mentioned in prior ICstudies.


3.2. Manual versus electronic searching

With one exception, prior studies have manually analysed annual reports for IC information,which has inevitably restricted the sample size due to the labour-intensive data collection pro-cess. Bontis (2003) introduced a different approach by performing an electronic search on 38 ICkeywords for a large sample of 10,000 Canadian companies. Minimal levels of disclosure werefoundonly 74 occurrences by 68 companies! Although words have the advantage of being cat-egorised more easily and large databases can be scanned for specific words (Gray et al., 1995),there are problems in using electronic word searches to investigate IC disclosure. Weber (1990)highlights the problems with synonyms and words with multiple meanings when using frequencyword counts. Expanding the number of keywords and a keyword in context (KWIC) search wouldreduce, but not eliminate, these problems. The use of pronouns (words used instead of nouns toindicate something without naming it) also creates problems when identifying keywords. Milneand Adler (1999) discredit an electronic keyword search method when noting that understand-ing of disclosures is best achieved by consideration of whole sentences, with individual wordsunlikely to convey much meaning. Moreover, a keyword search is unlikely to detect IC disclo-sures expressed in company-specific terms. An example would be an electronic search on theterm company names, which is unlikely to identify specific company names.

3.3. The annual report material analysed

Prior studies are generally silent or vague about exactly what parts of the annual report areanalysed. Possibilities include the entire annual report, including notes to the accounts; the frontend of the annual report; or selected sections from the front end, such as the chairmans state-ment, CEO review or MD&A/OFR. Several studies note that they analyse voluntary disclosuresonlypresumably this includes all sections of the annual report although this is not made clear.In general, therefore, the population (i.e. the set of units being studied) is not carefully defined.This creates interpretation difficulties as findings across studies relate to unknown and potentiallydifferent populations.

3.4. The volume of disclosure: presence/absence versus count of occurrences (with/withoutrepetition)

The importance attached by a reporting entity to different categories of information is assumedto be reflected by the extent of information disclosed (Krippendorff, 1980). This is the fundamentalpremise of content analysis. Detecting the presence or absence of a particular item does notcapture the volume of disclosure of that item. However, recording the presence or absence of eachitem against a predetermined check list does enable the range (i.e. variety) of disclosure to becompared across reporting entities. To capture the volume of disclosure requires a count of thenumber of times each item on the checklist occurs. Guthrie and Petty (2000) appear to documentonly the presence or absence of information pertaining to each IC sub-category for a particularcompany (i.e. two different pieces of information relating to the same IC sub-category wouldcount only once). However, the precise detail of the content analysis research method adoptedis, unfortunately, not transparent as no reference is made to multiple disclosures. Both Brennan(2001) and April et al. (2003) appear to follow suit, with only the latter explicitly stating thatIC sub-categories were mentioned multiple times in the reports they analysed, but the number ofoccurrences was ignored on the basis of disclosures being mostly repetitions (i.e. duplications).


Recording only that an IC sub-category is mentioned at least once is a very partial analysis ofthe amount of IC disclosures in corporate annual reports. Different pieces of information relatingto the same IC sub-category are ignored. Hackston and Milne (1996) highlight that a simplepresence/absence approach may be misleading as it treats a company making one particulardisclosure as equal to one that makes 50 disclosures. The extent to which IC disclosures arerepeated is also of interest. It is common for the same information to appear in different sectionsof annual reports. While this introduces redundancy, repetition is a communication strategy usedfor emphasis and reinforcement and signals the importance placed by management upon thesemessages (Beattie & Jones, 2001; Lothian, 1976).

Bozzolan et al. (2003) claim to replicate the Guthrie and Petty (2000) study for a sampleof Italian companies. Indeed, they make a direct comparison between the findings of the twostudies and conclude that, on average, Italian companies disclose more IC information than thosein Australia. This result was described by the researchers as unexpected, given Guthrie andPettys sample comprised Australias largest listed companies and a company held as an exampleof best practice in the field of IC reporting. Explanations are offered in terms of recent Italianinitiatives leading to increased attention to IC and an analysis based on reports published 3 yearson from Guthrie and Petty for companies where IC categories are structurally higher. However,an additional explanation is that the findings are simply not comparable, given that Bozzolan etal. (2003) appear to have recorded multiple disclosures (albeit not repetitions) whereas Guthrieand Petty (2000) do not.

3.5. Location and type of IC disclosure

According to Gray et al. (1995), there is no single, unique choice why any particular disclosurelocation in the annual report should be preferred. However, the location of IC disclosure in thedifferent sections of the annual report may be informative in terms of, for example, the importanceattached to the information, reader attention and auditor confirmation. Analysing the different sub-categories of IC information found in the same location may be useful in identifying underlyingrelationships within the IC concept. However, there are interpretation problems in relation torepeated information and similar information which might be legitimately expected to appear inseveral sections of the annual report (Gray et al., 1995). There is a lack of consideration in relationto the location of IC disclosures in prior studies. For example, results reported by Guthrie andPetty (2000) contain no reference to the location of disclosures despite an indication in the paperthat this information was recorded.

Investigating only the volume of disclosures is potentially misleading when it is the credibilityor quality of disclosure that is important (Toms, 2002). Further, content analysis measuring thevolume of disclosures may be insufficient for the purposes of identifying underlying relationships(Hasseldine, Salama, & Toms, 2005). A quality-adjusted method of content analysis is suggested,in which disclosures are counted but also weighted to reflect their likely significance (Hasseldineet al., 2005).

Prior studies have investigated the type of IC disclosure, mainly in terms of whether thedisclosure is quantitative or qualitative. Guthrie and Petty (2000, p. 247) found that nearly everyinstance of reporting involved the intellectual capital attribute being expressed in discursive ratherthan numerical terms. Bozzolan et al. (2003) introduce a weighting scheme in the developmentof their IC disclosure index by counting qualitative disclosures as 1 and quantitative disclosuresas 2. This is intended to proxy the quality of disclosure. According to Toms (2002), in thecontext of CSR disclosures, it is difficult for competitors to imitate quantified disclosures. He also


suggests that quantified disclosures are more likely to be accurate and represent actual activities.However, the linking of quality with quantitative information might be inappropriate in the ICcontext. Guthrie and Petty (2000) highlight the difficultly involved in trying to quantify IC, whenin many instances it is a qualitative item. They suggest that many companies are interested inunderstanding where real firm value lies rather than assigning monetary amounts. This wouldappear to suggest that in the absence of quantitative disclosures, qualitative disclosures requireclose attention in the IC context. The quality or significance of qualitative IC disclosures doesnot appear to be discussed in prior studies. Toms (2002) suggests that rhetoric and non-verifiabledisclosures should carry less weight as they can be made in large volume without commitment. Itis not clear whether prior IC studies screen out these disclosures in the coding process (i.e. theyare not included in the IC count), or whether they are included and given the same weight as moreinformative, higher quality disclosures.

Abeysekera (2006) notes that the majority of prior IC disclosure fail to consider the possibleexistence of intellectual liabilities. Abeysekera and Guthrie (2005) acknowledged and incorpo-rated both IC assets and liabilities in their study by using a coding system of 1 for intellectualliabilities, 0 for no IC information, and +1 for intellectual assets. However, it is not transparenthow qualitative disclosures were categorised into asset and liability types, i.e. what constitutesan intellectual liability. Unfortunately, there appears to be no mention of the extent and natureof intellectual liability disclosures found. It is, therefore, not clear whether any intellectual lia-bilities were disclosed or not. A lack of intellectual liability disclosure might justify the lack ofconsideration to intellectual liabilities in prior studies.

3.6. Unit of analysis and unit of measurement

Using content analysis to investigate IC disclosure involves deciding what should form thebasis for coding (unit of analysis) and what should form the basis for measuring the amount ofdisclosure (unit of measurement). Milne and Adler (1999) suggest that sentences are the mostreliable unit of analysis. On this basis, each sentence in the annual report would be analysed todetermine if it provides an IC disclosure (or not) and, if so, to which IC sub-category it relates.A comparison of coding decisions made between multiple coders can then be used to establishreliability. Coding sentences would appear relatively straightforward if sentences containing ICdisclosures referred to only one sub-category of IC information. However, information pertainingto different IC sub-categories could well be disclosed within the same sentence. In this situation, adecision has to be made in relation to the dominant IC sub-category. Additional coding rules wouldbe required in order to identify dominant themes. Alternatively, the unit of analysis becomes partsof sentences or words. This problem is accentuated if paragraphs or areas of pages are used as theunit of analysis, as the likelihood that different sub-categories of IC information are discussed isincreased.

Beattie, McInnes, & Fearnley (2004a, p. 32, 2004b), in their detailed analysis of annual reportnarratives, frequently found it necessary to split sentences into text units, each group of wordscontaining a single piece of information that was meaningful in its own right. The extract inBrennan (2001) cited above is a good illustration of this issueit could be argued that the namesof five customers could be classed as five separate pieces of information. In using text units asthe unit of analysis, each piece of information in a sentence is coded based on the context of thatsentence. Milne and Adler (1999) discredit the coding of single words on reliability grounds asthey have no meaning to provide a sound basis for coding without a sentence or sentences forcontext.


According to Unerman (2000), the use of sentences along with words and characters is partialin that it will only capture narrative disclosures. Other visual forms of communication such asgraphs, tables and pictures have been found to provide an immediate and effective means ofdisclosure in corporate annual reports (e.g. Beattie & Jones, 1992; Graves, Flesher, & Jordan,1996; McKinstry, 1996). This appears to suggest that coding should also be applied to thesenon-narrative forms of disclosure. For example, the disclosure reproduced in Fig. 1 concerningemployee satisfaction (a component of human capital) (taken from EMAP plcs 2004 annualreport) would not be fully captured by an analysis of only narrative forms, even if the captionswere included. In a recent study of pictures contained in the annual reporting documents of the top100 UK listed companies, it has been found that approximately 94% of pictures communicatedintangible aspects of companies businesses (Davison & Skerratt, 2007, p. 9).

The choice of unit of analysis has reliability implications. In order to measure the reliabilityof coding decisions (with calculations such as Krippendorffs , for example) the total numberof coding decisions each coder makes and the coding outcome of each decision is required to beknown (Milne & Adler, 1999). If the unit of analysis is words, sentences, paragraphs or pagesthen there is no dispute over the number of coding decisions to be made. The number of codingdecisions simply equates to the number of words, sentences, paragraphs or pages contained inthe document being analysed. However, the use of text units involves a multi-stage process. First,all the sentences would need to be coded as containing an IC disclosure or not. At this stage,the number of coding decisions would equate to the total number of sentences. Second, all thesentences agreed to contain IC information would need to be coded in relation to how many piecesof information they were thought to contain. At this stage, the number of coding decisions wouldbe equal to the number of sentences agreed to contain IC information. Finally, all the agreedpieces of information would need to be coded to an IC sub-category. At this stage, the number ofcoding decisions would be equal to the agreed number of IC information pieces.

The unit of measurement is also an important consideration as one of the key assumptions under-lying content analysis is that volume signifies the importance of items being disclosed (Unerman,2000). Milne and Adler (1999) advocate that each coded sentence should count as one disclosure,i.e. sentences form the basis for both coding and measuring. However, in addition to the issue ofmultiple IC sub-categories discussed above, this ignores the fact that sentences vary in length andin the extent of IC information they contain. Grammatical differences might result in the samemessage being conveyed using a similar number of words, occupying a similar amount of annualreport space but using a different number of sentences (Unerman, 2000). Counting the number ofwords in each sentence could solve this problem. However, in doing so it creates problems of itsown when determining which individual words are IC disclosures and which are not (Hackston &Milne, 1996). Beattie et al. (2004a) used their text units for both coding and measuring narrativedisclosures. Measuring sentences, words or text units fails to capture differences in font choice,size, boldness, etc., which could all be used to stress the importance of the item being disclosed.Unerman (2000) suggests measuring proportions of pages as a solution to grammatical differences,font size and non-narrative disclosures. However, differences in print sizes, column sizes and pagesizes across annual reports could render comparisons across companies on the basis of page pro-portions unreliable (Ng, 1985, cited in Hackston & Milne, 1996). Further, boldness or emphasisthrough colour in both non-narrative and narrative disclosures would still not be captured.

Choices regarding unit of analysis and unit of measurement do not arise in the prior IC studieswhich ignore multiple disclosures and focus merely on presence/absence. In the other studies,Bozzolan et al. (2003) chose sentences as the unit of analysis and measurement. The choice madein Abeysekera and Guthries (2005, p. 157) paper is not clear: they state that in the word count


Fig. 1. IC disclosure from EMAP PLC 2004 annual report.


category, the method of counting lines (sentences) was chosen as the context unit instead of word,paragraph or page. They go on to highlight the benefits of sentences as the unit of analysis andalso say that line count methods provide a more appropriate starting point from which to convertcharts, tables and photographs into equivalent lines for comparison. This appears to infer thatlines and sentences are the same, when this is not the case as one sentence may span several lines,and one line may contain more than one sentence. An added problem is that annual reports varyin the number of columns per page; line counts will be inflated where multiple columns exist.

In the next section, Next plcs 2004 annual report is used to illustrate some of the above issues.It is pertinent to stress that this analysis was undertaken solely for the purpose of highlighting thespecific problems when using content analysis. It does not claim to provide an assessment of theextent and nature of IC disclosure by Next plc. Moreover, the coding procedures adopted are notclaimed to be ideal or even the best and do not offer a solution to all the issues involved.

4. Analysis of IC disclosuresan illustrative example: Next plc

4.1. Choice of illustrative example

Next plc (2004 annual report) was chosen on the basis of its high market-to-book ratio. Theoriginal IC definition used in the US was the difference between a firms market and book values(Stewart, 1997, quoted in Chaminade & Roberts, 2003, p. 746). Similarly, Edvinsson (1997), theformer director of IC at Skandia, defines IC as overall value less financial capital. In an analysisbased on the FTSE 100 companies in 2002/2003, Next plc exhibited the third highest market-to-book ratio (8.21) (Beattie & Thomson, 2005). As noted in the introduction, the difference betweenmarket and book values can result from factors other than unrecorded intangible assets. Brennan(2001) highlights factors such as unrealistic tangible balance sheet assets and fluctuating dailyshare prices causing market-to-book ratios to be unreliable short-term measures of IC. Fig. 2shows the trend in market-to-book ratio for Next plc over the 18-year period 19852003. Theconsistently high ratio in recent years indicates that fluctuating daily share prices are an unlikelyinfluence. In addition, Next plc operates in the retail industry which is characterised by relativelystandardised tangible assets, minimising the problem of unrealistic tangible balance sheet values.

4.2. Approach taken to illustrate specic issues

Following Beattie et al. (2004a), the unit of analysis and the unit of measurement was thetext (or table) unit. The use of text (table) units is not offered up as the perfect solution in the

Fig. 2. Next plcs market-to-book values: 19852003.


use of content analysis. It is more complex than the use of sentences, and increased complexityhas the potential to decrease reliability. However, reliability can still be measured when usingtext units, and it has the following advantages: it enables the extent to which different categoriesof IC information are disclosed to be investigated; it avoids making decisions in relation to thedominant IC category disclosed in each sentence; and does not exclude the IC information whichis less dominant.

The entire content of the annual report (excluding financial statements) was manually analysedto identify IC information. Each sentence, headline and table (there was no pictorial material) wasclassed as containing IC information or not. IC information identified in the form of sentences,headlines and tables was then split into text units (pieces of information provided) and coded toIC sub-categories obtained from a review of the literature by Beattie and Thomson (2004) andshown in Table 2. This list is not definitive; typically researchers have subsequently modified theirIC frameworks (e.g. Abeysekera & Guthrie, 2005). However, it provides a relatively extensivebase from which to identify IC disclosures and takes a broad view of the IC concept for thepurpose of illustration.4 It is recognised that an increase in IC sub-categories potentially decreasesinter-coder reliability (Milne & Adler, 1999); hence this illustration does not claim to reliablyreport the extent and nature of IC disclosure made by Next plc. In analysing disclosures, relevantinformation was identified which did not fit into the original set of sub-categories. Accordingto Gray et al. (1995) this is a common situation and it would be inappropriate to ignore suchinformation. Consequently, in this situation, new sub-categories were created and existing sub-categories modified (described below). Coding rules were developed after an initial analysis of theinformation disclosed. Text/table units were coded based on the context in which they appeared(i.e. within a sentence, headline or table). Coding was undertaken by one of the authors andverified by the other author. An electronic word search of Next plcs 2004 annual report in pdfformat using Adobe Acrobat was also conducted.

4.3. Concept boundary problems

Inevitably, the volume and nature of IC disclosure found is a product of the various codingrules adopted and the subjective judgement of the coder in applying them. However, increasing thetransparency of the methods applied facilitates comparisons across studies. For example, what isto count as employees? Is it all human assets receiving remuneration from the company or shoulda distinction be made between the board of directors and other employees? The composition ofindividual boards of directors undoubtedly impacts in varying degrees on the success or otherwiseof a company. Adopting a broad view of the IC concept in this illustration, the informationconcerning directors (executive and non-executive) and employees was included.

During an initial analysis of Next plc, a considerable amount of information was found inrelation to employee remuneration. This sub-category had not previously been identified from theliterature (Table 2). Should all remuneration information be classed as IC? An argument for thiscan be made on the grounds that remuneration procedures influence the attraction, retention andsatisfaction of employees and hence the value of human capital. Subsequently, this informationwas recorded in a new sub-category termed remuneration procedures for the purposes of this

4 It is worth noting that some simplification in inherent in all classification systems, otherwise all that is produced is adescription of reality which serves little purpose. Classification systems can be created at various levels of aggregationand may focus on different attributes. The suitability of any given classification system depends upon the purpose forwhich it was created.


illustration. Abeysekera and Guthrie (2005) appear to have also taken the same view as theyinclude employee compensation plans, employee benefits and employee share option ownershipplans in the IC sub-category employee welfare. However, in the analysis of Next plc remunerationprocedures were considered as structural capitalinternal procedures to attract, retain and satisfyhuman capital, independent from the human capital itself. Abeysekera and Guthrie (2005) adopt adifferent perspective by considering employee welfare to be part of human capital. Alternatively,other researchers could decide that the remuneration procedures disclosed across companieswere sufficiently similar not to warrant inclusion in any IC category. The potential for adoptingalternative perspectives is evident, as is the essential need for transparency to compare findings.

Adopting a broad view of IC, 906 pieces of IC information were identified in Next plcs 2004annual report. This information was coded across 20 sub-categories of human capital, 16 sub-categories of structural capital and 23 sub-categories of relational capital, as shown in Table 5,columns 1 and 2. In this table, a subjective, summary description of the nature of IC informationpertaining to each IC sub-category is offered in column 3. The final column identifies possiblealternative classifications and serves to highlight the difficulty involved in allocating informationto only one category. For example, the following sentence could be categorised as providinginformation in relation to either human capital or as procedures, part of structural capital:

The Group has developed policies for recruitment, training and development of personnel.[Source: Corporate Social Responsibility Statement, p. 16].

Similarly, the following sentence could be categorised as either corporate culture or manage-ment philosophy:

The Board promotes the development of a strong control culture within the business.[Source: Corporate Governance, p. 14].

In this example, both sub-categories relate to structural capital, so perhaps the clustering of alarge number of sub-categories into a smaller number of sub-categories (i.e. corporate culture andmanagement philosophy information contained in the same sub-category of structural capital) asadvocated by Abeysekera and Guthrie (2005) would eliminate the problem.

For the data in columns 1 and 2 in Table 5 to be meaningful, researchers need to develop andreport comprehensive coding rules and give examples of information allocated to each IC sub-category. To illustrate, Fig. 3 reproduces an annual report extract in conjunction with the relevantcoding rules adopted and the IC content identified.

4.4. The annual report material analysed

The Remuneration Report contained the most IC information (21.3%), mainly in relation tocompetitors/competitor names, expert teams, remuneration procedures and staff profile. A sig-nificant amount of IC information (20.4%), predominantly relating to components of relationalcapital, was found in the Chief Executives Review. Smaller numbers of IC disclosures were evi-dent in the Chairmans Statement (2.8%), the Corporate Social Responsibility Statement (7.6%),and the Directors Report (8.3%). It is clear that many IC disclosures are found in sections ofthe annual report that are subject to mandated requirements (e.g. Directors Report, notes to theaccounts). It is, therefore, vital that the boundaries of the material being analysed are clearlydefined and not drawn too narrowly, otherwise comparisons across studies are impossible and ICdisclosures could be missed.


Table 5Manual content analysis of Next plcs 2004 annual report: Occurrence frequency and nature of IC information

IC Sub-Category Frequency Nature of information Alternativecategorisation

Panel A: Human capitalEmployee attitude 6 Enthusiasm of all employees (repeated

twice). One non-executive directordescribed as having an independentapproach, being probing and persistent.Another described as having a commonsense approach

Employee capability 3 Strength of management team. Twoboard members are eligible forreappointment

Recruitment/staffturnover

Employee commitment 4 Dedication of all employees (repeatedtwice). Two board members willing to bereappointed

Recruitment/staffturnover

Employeecommunicativeactivities

32 Giving employees group information.Taking into account employee views andsuggestions. Frequency and nature ofboard meetings. Open discussions andregular information provided todirectors. Next brand trading meeting.Frequency of Audit Committee,Remuneration Committee andNomination Committee meetings

Expert teams/brands

Employee development 1 Developed policies for development ofpersonnel.

Structural capital(procedures)

Employee experience 28 Previous positions held outside Next bydirectors and officers. Ranging frommarketing experience, accountingexperience to experience working inparticular companies

Staff profile/workrelated competencies

Employee motivation 1 A well motivated management team Employeeattitude/employeecommitment

Employee sensitivity 1 One of the non-executive directorsdescribed as having a thoughtful way

Employee attitude

Employee skills 2 Skilled management team. Design skillscan add value in developing andextending product range

Equality 10 Statements in relation to disabledemployees, equal opportunities, nodiscrimination, equal pension rights formembers of either sex

Expert teams 74 Remuneration and nominationcommittees: members and function.Operational management. Auditor teamre supplier compliance. Responsibilityfor central treasury function

Taking responsibility/organisationalstructure

Human assets 1 People area key asset to the businessProductivity 2 Objectives and targets set for the chief

executiveRecruitment 2 Another non-executive director required.

Policies in place for recruitmentStructural capital(procedures)


Table 5 (Continued )IC Sub-Category Frequency Nature of information Alternative

categorisation

Staff profile 90 Positions held: who holds them, duration,age profile. Directors ages and serviceyears. Staff numbers, average per Nextbrand, Ventura and other and full timeequivalents

Staff turnover

Staff turnover 20 Retirement and length of service.Termination notice and turnover in boardmembers

Taking responsibility 19 By particular people or groups includingthe chief executive and the board

Expert teams

Training 1 Policies developed for training Structural capital(procedures)

Work-relatedcompetencies

17 Current positions held outside Next bydirectors and officers. Non-executivedirectors bearing considerable judgement

Work-relatedknowledge

1 Non-executive directors bringconsiderable knowledge

Total 315

Panel B: Structural capitalInternal communicative

activities1 Significant risk issues are referred to the

boardProcedures

Corporate culture 1 Board promotes a strong control culture Managementphilosophy

Customer support 1 The majority of the increase in Venturasprofit came from customer serviceactivities

Financial relations 7 Bankers and stockbrokers. Investments ofcash surpluses with bank and companieswhich fulfil credit ratings and investmentcriteria. Board role is to represent andpromote interests of shareholders

Financial strategy 21 Use of debt and share buybacks. Use ofcapital. Borrowing terms, managinginterest rates, hedging, interest rate swaps,options, forward rate agreements

Infrastructure 22 Warehousing and call centres. Investmentsin additional capacity. Sourcingcompanies

Innovation 1 Trailing a wedding list service Markets(relational capital)

Managementphilosophy

4 Style, quality and value of ranges remainhighest priority. The product is believed tomake the brand successful

Brands

Management processes 15 Risk management process. Process ofcommission of CSR report

Social matters

Operation process 2 Increased operating efficiency. Highlymechanised warehouse

Infrastructure

Organisationalflexibility

3 Increased number of customers who areable to use directory accounts to makepurchases in store. Using treasury sharesto satisfy employee share options

Customers(relational capital.)Financial strategy



categorisation

Organisationalstructure

2 Division of responsibilities betweenoffices of chairman and chief executive.Groups management structure

Taking responsibility

Procedures 10 Business continuity plans. Expenseclaims. Safety and Ethicalresponsibility of Suppliers. MinimisingRisk. Health and Safety

Quality improvements 4 Style, quality and value. Bettersourcing and selection. Improved stockavailability. Reduced product returnrates

Remunerationprocedures

137 Detailed rules in relation to: share saveemployee option scheme; remunerationpackages of directors; how salaries areset; bonus schemes; pension schemes;directors benefits

Systems 9 Internal control

Total 240

Panel C: Relational capitalBrands 13 Turnover, operating profit, earnings per

share for 2004 and 2003 for the Nextbrand. Sales increase. Staff costs forthe Next Brand

Business collaborations 4 Choicean associated company whichoperates a chain of 13 discount stores.Cotton Traders an associatedcompany selling its own brand product

Company names

Client names 2 Pension Credit Services and BritishGas

Company names

Client profile 5 Ventura commenced providing serviceon behalf of several new clients.Increased demand from existing andnew clients. Good progress inbroadening client base. Winning newclients

Collaborations 1 Working in partnership with suppliers SuppliersCommunity 12 Charitable donations. Corporate and

charity sponsorship programme.Committee of employeerepresentatives to ensure diverse rangeof charitable causes supported

Competitor names 27 List of 20 comparator companies. JJBsports, Arcadia, Debenhams andSelfridges removed, replaced byBurburry, Signet and Woolworths

Company names

Competitors 3 Comparator group most comparable interms of size or nature. Totalshareholder return of Next rankedbetween fifth and sixth compared to thegroup of 20 other retailers

Customer knowledge 1 Understanding of our customers Customers



categorisation

Customer satisfaction 11 For customers: products of good qualityand value; Improvements in value;greater choice; more comfortableshopping environments; improved andexpanded store based services fordirectory customers; competitive prices;purchase and return in the way which ismost convenient

Customers

Customers 12 Active customers, growing the base,growth expectations, account balancesfor directory customers

Distribution channels 106 Stores and Next Directory (HomeShopping)

Environmentalactivities

8 Waste stream management. Recyclingprocedures

Structural capital

Ethical matters 5 Ethical tradingExternal

communicativeactivities

11 Communication through annual reports,yearly and half yearly announcements,regular trading updates to stockexchange. Company website,opportunities at AGM, shareholdersviews included in board reports

Shareholders

Financial contracts 7 Borrowing requirements and termsFranchising agreements 33 Turnover, operating profit, earnings per

share for 2004 and 2003 for the NextFranchise. Number of franchise stores inMiddle East, Japan. Success in Iceland

Market channels 40 Turnover, operating profit, earnings pershare for 2004 and 2003 for geographicalsales regions

Market intensity 1 Development of product rangesNegotiation 1 Re-negotiation of Venturas contract at

lower marginsSocial matters 18 Social responsibility. Health, safety,

welfare and fire preventionStakeholders 15 Substantial shareholder informationSuppliers 15 Support of suppliers, improvements,

payment, sources, supplier code ofpractice

Total 351

4.5. The volume of disclosure

The use of presence/absence of an IC sub-category does not capture the volume of IC informa-tion being disclosed. Of the 59 IC sub-categories identified as having at least one IC disclosure, 45had multiple disclosures. Only 3.3% of IC information was repeated. Given the relatively smalloccurrence of repeated disclosures, their treatment may not be a critical issue. The maximumnumber of disclosures for a single sub-category was 137 in relation to remuneration procedures.If multiple disclosures were ignored, neither the absolute nor the relative volume of information


Fig. 3. IC extract and analysis 1.

provided about remuneration procedures would be captured. Ignoring multiple disclosures alsofails to capture different IC disclosures relating to the same IC sub-category. For example, six dis-closures were found relating to employee attitude: all staff were said to be enthusiastic (repeatedtwice), a particular member of staff was said to maintain an independent, probing and persistentapproach (counted as three disclosures) and another member of staff was said to have a commonsense approach.

4.6. Type of IC disclosures

In this illustrative study of Next plc, two type attributes were recordedwhether the dis-closure was quantified or non-quantified and whether it was a fact (capable of verification) orjudgement (unsubstantiated statement) (Beattie et al., 2004a, p. 31). For example, the followingsentence, coded as an IC disclosure in the distribution channels sub-category, was recorded asboth quantified and factual:


This years portfolio of new space is exceeding its appraised sales target by 12% [Source:Chief Executives Review, p. 4].

The following example, coded as an IC disclosure in the franchising agreements sub-category,was recorded as neither quantified nor fact. As Next plc failed to define success, the statementrepresents an unsubstantiated judgement.

The success of our franchise in Iceland [Source: Chief Executives Review, p. 5].The claimed superiority of quantified disclosures over non-quantified disclosures has been

previously noted. Almost all quantified disclosures were also recorded as factual and could thusbe considered as higher quality IC disclosures. However, although the following example of ICdisclosure in the distribution channels sub-category was recorded as quantified, it is based oncompany expectations and thus it is considered a company judgement:

We currently expect to increase net selling space by around 420,000 square feet in the yearahead. [Source: Chief Executives Review, p. 5].

In weighting disclosures, researchers need to consider how the quality of a quantified judgementcompares with a non-quantified fact. The following example was recorded as a non-quantifiedfact in the work-related competencies sub-category, as the duration these positions have beenheld for is not disclosed.

He (the chairman) is also a non-executive director of Aggregate Industries plc andLeicesterFootball club plc. [Source: Directors and Officers, p. 9].

In this example, it is the duration which adds quantification to the disclosure. However, whetherin this example, quantification would add further value to the disclosure is debatable. If thechairman is disclosed as having held these positions for a length of time, it could imply that hisrelated long-term experience may be of value to his current position with Next plc. If he has justbeen appointed in these positions, it could imply that his experience is limited and, therefore, oflittle benefit to his position with Next plc. Irrespective of the length of related experience, it isdifficult to judge the extent to which it will contribute to Next, if at all.5 This example highlights thesubjectivity and difficulty involved in pinpointing which disclosures are more informative/valuableand in weighting disclosures.

Recording IC disclosures on the basis of quantified/non-quantified and fact/judgement fails tofully capture the significance of the disclosure. The following two examples were both recordedas non-quantified judgements. However, the former does provide a degree of future intentions,whereas the latter is more a sweeping statement, of the form which Toms (2002) refers to asgeneral rhetoric.

. . .advancing the underlying operating prot of the Next Group will mainly be achievedthrough. . ..expansion of our selling space [Source: Chief Executives Review, p. 4].The success of Next owes somuch to the strength and skills of awell-motivatedmanagementteam. [Source: Chairmans Statement, p. 2].

In the analysis of Next plc, approximately half of the information found (51%) was quantifiedand 68% was factual. Approximately 33% was neither quantified nor factual. Measuring and

5 Hackston and Milne (1996, p. 108) propose that all coding be based on stated (and not implied) disclosures.


comparing IC disclosures on the basis of volume offers a partial view. The weighting of differenttypes of disclosure requires consideration. Perhaps their quality could be assessed and recordedin terms of how useful the disclosure is for decision-making? To take account of IC disclosuretype, further coding rules are required. In the CSR literature this has led to the use of multipledigit codes in content analysis. For example, Hackston and Milne (1996) suggest a four digitcode in relation to CSR disclosures. The first digit records the existence of a CSR disclosure, thesecond digit records the category of disclosure, the third digit records the type of disclosure andthe fourth digit records whether the disclosure is good, bad or neutral news. They suggest thatthe use of multiple digit codes facilitates the recording of coding decisions and permits formalreliability calculations to be performed. This method would also permit the type of disclosure bycategory to be analysed.

4.7. Unit of analysis and unit of measurement

Coding and counting text units (parts of sentences capturing a piece of information) avoidsthe problem of coding sentences which contain different sub-categories of IC information. Forexample, consider the following sentence:

The success of Next owes so much to the enthusiasm and dedication of our employees andthe support of our suppliers [Source: Summary of Performance, page 1]

Using the text unit approach, this sentence provided three pieces of information: Next attributessuccess to enthusiastic employees, Next attributes success to dedicated employees and Nextattributes success to the support of suppliers. This sentence was coded as two pieces of informa-tion relating to human capital (employee attitude and employee commitment) and one piece ofinformation relating to relational capital (suppliers). Although sentences are broken down accord-ing to how many pieces of information they contain, and each piece of information is allocated toa sub-category of IC information, the piece of information remains in the context of the sentencethroughout this process. In the example above, enthusiastic employees, dedicated employees andsupplier support were all categorized in the context of what the success of Next is being attributedto. If this sentence were coded to one IC sub-category, i.e. if sentences were used as the unit ofanalysis, a dominant sub-category would need to be identified. The sentence refers to employees(human capital) and suppliers (relational capital). If human capital were chosen, then a furtherdecision would need to be made in relation to sub-category, in this case attitude or commitment(although this decision could be eliminated if one sub-category represented both attitude andcommitment). Without quantifying the pieces of information contained in the sentence (i.e. twopieces about employees, one about suppliers, so employees dominate on the basis of volume),how would the coder decide the dominant theme? Would employees dominate because they arementioned first before suppliers in the sentence or would the choice depend on the coders per-ceptions/intuitions of what is more important? Whatever category is chosen to dominate, thedisclosure in relation to the other is lost.

Across all the material analysed, 18.1% of the IC disclosures found were in table format,leaving 81.9% in text format (including 3.9% list items). Coding text units facilitated the inclusionof IC information provided in forms other than sentences, as shown in the examples contained inFigs. 46.

The 906 pieces of IC information identified were spread across 260 sentences identified ascontaining IC information, indicating the major impact that the choice of unit of analysis and unitof measurement can have on the volume of IC disclosures identified. In this illustration, there



was no use of colour (in text or pictorial format) and only 2% of IC text was emphasised inbold headline format. However, as colour and headlines are prominent features in many corporateannual reports, this use of emphasis could still be captured when analysing text/table/visual unitsby including additional type attributes.

4.8. Manual versus electronic searching

In total, 105 IC terms were electronically searched for after removing 23 synonyms from thelist in Table 2. Thirty-five of these terms were found in Next plcs 2004 annual report, while 70generated a zero count (Table 6). The total frequency count across the 35 terms was 264, only29% of the total number of information items obtained from the manual search. On the basis ofquantity of information found, the electronic search is inferior. Further problems were revealed


Table 6Electronic content analysis of Next plcs 2004 annual report

Search terms IC category Totalhits

KWIC review Pieces of KWICverified ICinformation from search term

IC Hits Not IC Hits HC SC RC

Employees HC 39 23 16 24 12 3Knowledge HC 1 1 0 1Commitment HC 12 1 11 1Communicative/communication HC/SC 4 3 1 2 1Development HC 20 7 13 4 3 1Productivity HC 1 1 0 1Teams HC 3 2 1 6Training HC 3 3 0 2 1Experience HC 16 11 5 22 2Personnel HC 2 2 0 1 1Recruitment HC 2 1 1 1Skill HC 2 2 0 2Responsibility HC 11 4 7 2 2 2Staff HC 6 3 3 1 1 2Culture SC 1 1 0 1Processes/process SC 8 8 0 2 9Operations SC 11 6 5 5 1 1Flexibility SC 1 1 0 1Structure SC 3 3 0 2 1Procedures SC 8 3 5 4 1Quality SC 5 4 1 2 2Research SC 1 1 0 1Systems SC 3 3 0 2 1Strategy SC 1 1 0 1Brands RC 12 10 2 8 1 17Customers RC 25 19 6 2 24Distribution channels RC 1 1 0 1Franchise/franchising RC 8 7 1 17Clients RC 4 4 0 5Commercial RC 2 0 2Competitive/competition RC 2 2 0 1 1Environmental RC 4 4 0 4

when a keyword in context review was undertaken to confirm the IC content. Of the 264 termsfound, 107 (40%) were judged to have no IC content, and may be described as false hits.

This may appear to suggest that, excluding false hits, only 17.3% of IC information identifiedin the manual analysis was captured by the electronic search. A mitigating factor, however, is thatin some cases one electronic hit provided more than one piece of IC information. For example,experience (one hit) was located in the following sentence:

Previous experience includes twenty years in a large home shopping and consumer servicesgroup and ve years as Chief Executive of a UK home shopping plc. [Source: Directorsand Officers, p. 9]

In the manual search, this sentence was recorded as providing 4 pieces of information in relationto employee experiencetwo types of experience for two time periods.


The electronic search terms used related to the three IC categories of human capital, structuralcapital and relational capital. However, an electronic hit on one IC sub-category was often foundto yield information relating to another. For example, markets/marketing in the relational capitalcategory was located twice in the following sentence:

Previous experience includes a career in advertising, marketing and market research.[Source: Directors and Officers, p. 9]

In the manual search, this sentence provides no information in connection with relational capitalbut documents employee experience, part of human capital. Findings highlight the inadequacyof electronic searches in the investigation of IC disclosures in annual reports. They substantiateMilne and Adlers (1999) suggestion that an understanding of disclosures is best achieved byconsideration of whole sentences. The findings of very low IC disclosure from the study byBontis (2003) which uses an electronic search and a limited item list now appear less surprising.

5. Summary and conclusions

IC is now acknowledged as the major contributor to the market value of many companiesoperating in service and knowledge industries, yet it is generally not reflected on the balancesheet. The opportunity exists, however, to report IC in the narrative sections of corporate annualreports. Prior content-analytic studies have generally concluded that these sections contain lowlevels of IC disclosure (e.g. Bozzolan et al., 2003; Brennan, 2001; Guthrie & Petty, 2000). This issomewhat surprising given the theoretical motivations that companies have for disclosing IC infor-mation. However, specific issues arising in the use of content analysis to identify IC disclosureshinder the interpretation and comparison of findings across studies. These issues include: conceptboundary problems and coding reliability; annual report material analysed; volume of disclosure;type of disclosure; the unit of analysis and unit of measurement; and manual versus electronicsearching.

This paper highlights and illustrates the problems associated with these specific issues throughan analysis of prior studies and the use of an illustrative example (Next plcs 2004 annual report).Next plcs 2004 annual report was manually analysed for IC information using a detailed list ofIC sub-categories (Beattie & Thomson, 2004). An electronic keyword search was also conductedto facilitate comparisons between manual and electronic methods.

The depth and breadth of the IC concept is evident in the academic literature. The concept hasdeveloped over time, yet the requirement of a generally agreed taxonomy of IC terms (as calledfor by Zambon, 2005) has still to be met. Consequently, the identification and categorisation of ICinformation using content analysis to identify IC disclosures is, at present, highly subjective. Thesubjectivity in the categorisation of information found in Next plcs annual report is illustratedin the present paper. In this context, transparency is essential for meaningful interpretations andcomparisons of findings across studies. To date there is a lack of explanation of the nature of ICinformation allocated to IC categories and sub-categories in studies conducting manual contentanalysis. Future studies would benefit from increased transparency in relation to the IC informationfound and how it is categorised. Not only would this action help to address the problems ofcomparability across IC disclosure studies (as called for by Abeysekera, 2006), but it wouldalso provide real practical examples of the IC concept. Through the communication of practicalexamples and awareness of researchers perceptions, the IC concept would be given the opportunityto evolve, and shared meanings the opportunity to developa necessary requirement for contentanalysis to be effective (Gray et al., 1995). It would assist in the development of reliable coding


instruments, which in conjunction with the demonstration of consistent coding decisions wouldfurther address comparability issues.

In the analysis of Next plc, IC disclosures were found across the various sections of the annualreport, including sections that are subjected to mandated requirements. It is essential that futurestudies clearly define the boundaries of the material being analysed to permit valid comparisons.It is important to recognise that the boundaries have a direct impact on the volume of IC found.

Multiple disclosures of IC information were evident for 45 out of the 59 IC sub-categoriesidentified in the analysis of Next plc, however only 3% of all disclosures were found to beduplicates. The maximum number of disclosures for a single item was 137 in relation to remu-neration procedures followed by 106 for distribution channels. If multiple disclosures areignored, neither the absolute nor the relative volume of information provi