what financial and non-fínancial information on intangibles is...

Accounting and Busine.'is Research, Vol. 38. No. 3 2008 International Accounting Policy Forum, pp. 217-256 217

What financial and non-fínancialinformation on intangibles is value-relevant?A review of the evidenceAnne Wyatt*

Abstract—This paper evaluates what we have learned about the relevatice and reliability of financial and non-fi-nancial information on intangibles from the value-relevance literature. Because value-relevance studies do not eas-ily allow judgments about the reliability of information on intangibles, and this is an issue of central interest, thispaper takes a rather wide look across a range of literatures to try to piece together some indirect evidence on bothrelevance and reliability. The evidence from a package of value-relevance and triangulation studies suggests re-search and development (R&D) is generally not reliably measured and may be less relevant in some contexts thanothers as well (e.g. established versus growth firms). Further purchased goodwill and some non-financial measuresof brands and customer loyalty do not appear to be reliably measured. While a large number of financial and non-financial information is value-relevant, it is difficult to make categorical judgments about most other items, as dif-ferences in value-relevance could be due to different relevance or reliability, or both. Several rich areas for futureresearch include designing direct tests of reliability, focusing on settings where intangibles are changing due toshocks, finding new economic benchmarks to test reliability, and studying the impact of accounting discretion andfactors such as strategy and capabilities on value-relevance tests of information on intangibles. Two regulatory is-sues arising from this review paper are the gap in the reporting of separate line items of expenditures on intangi-bles; and the possibility that giving management discretion, with regulatory guidance, to report intangibles mightfacilitate more value-relevant information on intangibles.

Key words: value-relevance, intangibles, accounting regulation

1. IntroductionThis paper evaluates what we have learned aboutthe relevance and reliability of financial and non-financial information on intangibles from thevalue-relevance literature. The paper provides arather wide-ranging view across the literatures inseveral disciplines, including economics, account-ing, and management. This approach is motivatedby the difficulty of testing for reliability using thevalue-relevance design. Since the reliability of in-formation on intangibles for valuation is an issueof central interest, this paper provides an indirect,second-order assessment of reliability by piecingtogether the evidence from a large number ofstudies with different research questions and de-signs, and different measures of intangibles infor-mation and value.

Information is value-relevant when it is associ-

*Anne Wyatt is Associate Professor at the School ofAccounting, University of Technology, Sydney. E-mail:[email protected]. She is grateful to the Editor, PaulineWeetman, and an anonymous referee, and Pete Clarkson,David Emanuel, Paul Healy, Richard Macve, Andy Stark,Stephen Zeff, and ICAEW staff, Robert Hodgkinson andBrian Singleton-Green, for helpful comments on drafts of thispaper. Support from the ICAEW and the School ofAccounting, University of Technology, Sydney is gratefullyacknowledged. Thanks to Mark Russell for the timely researchassistance.

ated with investors' valuation of the firm as re-flected in the firm's stock price. However, intan-gibles are generally unverifiable and uncertain bynature. Regulators and some researchers thereforehold reservations about financial disclosures onintangibles, including the costs and benefits to thefirms, and the reliability for investors.' Value-rel-evance studies provide some insights on theseconcerns. If the information items of interest aresignificantly associated with the information set

' For example, the Financial Accounting StandardsCommittee of the American Accounting Association com-mented on the Financial Accounting Standards Board's'Proposal for a New Agenda Project: Disclosure ofInformation about Intangible Assets Not Recognized inFinancial Statements, August / 7 ' , September 28, 2001: 'If theFASB is to step in and (say) mandate the disclosure of certaininformation on intangibles, a question that seems relevant is:why have firms chosen not to disclose this information volun-tarily. One answer is that there are likely to be costs associat-ed with such disclosures, including both costs associated withmeasuring intangibles and proprietary costs of disclosing suchinformation to competitors. Another answer may be that thebenefits of these disclosures are not very large, perhaps be-cause these disclosures are not very informative to investorsdue to low relevance or imprecise measurement. Whatever thecase, it seems to us that the relatively low levels of voluntarydisclosure in the intangibles area raise the possibility that dis-closures in this area do not provide net benefits.'

218 ACCOUNTING AND BUSINESS RESEARCH

that was used by investors to value the company,we can infer that the information is relevant (ei-ther directly or indirectly in a confirmative sense)for valuing the company. This statistical associa-tion with stock price also suggests that the infor-mation is reliable enough to be value-relevant.However, there are limits to what can be learnedabout reliability. Conclusions from value-rele-vance studies are not reliable if important factorsare left out of the tests. We cannot tell whether in-vestors actually used the information item of in-terest or whether one accounting method isoptimal relative to another, or easily understandwhy information is value-relevant. Overall, it isdifficult to directly test reliability and only a fewstudies do this.

Intangible investment is increasingly viewed bysome as an important category of investment.^This view reflects an increasing tendency fortechnology to be embodied in intellectual proper-ty (IP) and labour where previously it resided infixed assets. Without a long and comprehensivefinancial data series, however, it has proved im-possible to conclusively test this hypothesis. At amore fundamental level, it is easy to argue that ex-penditures on intangibles are important becausethe stock of physical resources is finite and eco-nomic activity can only be sustained by the appli-cation of intellectual inputs (Webster, 1999). Thispresents a prima facie case for the value-rele-vance of at least some of the firm's expenditureson intangibles and the non-financial informationbearing on the value and uncertainty associatedwith these expenditures.

Intangibles are also at the centre of an informa-tion gap that arises from the forward looking anduncertain nature of economic activity. In fact, allof the firm's investments, tangible and intangible,are uncertain by definition, since investment ex-penditures are outlays made in anticipation of fu-ture benefits (Fisher, 1930). However, whiletangible assets tend to be standardised with controlrights and a predictable stream of inflows, intangi-ble assets tend to be heterogeneous and uncertainand subject to long development periods withoutcontrol rights (Webster, 1999). This uncertaintyengages managers and investors in a constantsearch for information to improve their foresightand decisions. Management have a central role ingenerating estimates of the future as they designand execute their firm's strategy (Knight 1921,Part III). These estimates embody a range of ex-pectations about investor and consumer behav-iours and wider economic conditions; and they arepartially (and imperfectly) revealed in GenerallyAccepted Accounting Principles (GAAP) financialreports and the firm's interactions with the busi-ness environment.

In this paper, the literature is organised into six

categories of intangibles that relate to the firms'core value drivers and five different measurementapproaches that reflect the influence of GAAP, re-searcher defined intangibles and non-financialinput and output metrics. The review in this paperindicates that expenditures on R&D are value-rel-evant but appear to be less reliable than tangibleitems and to vary in the ability to signal future ben-efits. Purchased goodwill and some non-financialmeasures of brands and customer satisfaction areusually value-relevant but do not appear to be reli-able indicators of future benefits. A wide variety ofother financial and non-financial information onintangibles is value-relevant. However it is diffi-cult to know whether variation in the size of the re-gression coefficient is due to differences inrelevance or reliability, or both. Overall, it is diffi-cult to obtain robust tests of reliability and ad-dressing this gap is a key area for future research.

One gap in financial information that is evidentfrom the review in this paper is the reporting ofseparate line items of expenditures on intangiblesin the income statement. It is often argued thatvalue creation is reflected in earnings. However,earnings are a summary number that is not neces-sarily useful for addressing the question of howvalue is created. For this purpose, informationabout value driving expenditures is relevant. Thereis also evidence that accounting regulators mightbetter facilitate value-relevant disclosures on in-tangibles if they give discretion to management toreport their firm's economic reality.

Section 2 begins with some background on theclassification and economic properties of intangi-bles. Section 3 provides an overview of studiesthat examine the value-relevance of financial andnon-financial information relating to intangibles.These sections consider how we might interpretthis evidence in the light of wider economic con-ditions and other factors, such as omitted variablesrelating to the firm's competencies and strategy.Section 4 concludes with a discussion of what wehave learned from the value-relevance tests, alongwith research design issues and some directionsfor future research.

2. BackgroundThis section outlines the categories of intangiblesused to structure the review in this paper, the eco-nomic properties of investments in intangibles, andthe design of value-relevance studies, including adiscussion of the concepts of relevance and relia-

^ This trend has been ascribed to authors including the follow-ing (see Webster 1999): Kendrick (1972), Caves and Murphy( 1976), Magee ( 1977), Grabowski and Mueller ( 1978), Reekie andBhoyrub (1981), Rugman (1981), Hirschey (1982), Caves (1982),Cantwell (1989), and Abramovitz (1993).

Vol. 38 No. 3 2008 International Accounting Policy Forum. 219

bility, which are the focal point of the value-rele-vance literature.

2.1. Classification of intangiblesThe literature review in this paper canvasses

a broad range of studies from the accounting,economic and management literatures. Papers in-cluded in the study are not exhaustive but are rep-resentative of the research questions and researchdesigns observed in the different areas of study.The papers are classified according to six cate-gories of intangibles as follows:

Technology resources1. R&D expenditures and related IP

Human resources2. Human capital

Production resources3. Advertising, brands and related IP4. Customer loyalty5. Competitive advantage6. Goodwill

These six categories of intangibles relate tothree broad categories of the firm's resources:technology, human, and production resources.^Category one, R&D expenditures and the IP off-shoots, such as patents, are aimed at developingtechnology, which is defined as a body of knowl-edge about how to do or make something(Metcalfe, 1998). Category two, human capital,relates to the resource generated by investing inemployees. Categories three-six (advertising,brands and related IP; customer loyalty; competi-tive advantage; and goodwill), relate to produc-tion resources the firm has generated or acquiredfrom prior periods' intangible investments. Thesesix categories overlap, but, in the big-pictureview, relate to the three broader elements of thefirm's activities and resources, as outlined above.The six categories of intangibles are not exhaus-tive. Arguments can be made that other categories,such as environmental and social responsibility,are also important. The rationale for what is in-cluded is the need to be selective given the largenumbers of papers but at the same time provide an

' Researchers have come up with a variety of classificationsof intangibles which are often bundled under the label of 'in-tangible capital'. See, for example, Abemathy and Clarke(1985), Webster (1999), Commission of the EuropeanCommunities (2003), Ashton (2005), and Hunter, Webster andWyatt (2005).

"* Relevant information has predictive value and/or confir-matory value, and therefore has the ability to influence theeconomic decisions of users and is provided in time to influ-ence those decisions. Reliable information is free from delib-erate bias and material error and is complete. If information isreliable then GAAP maintains that it can be depended on tofaithfully represent what it purports to represent or could rea-sonably be expected to represent.

accurate account of the types of information thatresearchers have studied.

These six categories of intangibles are further par-titioned according to five measurement categories.

1. Management reported assets (financial measures);

2. Researcher estimated assets (financial measuresor non-financial metrics);

3. Annual expenditures (financial measures);

4. Input metrics (e.g. non-financial metrics, suchas the number of scientists);

5. Output metrics (e.g. non-financial metrics, suchas the number of patents).

The measurement categories reflect (1) the eco-nomics of the value creation processes and theresearchers and practitioners' interests in the iden-tification of value drivers and their empiricalmeasures; (2) the influence of GAAP on the re-porting of intangibles and the research problems ofinterest to practitioners and researchers; and (3)the influence of management discretion.

Annual expenditures and management reportedassets

The logical starting point for researchers and in-vestors is to identify how much has been spent onintangibles and the types of activities and rents in-volved. Once this information is known, rates ofreturn from different types of expenditures can becomputed, illuminating some of the drivers of firmperformance.

However, this financial data is not availableunder current GAAP. Further, the factors thatcause expenditures to give rise to future rents arenot fully understood and change over time. In ad-dition, management do not necessarily think interms of 'intangibles' and do not always have in-centives to voluntarily provide expenditures dataof this type (e.g. for competitive reasons). Lineitem disclosures of expenditures on intangibles aretherefore primarily limited to R&D with some re-search on advertising and labour costs where pos-sible (e.g. advertising costs have not beenavailable in the UK and are only patchily disclosedin the US). Hence, the research able to be under-taken on annual expenditures on intangibles islimited to a narrow range of expenditures.

Management reported assets are also limitedunder GAAP due to regulators' concerns about thereliability and verifiability of these items.'* The rel-evance of the information for evaluating perform-ance and value is seldom disputed. Reliability isthe regulator's concern: does the recorded numberreflect expected future benefits and what is theprobability these expected benefits are realisable?As a result of these concerns, GAAP is conserva-tive. This conservatism manifests as a two-wayclassification, acquired assets and internal


expenditures, from which only the acquired intan-gibles can usually be recorded as assets.^ This lim-its the research able to be undertaken on internallygenerated intangible assets. There are insightsavailable on the value-relevance of managementreported assets of both types (acquired and inter-nal) from settings and time periods where GAAPallowed more liberal reporting of intangible assets(e.g. Australia, prior to the 2005 adoption of inter-national accounting standards, or Ely and Waymire(1999), who study New York Stock Exchange list-ed companies allowed to report intangibles in the1927 pre-SEC era).

Researcher estimated assets, and input and outputmetrics

Due to gaps in financial reporting under GAAP,to attempt the difficult task of studying optimal ac-counting methods, or to study the value drivers,some financial and non-financial measures of in-tangibles have to be estimated by the researchersthemselves (e.g. the construction of R&D assetsfrom R&D expenditures in Lev and Sougiannis,1996. or the managerial skills measure in Abdel-khalik, 2003). Researchers studying value creationprocesses also focus on input or output metricssuch as the number of scientists associated withthe company (input metric: see Darby, Liu andZucker, 1999) or patent metrics as a measure oftechnological innovation (output metric: see Hall,Jaffe and Trajtenberg, 2005).

In summary. Table 1 in the appendix summaris-es the literature using the six intangibles categoriesand five measurement classifications. The studiesare also grouped on the value-relevance measureemployed by the researcher: stock price levels,stock returns or financial performance measures.Percentages of firms with significant coefficientsfor the financial or non-financial information onintangibles and coefficients that are smaller thanthose on other tangible assets in the test (as only arough guide to reliability) are provided in the bodyof the table.

We now briefly look at the economics of intan-gibles in the next section to get a feel for the prop-erties that impact the relevance and reliability ofinformation relating to intangibles.

2.2. Economic properties of intangiblesExpenditures on intangibles are usually invest-

ments since they are made in anticipation of futurebenefits (Fisher, 1930). Expenditures such as R&Dand advertising may be employed directly in pro-duction to generate innovations and product mar-ket share. In addition, these expenditures can alsogive rise to intermediate (produced) assets, whichare used in production. For example, intellectualproperty (IP) outputs are used in production togenerate future rents in various ways, such asthrough the ability to charge a price premium or

control costs. R&D and advertising can generatepatents, trademarks, brands or designs that provideproperty rights over innovations or generate mar-ket share and thereby permit the firm to appropri-ate the expected benefits from the earlier R&D andadvertising investments.

There are complex lead-lag relations betweenearly investments, intermediate (produced) assets,capital investments to produce the goods, and fu-ture expected benefits that are challenging for re-searchers to observe and model. Successfulinvestments generate a range of intangible assetsand future rents for the firm right across the valuechain. But not all of the firm's outlays are success-ful in creating value.

From the investors' perspective, stock price re-flects the capital market's expectation of the firm'sfuture cash flows from the firm's investments.Investor expectations are formulated from a di-verse set of information. This information set pres-ents some problems for investors and managers.Chief among these is the fact that this set can neverbe complete because the future is uncertain.* Theavailable information is imperfect and not held asa complete unit. Instead, the information exists as'dispersed bits of incomplete and frequently con-tradictory knowledge' in the hands of individuals(Hayek, 1945:519).

'At the bottom of the uncertainty problem ineconomics is the forward-looking character ofthe economic process itself. Goods are producedto satisfy wants; the production of goods re-quires time, and two elements of uncertainty are

^ Under the previous UK standard, FRS 10 Goodwill andIntangible Assets, which is now superseded by IFRS 3Business Combinations, intangible assets are non-financialfixed assets that do not have physical substance but are iden-tifiable and are controlled by the entity through custody orlegal rights (para. 20). Internally developed intangible assetscan be capitalised if there is control and a readily ascertaina-ble market value. According to FRS 10, 'readily ascertainablemarket value' is the 'value of an intangible asset that is estab-lished by reference to a market with a homogenous populationof assets and the market is active as evidenced by frequenttrades for that population of assets.' Since active markets ofthis type generally do not exist for intangibles, only acquiredintangibles can be routinely capitalised. SSAP 13 Accountingfor Research and Development allows capitalisation of devel-opment costs only if future benefits are virtually certain. TheInternational Accounting Standards Board in IAS 38Intangible Assets has asymmetric rules for acquired intangi-bles and internal expenditures on intangibles. Under IAS 38,there is a presumption that acquired intangible assets aremeasured reliably and are therefore capitalisable assets.However, to be treated as assets, internal expenditures on in-tangibles must pass six additional tests as set out in IAS 38paragraph 57.

^ Shackle (1974: 3) points out that managers do not actual-ly know their circumstances in the sense of having complete orperfect information; there is a ' . . . lack of knowledge, unlikeactuarial probability calculations which require substantialknowledge' (Shackle as cited in Ford (1994: 82); see Knight,1921).

Vol. 38 No. 3 2008 International Accounting Policy Forum.

introduced, corresponding to two different kindsof foresight which must be exercised: First, theend of productive operations must be estimatedfrom the beginning. It is notoriously impossibleto tell accurately when entering upon productiveactivity what will be its results in physical terms,what (a) quantities and (b) qualities of goodswill result from the expenditure of given re-sources. Second, the wants which the goods areto satisfy are also, of course, in the future to thesame extent, and their prediction involves uncer-tainty in the same way. The producer, then, mustestimate (1) the future demand which he is striv-ing to satisfy and (2) the future results of his op-erations in attempting to satisfy that demand.'Knight (1921, III.VIII.8)

The extent of the problem for managers dependson factors such as the technical difficulty of thefirm's products and processes, the extent to whichthe firm's assets and routines are standardised andpredictable versus non-standard and unpredictable,and the strength of property rights (Dosi, 1988).Investors have the problem of decision makingunder uncertainty, which is compounded by infor-mation asymmetry between managers and in-vestors, and among investors themselves. Investorsalso have different levels of sophistication and in-centives to search for value-relevant information.Some investors are therefore more informed thanothers. These information asymmetries are exacer-bated by the natural optimism of managers abouttheir firm's prospects.

What economists have learned about productionand growth is important for value-relevance studiesbecause information is value-relevant only if it iscapable of reflecting some aspect of the firm's eco-nomics. For example, we expect expenditures ontraining to be value-relevant if the expenditures areassociated with increases in the skills and produc-tivity of employees. Economists find that expendi-tures on intangibles are important for building thefirm's capabilities to exploit emerging opportunitiesand meet profitability goals (Cohen and Levinthal,1989).^ These expenditures help to differentiate thefirm's value creating activities and routines so theyare hard for rivals to copy and help reduce the num-ber of uncontrolled factors impacting the firm's op-erations (Webster, 1999). As a result, intangiblesare contingent by nature and intrinsically exposedto economic states. For example, the benefits fromtraining staff are contingent on the state of thelabour market as well as the firm's own ability to at-tract, retain and motivate the employees.

Expenditures on intangibles are distinguishedfrom tangible investments based on the heteroge-neous and non-standardised nature of intangibleinvestments (Webster, 1999). Heterogeneity andstandardisation are a function of how often taskshave been performed before and the ease of copy-

221

ing. Plant, property and equipment are relativelystandardised compared with payments for intellec-tual inputs from employees and payments for pro-duced intangible assets from outside the flrm. Theoutputs produced from the intellectual inputs ofemployees, and intangible assets purchased fromoutside the firm, are more difficult to control andpredict compared with the outputs from machines(Webster, 1999).

A further source of uncertainty is the inability toassign property rights over people and over sometypes of assets (e.g. R&D). The value is often tiedup with people who cannot be owned or attributa-ble to rents that are easily dissipated by rival firms(e.g. brands) (Webster, 1999). Property rights overintangible investment may be unavailable for ex-tended periods while a project is developed. Bycontrast, investment in tangible assets occurs whenthe company is ready to produce products.

In summary, the economic properties of intangi-bles reflect several fundamental uncertainties, in-cluding an intrinsic exposure to changingeconomic states, an unstandardised and heteroge-neous nature, and difficulty obtaining propertyrights. Expenditures on intangibles are thereforeless reliable by definition compared with tangibleassets. Accordingly, while financial informationon intangibles is likely to be relevant for valuingthe firm, it is less likely to be reliable, especially inthe earlier stages of the investment. Non-financialinformation is likely to be value-relevant if it issufficiently salient to the firm's economic realityand precisely measured to be informative about theearnings effects of the firm's interaction with itsenvironment.

2.3. Relevance, reliability and value-relevancetests

value-relevance studies test for an associationbetween information items of interest and a stockprice or financial measure of value, for example,the market value of equity, stock returns orfuture earnings. The tests rely on stock marketefficiency.^

' The intangible investment is not confined to R&D but in-volves a bundle of expenditures and activities of differenttypes, including strategic planning, design, feasibility, devel-opment, production, marketing, distribution, advertising, cus-tomer service, management of intellectual property portfolios,and building of organisation and information infrastructuresand routines (Abernathy and Clark, 1985).

"The assumptions include (see Lev and Ohison, 1989:297-298): prices in the pre-disclosure economy are unbiasedestimators of the prices in the post-disclosure economy; indi-viduals have homogeneous information and identical beliefsabout the implications of the intangibles information; and theeconomy is efficient in the sense that more information is(Pareto) better than less information so that nobody is worseoff while additional trading on the information would makesome better off.


The regressions used for value-relevance testsassociate the value-relevance measure on the left-hand side of the equation with the informationitems of interest and other variables on the right-hand side of the equation.

Market value of equity =bg + (,|Book value of equity +

¿"jlnformation of interest + error

Stock return =bg + ¿/Earnings +

in Earnings +fcjinformation of interest + error

Financial performance =

¿>2lnformation of interest + error

( 1 )

(2)

(3)

The existence of a statistical association is deter-mined by looking at the estimated regression coef-ficients, the ' è ' s in equations ( l ) - (3 ) and testingwhether they are significantly different from whatwas expected. If the test statistic is significant, wecan infer that the information of interest is associ-ated with the value-relevance measure on the left-hand side.

2.5.7. What can be inferred?We can infer that the information item of interest

is associated with the information set that investorsused to value the firm's equity, and the informationitem is therefore value-relevant. But there are atleast two things we cannot infer. First, we cannotinfer that investors actually used the informationof interest to value the firm. Second, we cannotinfer from the statistical test alone that the infor-mation of interest causes the level of market value,changes in stock price, or financial performance.The statistical test only tells us whether the value-relevance measure and the information of interestare statistically associated.

It is also difficult to infer optimal accountingpolicies from value-relevance tests. Holthausenand Watts (2001) argue this is because value-rele-vance tests provide a statistical association that isnot backed up by theory and modelling of the un-derlying links between accounting, standard set-ting and value.

A factor adversely impacting the inferencesavailable from value-relevance studies is the prob-lem of omitted correlated variables. That is, factorsof varying importance that are associated with theleft- and right-hand side variables are not includedin the equation. For example, Holthausen andWatts (2001) point out that expected future rentsare omitted from value-relevance regressions.Omissions like this can distort the 'b' coefficientsin equations ( l ) - (3 ) and lead to erroneous conclu-sions. One common problem is the effect of differ-

ences in firm size on the test results. Size differ-ences can produce significant results that have lit-tle to do with the intrinsic attributes of the item ofinterest. For example, a significant test for the ' è 'coefficient on goodwill might just show that largecompanies have more goodwill.

There are also trade-offs in the choice of thevalue-relevance measure. Shevlin (1996) points outthat, for investment variables, the sign is more in-tuitive in the stock levels regression compared withthe stock returns model. That is, an unexpectedchange in investment can be good or bad news, buta significant association for the total amount of theinvestment (e.g. R&D expenditures) is expected tobe positively related to value-relevance metrics.Further, there may be little change in the variable ofinterest in a narrow return interval (e.g. change instock price over a quarter or a year). For example,customer satisfaction for an established brand com-pany such as Coca-Cola might be highly value-rel-evant in a stock price levels regression but thevalue might change very little on an annual basis.FinaWy, financial measures are limited to the extentthat they do not reflect the capitalised value of theexpected benefits from intangible assets. We needto consider the impact of these trade-offs in draw-ing conclusions from value-relevance studies.

2.3.2. Concepts of relevance and reliability invalue-relevance studies

Value-relevance tests are joint tests of relevanceand reliability (e.g. Barth et al., 2001). However, itis difficult to infer the amount of 'value-relevance'that is due to relevance and the amount that is dueto reliability.

Figure 1 graphically depicts the relevance andreliability concepts. Conceptually, relevance re-lates to two aspects of the underlying economics ofthe investment. The first is a value construct ofsome kind, such as expenditures on R&D or ac-quisition of goodwill. The second is the process bywhich value is expected to be created (e.g. theR&D is expected to produce value by generating anew large pharmaceutical molecule to be used in adrug with a known purpose). Relevance of infor-mation is decreasing to the extent that the valuecreation process is ill-defined. An example is pur-chased goodwill, which is a residual from a com-mercial transaction that relates to an unspecifiedvalue creation process. When the relevance link isweak (as defined in terms of an ill-defined valuecreation process), such as in the case of purchasedgoodwill, then the next link in the value-relevancechain, the reliability link, is also going to be weak.This way of thinking benchmarks reliabilityagainst the uncertainty of the value creationprocess. That is, it is going to be difficult to meas-ure something that defies definition in terms ofhow the value is going to be created.


Figure 1

Retevance Reliability Value relevance

Valueconstruct

i 1Value

creationprocess

Measure

r

Value

1. Impacted by uncertainty about link from valueconstruct to value creation (e.g. purchased goodwill)

2. Impacted by GAAP (e.g. full expensing of R&D)

3. Impacted by management discretion

In Figure 1, a reliable measure is capable of con-veying information about the future benefits ex-pected to flow from an underlying investment.Usually, the measure relates to a quantum of ex-penditure, but the measure can (less commonly) bea revaluation amount. Reliability refers to twolinks. The first is the relevance link. For reliablemeasurement, there must be a reasonably well de-fined value construct link to a known value cre-ation process. Second, a measure is required that iscapable of reflecting the economic substance ofthe value construct and process. An example of ameasure that falls down at the reliable measure-ment stage is R&D expenditures. Specifically, theindividual firm aggregates various types of expen-ditures into their R&D measure and has a goodidea of the value creation process. However, forexternal parties to the firm, the R&D expendituresnumber reported in the income statement are toogeneral to convey highly reliable information toinvestors about value creation and expected bene-fits. This is because the R&D aggregates expendi-tures relating to different kinds of undisclosedvalue constructs and value creation activities, notall of which are going to lead to expected futurebenefits. Hence, conceptually, the R&D expendi-tures provide relevant information about value cre-ation, but the measure is not a reliable indicator offuture rents.

Reliability is affected by a number of factors.One is GAAP rules. For example, as discussed, theaggregate nature and full expensing of R&D ex-penditures adversely impacts the reliability of theR&D measure. A second factor is economic un-certainty, which adversely impacts reliability if itcauses the link from the value construct to valuecreation to be ill-defined (e.g. there is uncertaintyabout how basic research will generate value)

and/or generates uncertainty about the probabilityof future benefits. Reliability is also affected by athird factor, management discretion. The effect ispositive if management have incentives to com-municate credibly with investors and potentiallynegative if management's interests are not alignedwith shareholders' interests.

To distinguish relevance and reliability effects invalue-relevance studies, and to obtain a direct testof reliability, it is necessary to develop the rele-vance and reliability links in setting up the study.This is a difficult task. One of the few studies thatdo this is Healy, Myers and Howe (2002). Theysimulate financial accounting data for a sample of500 pharmaceutical companies. Simulation en-sures that the value creation process, the R&D ex-penditures, and the firm value are known. As aresult, Healy et al. (2002) are able to examine thevalue-relevance of R&D accounted for in differentways under GAAP, as well as the effects of eco-nomic uncertainty and management discretion.

Another way to think about reliability is to com-pare the regression coefficient for the intangibleitem with the size of the coefficient for more reli-able assets. As a rough guide, we expect more re-liable information to have a larger coefficient.However, this comparison is difficult because sizedifference in the b' coefficients could be due todifferences in relevance or reliability, or both. Veryfew of the value-relevance studies directly addressthe question of reliability.

As a second-best alternative, we can get someindirect insights by selecting a wide range acrossthe literature from different disciplines (e.g. eco-nomics, accounting, management, and marketing)to let the overlapping nature of the studies tell thestory. This is the approach taken in the review fol-lowing in Section 3.


3. What information on intangibles isvalue-relevant?Section 3 reviews a cross-section of the literatureorganised according to the six categories of intan-gibles: R&D and related IP; advertising, brandsand related IP; customer loyalty; human capital;competitive advantage; and goodwill. Each cate-gory of intangibles is reviewed according to thefive types of measures that were introduced inSection 2.1.

3.1. R&D and related IPThis section considers R&D and its IP output,

mainly comprising patents. IP refers to the lawsgoverning creations of the mind, including inven-tions, literary and artistic works, and symbols,names, images, and designs used in commerce.The two categories of IP are industrial property(patents, trademarks, copyright, designs, circuitlayouts, and plant breeder's rights) and copyright(literary and artistic works, such as novels, poemsand plays, films, musical works, drawings, paint-ings, photographs, sculptures, and architectural de-signs).' IP is designed to grant a short-termmonopoly so the entrepreneur can capture valuefrom their investments.'"

Most research focuses on patents becausepatents measure knowledge creation.

Expenditures on successful R&D give rise toproduct and process innovations and a productpipeline to ensure a sustainable earnings streaminto the future and value. R&D expenditures areinputs that also give rise to IP outputs, which maycomprise patents, trademarks and designs. IP isemployed in production to produce goods andservices, help appropriate expected benefits due tothe short-term monopoly, and potentially support aprice premium.

3.1.1. R&D and IP — management reported assetsR&D and IP are important factors of production,

particularly for high-technology companies(Griliches, 1990), but are not routinely capitalis-able under GAAP in most countries (e.g. usuallyexpensed under IAS 38 Intangible Assets). On theone hand, producing technology engages the firmin relatively unstructured and uncertain problem-solving activities (Dosi, 1988). Adopting technolo-gy also engages firms in costly problem solvingand learning activities. The uncertainty associatedwith R&D and the associated projects leads to fullexpensing of R&D. Financial statements of R&Dintensive companies usually do not fully capturethe economics of these activities.

Consistent with this idea, Kwon (2001) findsmanagement reported GAAP numbers are lessvalue-relevant for high-technology firms than forlow-technology firms. His evidence suggests thatthis difference is due to the greater impact of

GAAP accounting conservatism in high-technolo-gy firms compared with low-technology firms.

Consistent with uncertainty about future eco-nomic benefits, Wyatt (2005) finds that R&D as-sets are not significantly associated with stockreturns in the Australian setting. This evidence isfrom the time period prior to the 2005 adoption ofinternational financial reporting standards (IFRS)in Australia. At that time, companies were allowedto capitalise applied R&D expenditures, but it wasnot mandatory. There was no accounting standardfor identifiable intangible assets at the time. As aresult, management had wide discretion to choosenot to record R&D assets and instead record awide array of other intangibles assets that aremuch more informative about the value creationprocess." In contrast to the lack of R&D assetvalue-relevance, Wyatt finds that the identifiableintangible assets are significantly positively relat-ed to stock returns. Her evidence further suggeststhat the identifiable intangible assets are signifi-cantly associated with technological factors driv-ing the firm's production function. However, theR&D assets are not. This finding suggests why theR&D assets are not value-relevant.'^

By contrast, Deng and Lev (1998) find R&D-in-process assets purchased and valued as part of anacquisition are value-relevant. They use a sampleof 400 companies acquiring R&D in-process as-sets in the years 1985-1996. Both Wyatt (2005)and Deng and Lev (1998) use contemporaneous

' See http://www.wipo.int/about-ip/en/. Recent IP develop-ments also include technological protection measures (TPM)and digital rights management (DRM). TPMs are used in ma-terial such as sound recordings, films and computer software,as well as electronic artistic and literary works (e-books).DRM is technology used to control access to digital works ordevices, to protect copyright in those works or the works usedon the devices. For example, the iTunes store incorporatesDRM into its music, to restrict copying. There is no knownevidence on the value-relevance of these IP.

'" IP is designed to 'create a market for knowledge by as-signing property rights to innovators which enable them toovercome the problems of non-excludability while at the sametime, encouraging the maximum diffusion of knowledge bymaking it public' (Geroski, 1995: 97). See Griliches et al.(1987); Trajtenberg (1990); Austin (1993); Hall et al. (2001).

" To outsiders of the firm, R&D expenditures are a bundleof unknown expenditures with unknown links to future bene-fits. R&D assets are only marginally more defined: while cap-italisation of assets is a signal of future benefits beyond thecurrent period, the aggregate nature of R&D precludes in-vestors making precise links between the expenditures andvalue creation in the absence of other information about thefirm's R&D success rate.

' Wyatt (2005) finds that management's choice to recordintangible assets is associated with the strength of the technol-ogy affecting the firm's operations, the length of the technolo-gy cycle time, and property rights-related factors that affectthe firm's ability to appropriate the investment benefits. Theseeffects are more important than other contracting and sig-nalling factors consistent with the underlying economics oper-ating as a first-order effect as envisaged by GAAP.


Stock return as their value-relevance measure. Themain points of difference are the economic signif-icance of the acquired R&D-in-process assets,which represent an average 75% of the acquisitioncost. By contrast, the R&D assets reported byAustralian companies are internal expendituresthat comprise an average of 1% (median 0%) ofthe firms' total assets. Further, compared with theUS companies, the Australian companies had dis-cretion to report more informative intangible as-sets, reducing the need to communicate withinvestors via the more uncertain R&D assets.

The US Financial Accounting Standards Board(FASB) has required full expensing of R&D costssince 1974 (Statement of Financial AccountingStandards No. 2). In a break from this strict stan-dard, the FASB issued Statement of FinancialAccounting Standards (SFAS) No. 86 Accountingfor the Costs of Computer Software to be Sold,Leased, or Otherwise Marketed in August 1985,which allows the capitalisation of software devel-opment costs once technological feasibility of thesoftware is established. Givoly and Shi (2007) ex-amine the capitalisation of software developmentcosts by IPO firms. They test and find capitalisa-tion is associated with lower underpricing of thestock on the first day of trading. They interpret thisas evidence that the capitalisers are subject to lessuncertainty about the success of their software in-vestments but do not test whether these capitalisedcosts and lower underpricing predict future per-formance. An open question is whether capitalisedsoftware is enough information to distinguish theprospects of the newly listed software companiesthat face significant technological and competitivepressures in the industry, without the benefit of aminimum efficient scale.

For a sample of computer programming and pre-packaged software firms surviving for at leastthree years, Aboody and Lev (1998) find the soft-ware assets reported under SFAS No. 86 are value-relevant. However, only 25% of the total softwaredevelopment costs are capitalised by these firms.Eccher (1998) suggests one explanation for this isthe working model approach to software develop-ment where software is completed (technicallyfeasible) late in the development period, resultingin few capitalisable costs. The firm's software de-velopment strategy therefore appears to be rele-vant to a full interpretation of value-relevance testsfor software costs.

A further issue examined by Aboody and Lev(1998) is the petition by The Software PublishersAssociation in 1996 to abolish SFAS No. 86.Aboody and Lev suggest this petition was moti-vated by increasingly negative effects of the ac-counting standard on earnings, in a maturingindustry. However, they also find higher analystearnings forecast errors for capitalisers compared

225

with expensers, which suggests there is some at-tribute of the capitalising firms that makes theirearnings harder to predict compared with expens-ing firms. Given there is more information in cap-italised intangible assets for growth comparedwith steady state firms, one possibility is the capi-talisers are growth firms that are more risky thanthe expensing firms.

Ken Wasch, president of The Software PublishersAssociation, sheds some light on the risk issue. Heargues there is significant uncertainty even at thepoint of technical feasibility about the success ofthe software 'due to factors such as the ever-in-creasing volatility in the software marketplace, thecompression of product cycles, the heightenedlevel of competition and the divergence of tech-nology platforms' (Aboody and Lev, 1998: foot-note 3).

Economic studies show these factors do impactthe firm's success rate and performance (e.g.Audretsch, 1995). For example, Agarwal and Gort(2001) fmd the average cycle time from an idea toa viable product has decreased from 33 years in1900 to 3.4 years in 1967-1986. Further, as thecycle time decreases, costs and the level of compe-tition escalates (Scherer, 1966; Graves, 1989; Ittnerand Larcker, 1997).'^ In the US telecommunica-tions industry. Banker, Chang and Majumdar(1995) find increasing competition is associatedwith declining profitability. Consistent with techno-logical innovation and competition conditionsimpacting the success rate of R&D, the value-rele-vance of R&D varies across time periods and in-dustries (Hall, 2000) and across technologicalsectors (Greenhalgh and Rogers, 2006).''' Cohenand Klepper (1992), among others, show there areusually only a small number of high performingR&D companies in an industry. Consistent withthis evidence, Ceccagnoli, Arora, Cohen and Vogt(1998) find that differences in the firms' capabilitiesaffect their ability and incentives to generate inno-vations from R&D and absorb the innovations of ri-vals (i.e. take advantage of rivals' R&D). Severalstudies find financial leverage is negatively relatedto the level of R&D consistent with a life-cycle ef-fect (Bernstein and Nadiri, 1982; Hall, 1991).

'•* Examples of additional organisational competencies lead-ing to shorter cycle times and superior performance are the useof cross-functional teams, customer involvement in the inno-vation process, advanced design tools, and higher quality ofresulting products (see Ittner and Larcker, 1997).

'" Hall (1993a, 1993b) finds the late '60s and '70s were pe-riods of higher valuation of R&D in capital markets, which de-clined abruptly during the eighties in the United States. It isgenerally held this decline relates to company restructuringand the declining value of R&D assets due to rapid technicalchange in particular industries, including electrical equipment,computing, electronics and scientific instruments (i.e. R&Dbenefits were relatively short-lived).


In summary, R&D and IP are not commonly re-ported as assets by managers under GAAP, and,therefore, evidence on value-relevance is limited.Assets found to be reliable enough to be value-rel-evant include independently valued R&D-in-process, capitalised IP, as part of the identifiableintangible assets voluntarily recognised byAustralian firms, and capitalised software R&Dunder SFAS No. 86. R&D assets are not value-rel-evant in the Australian setting, when companiescould report more informative identifiable intangi-bles, which are presumably more reliable thanR&D assets (i.e. some of the identifiable intangi-bles are output from successful R&D). No otherspecific inferences about reliability are possiblefrom this evidence. Finally, a range of factors areimportant for understanding the probable successand hence reliability of R&D, including cycletime-based competition, industry structure, firm-specific capabilities, life-cycle stage, and technol-ogy-conditions.

3.1.2. R&D and IP - researcher estimated assetThe US FASB has been sceptical about the reli-

ability of R&D. In SFAS No. 2 Accounting forResearch and Development Costs (para. 41), theFASB states that a 'direct relationship between re-search and development costs and specific futurerevenue generally has not been demonstrated'. Levand Sougiannis (1996) examine this propositionby estimating the R&D assets and amortisationthat would have been reported by US companieshad they been allowed to capitalise R&D. They re-state the earnings and book value of shareholders'equity using these estimates of R&D assets andamortisation and find the {pro forma) R&D assetsare value-relevant. They also find an associationwith future period stock returns, suggesting thatthe pro forma R&D is relevant but not that reli-able. This forward relation leads them to concludethat R&D intense stocks either are mispriced be-cause investors do not understand conservative ac-counting or attract a risk premium due to theuncertainty associated with the R&D outcomes.

The risk premium conclusion is consistent withBoone and Raman's (2001) evidence that R&D in-tensive firms have higher bid-ask spreads com-pared with less R&D intensive firms. Chambers,Jennings and Thompson (2002) argue that, if thefull expensing effects of R&D on the financialstatements mislead investors, then excess returnsearned from trading strategies associated withR&D intensive firms may be reduced or eliminat-ed by alternative R&D accounting policies thatbetter reflect the expected future benefits of R&Dactivities. They tested this hypothesis, and theirevidence suggests the ability to earn excess returnsfrom trading on R&D intensive companies is notdue to GAAP-induced mispricing. The excess re-

turns may therefore be a risk premium consistentwith the economic properties discussed in Section2.2. Lev, Sarath and Sougiannis (2005) also pro-vide evidence that the firms with a high growthrate of R&D relative to their profitability are sys-tematically undervalued. Both the risk and mis-pricing conclusions are consistent with R&Dbeing a relevant but not reliable indicator of ex-pected inflows from R&D. It is possible that therisk is at least partly due to the aggregate nature ofR&D, where the firm bundles a range of expendi-tures whose identity and links to future benefits arenot visible to outsiders.

The studies reviewed so far suggest that R&D isreliable enough to be value-relevant in general.However, as discussed at the beginning ofSection 3, it is difficult to obtain direct tests of thereliability of R&D using stock price or financialmeasures as an economic benchmark.

One of the few studies able to provide direct in-sights on the reliability of R&D is Healy, Myersand Howe (2002). They use a simulation modei fora pharmaceutical company to generate 32 years ofdata for 500 companies from formation to steadystate. Parameters from these processes are used toconstruct a cash flow model and financial state-ments. The model simulates the drug discovery(R&D input) process (year 1), to commerciallaunch of a series of products (year 14), to maturi-ty and expiry of patent (26 years), using underly-ing costs, probabilities of success and revenues.The model allows for some of the dynamic aspectsof the industry, for example, drugs that are signif-icant innovations, average or commercially unsuc-cessful drugs, rivals' competitive entry withcompeting drugs, and next generation drugs. Theeconomic value of the simulated R&D firm isknown, and the imposition of different accountingrules gives insights into the relevance and reliabil-ity of R&D.

Using the simulation data, Healy et al. (2002) in-vestigate the relevance and reliability of R&Dunder three methods of accounting: (1) immediateexpensing; (2) full cost capitalising of all R&D ex-cept basic research and expensing over the life ofthe drug once in commercial production; and (3)successful efforts made in capitalising successfuldrugs and writing down those found to be unsuc-cessful. They find the successful efforts method ofaccounting for R&D is most value-relevant.However, their tests suggest there are large meas-urement errors in the financial accounting dataunder all the accounting methods examined.Hence, their study suggests the R&D is relevant butnot that reliable due to economic uncertainty aboutthe success rate of the individual company's R&D.

In summary, researcher estimated R&D assetsare reliable enough to be value-relevant. However,R&D assets are also associated with future stock


returns, suggesting investors do not find expectedbenefits from R&D reliable enough to fully im-pound in this year's (contemporaneous) stockprice. Healy et al.'s (2002) evidence is consistentwith R&D assets being relevant but not a reliableindicator of future rents. The Healy et al. paper isone of the very few able to provide direct evidenceon reliability. An important area for future researchis one that devises research designs capable ofboth distinguishing between relevance and relia-bility and provides direct tests of reliability.

3.J 3. R&D and IP - annual R&D expendituresAnnual expenditures on R&D related IP are not

routinely reported. However, R&D expendituresare available from surveys and as a result of ac-counting standards in some countries, for example,the US standard SFAS No. 2 Accounting forResearch and Development Costs and the R&Ddata from the Industry R&D Survey conducted bythe US Census Bureau and National ScienceFoundation. Using these data sources, a large num-ber of studies find a positive significant relation-ship between R&D expenditures and investors'valuation of the firm as reflected in stock price.'^However, while R&D is generally value-relevant,taken as a package, the evidence below suggestsR&D expenditures are not that reliable as an indi-cator of the timing and magnitude of future bene-fits. In particular, investors do not appear to findit easy to evaluate the future earnings implicationsof the R&D expenditures, consistent with the un-certainty properties of intangibles outlined inSection 2.2.

One reason R&D expenditures are not reliableindicators of future rents is that these outlays donot directly produce a stream of revenues from thesale of products. Conceptually, earnings fromR&D are more variable because R&D involvessearch and discovery and problem-solving activi-ties whose success is uncertain (Dosi, 1988).Using the variability of future earnings (varianceof realised annual earnings over five years) as thedependent variable, Kothari et al. (1999) provideevidence that the benefits from R&D expendituresare more variable, and hence less reliable, than thebenefits from capital expenditures for a sample ofover 50,000 firm-year observations for1972-1992. Kothari et al. find a coefficient on cur-rent R&D expenditures about three times the coef-ficient on current capital expenditures (controllingfor leverage and firm size). Amir et al. (2002) findthat this greater future earnings variability effect islargely confined to firms in more R&D intensiveindustries and not to other industries.

Some doubt exists over the completeness of themeasures of R&D expenditures. Hansen and Serin( 1997) show that R&D expenditures are a hiddencost in some low-technology industries, for exam-

227

pie, process innovation costs that are not separate-ly reported but are bundled with production costsin manufacturing. Further, companies do not dis-close what types of expenditures are actually in-cluded in their R&D expenses. Some expenditureincluded in R&D has more direct implications forfuture earnings and earnings variability comparedwith other expenditures.

Overall, this evidence on the greater variabilityof earnings from R&D is consistent with the dif-ferent purposes of the R&D and capital invest-ments: R&D produces innovations while capitalexpenditures produce products that embody the in-novations. This relation has been demonstratedempirically using Granger causation tests (Lachand Schankerman, 1989; Lach and Rob, 1996).That is, R&D comes first and fixed capital invest-ment comes later, once fixed assets are needed toproduce the goods ready to sell to customers.Further, current R&D includes soon-to-be success-ful as well as some (potentially a lot of) unsuc-cessful expenditures, which suggests fairly clearlythat evaluating R&D expenditures alone, in the ab-sence of information about the probability of suc-cess, will not provide a lot of insights on value.

Sougiannis (1994) provides indirect evidence onthe reliability of R&D expenditures for a sample of573 US firms engaged in R&D between 1975 and1985. Sougiannis estimates two equations captur-ing (1) the R&D association with earnings, and (2)the R&D association with market value.'^Sougiannis finds one dollar of R&D is associatedwith a two-dollar gross profit increase over aseven-year period and a five-dollar increase inmarket value. However, he finds only the currentR&D expenditures are positively and significantlyassociated with the firm's market value of equity.By contrast, past R&D is unrelated to the marketvalue of equity. This result suggests no morebenefits are expected from the past R&D,

" See Ben Zion (1978); Ben Zion (1984); Griliches (1981);Hirschey (1982); Connolly, Hirsch and Hirschey (1986); Jaffe(1986); Ettredge and Bublitz (1988); Bublitz and Ettredge(1989); Chan, Martin and Kensinger (1990); Connolly andHirschey (1990); Griliches, Hall and Pakes (1991); Shevlin(1991); Hall (1993a); (1993b); Johnson and Pazderka (1993);Megna and Klock (1993); Chauvin and Hirschey (1994);Sougiannis (1994); Lev and Sougiannis (1996); Deng and Lev(1998); Stoneman and Toi vanen (1997); Aboody and Lev(1998).

' Earnings and price are endogenous in this specification;(I) (Earnings after tax & before extraordinaries, advertisingand R&D expense) = f (net capital stock measured as inflationadjusted items; PPE+inventories+intangibles+other invest-ments, advertising; and current and lagged R&D expendi-tures); (2) Price/BV equity = f (book value equity, abnormalearnings adjusted for R&D expensing and tax, R&D taxshield, current R&D costs, lagged R&D outlays). From thisanalysis he derives (a) the total effect of past and current R&Don earnings, (b) the indirect effect of R&D on stock pricethrough earnings, and (c) the direct effect of R&D on stockprice.


meaning none of the past R&D are assets and allthe benefits have already been received, whichseems implausible. A more plausible alternativeexplanation is that investors are uncertain aboutthe probability of future benefits from the pastR&D. Overall, Sougiannis' evidence suggestsR&D is value-relevant, but the time series of R&Dis not a reliable predictor of future rents.

Green et al. (1996) cast doubt on the value-rele-vance of R&D expenditures in the UK setting for1990, 1991 and 1992. They find the R&D expen-ditures are significant in 1991 but are not reliablyvalue-relevant in 1990 and 1992. However, theirdependent variable is the difference between themarket value and book value of shareholders' eq-uity (MVE-BVE). This variable provides a test ofthe relations between R&D and the excess marketvalue over the book value of assets, not the level ofthe firm's market value of equity. This excess maynot fully capture investors' expectations of thelead/lag relations between R&D and the expectedbenefits. Further, this excess variable is regressedon earnings, current R&D expenses plus a numberof control variables, some of which are likely to becorrelated with (proxy for) the level of R&D ex-penses or the risk of the R&D, leading to a value-irrelevance result.''' More recent UK evidence for1990-1994 (Stark and Thomas, 1998) and1990-2001 (Akbar and Stark, 2003) finds thatR&D is value-relevant in the UK setting.

Another explanation for the Green et al. (1996)results is a lack of power due to under (or no) re-porting of R&D expenditures by the UK firms.Stoneman and Toivanen (1997) encounter this prob-lem in their UK study. They employ a research de-sign that allows for sample selection bias due to thenon-reporting of R&D. For 1989-1995, they findR&D is value-relevant for UK firms. The valuationmultiple ranges between zero and 4.3, and they findthe multiple varies over time and across the firms.

One further issue in the UK setting, for R&Dvalue-relevance studies prior to 1996, is the UKGAAP impact on reported intangibles. Companiespredominantly wrote off goodwill to an equity ac-count in this time period but were able to reportidentifiable intangibles such as brands. If thegoodwill write-offs significantly understate intan-gible assets, then it is possible that regression co-efficients on assets in the model could be biaseddue to omitted correlated variables. Shah, Starkand Akbar (2007) consider the issue of omitted ad-vertising costs in the UK studies of the value-rele-vance of R&D. Advertising cost data had not beenavailable in the UK until the advent of theACNielsen MEAL data. This organisation moni-tors media outlets and assigns standard costs toadvertising activities. Using this data to controlfor advertising costs. Shah et al. (2007) find R&Dis (positively) value-relevant in the UK for all

firm size groups and for manufacturing and non-manufacturing sectors, as well as the R&D intensepharmaceutical, biotechnology, electronics, andelectrical equipment sectors.

Several studies examine factors that impact thevalue-relevance of R&D expenditures. One impor-tant factor is the project stage. For example,Shortridge (2004) provides evidence that the trackrecord of new drug approvals conditions the rela-tion between R&D and stock price for a sample ofUS pharmaceutical companies. Hand (2001) findsinvestors expect more successful outcomes frommore intense R&D and find more information inR&D for growth firms rather than establishedbiotechnology firms. Investors therefore appear tounderstand that the information in R&D about fu-ture benefits varies according to the firm's life-cycle and project stage.'^

A further issue in interpreting the R&D studies isthe model specification. In contrast to other ac-counting studies. Hand (2001) uses a Cobb-Douglasfunction (commonly used in economic studies)which allows for the diminishing marginal returnfrom R&D. More generally. Hall and Kim (1999)report that non-linear and log-linear functionalforms best approximate the R&D relation withstock price. A survey of the economic literature sug-gests the likely reason for non-linearity is techno-logical and firm life-cycles, and the changes inreturns to investment at inflection points in the cy-cles (Geroski, 2000). This non-linearity is corrobo-rated by accounting studies that suggest the relationbetween earnings, as a summary measure includingR&D and other expenditures, and stock returns isnon-linear (Cheng et al., 1992; Freeman and Tse,1992; Das and Lev, 1994; Subramanyam, 1996).

In summary, taken together, the evidence in thissection from a variety of different research designs,including the earnings variability tests, suggestsR&D expenditures reflect information that isvalue-relevant. However, the information is not asreliable in reflecting future benefits as the informa-tion conveyed by expenditures on tangible assets.Several plausible explanations are canvassed. Thedominant factor is the role of R&D which relates tofuture rents rather than current production rev-enues. Current R&D also includes both value-rele-vant successful and value-irrelevant unsuccessful

" Green et al. (1996) control variables comprise marketshare, concentration, debt-to-equity, average industry debt-to-equity, square of the difference between debt-to-equity of firmand industry, and annual variability of stock market returns.

"* Liu (2007) finds the firm's life-cycle is an omitted vari-able in commonly used discretionary accrual models, with theinferences from earnings management studies changing oncelife-cycle is included in the models. Anthony and Ramesh(1992) show that the stock price response to accounting per-formance measures, sales growth and capital investment is afunction of firm life-cycle stage.


expenditures. While the projects are incompleteand success rates are not known, the R&D is not re-liable as an indicator of value. It is possible thatR&D is also less relevant in some circumstances(e.g. less relevant for established than for growthfirms). Hence, evaluating R&D expenditures withprobable success rate indicators is important to getinsights on relevance, reliability and value. Someinformation that is relevant to this task includesnew drug approvals for pharmaceuticals, more in-tensive R&D, and growth firms' R&D.

Finally, it is not clear from the evidence to datethat reporting the aggregate, R&D expenditures, asa separate line item in the income statement is allthat informative about value creation. Outsiders donot know what is included in R&D and whetherthe inclusions relate directly to future rents. It ispossible to identify more informative categories ofexpenditures for separate line item reporting andas inputs to valuation (Hunter, Webster and Wyatt,2007).

3.1.4. R&D and IP - input metricsThe firm's production function is conditioned on

the state of science and technological knowledge(Jorgenson, 1989). To test the impact of technolo-gy as a conditioning factor on the value-relevanceof R&D and IP, Greenhalgh and Rogers (2006)condition their analysis on an augmented versionof Pavitt's (1984) technology sectors, comprisingthe following:

1. Supplier-dominated manufacturing and mining:usually smaller firms with weak in-house R&Dand engineering capabilities and innovationscoming from equipment and materials suppliers;

2. Production- and scale-intensive: large firmsproducing standard materials or durable goods;

3. Production-intensive, specialised suppliers: ma-chinery and instruments, tending to be smallertechnologically specialised firms;

4. Science-based: electronics, electrical and chem-icals, usually large firms with in-house R&D-based technology but the basic science isproduced elsewhere;

5. Information-intensive: includes finance, retail,communications, publishing, with in-housesoftware or systems development, with pur-chases of IT hardware and software;

6. Software-related firms: computer software andservices.

Using the log of the market value of equity as thedependent variable, Greenhalgh and Rogers findthe magnitude of the coefficient on R&D variessubstantially across technology sectors. It is lowestfor '6. Software' followed by '4. Science' (whichspends 45-55% of the total R&D across all six sec-

tors). The highest coefficient is for '2. Production-and scale-intensive' and '5. Information-intensive'.The 'Science' sector result is probably due to theaverage 20-year lag from an idea to a successfulscience innovation (Stephan, 1996). Including(total asset deflated) patent and trademark data inthe regressions does not affect these coefficients.They find the UK patents are less value-relevantthan the European Patent Office patents.Trademarks generally have positive and signifi-cant coefficients in each sector. They find the mostcompetitive technological sectors have the lowestmarket value of R&D. Within the most competi-tive technology sector (science-based manufactur-ing), firms with larger market shares (proxying forlower competition) have higher R&D valuations.

Matolcsy and Wyatt (2008) use patent metrics ag-gregated to the technology sector level (within in-dustries) as technology input metrics, to testwhether the association between the market value ofequity and current earnings is conditional on thetechnology conditions. Three technology conditions(within the industry dominating the firms' opera-tions) are considered: the success rate of past tech-nological investments, technology complexity, andthe technology development period. Using the mar-ket value of equity deflated by sales, they find thetechnology condition-earnings interactions arevalue-relevant. The results hold across a range ofhigh-, medium- and low-technology industries con-sistent with the predicted pervasive effects of tech-nology conditions on the firms' operations. Thethree technology conditions also predict future earn-ings, which is a pre-requisite for value-relevance.

In summary, the value-relevance of R&D variessubstantially across technology sectors, suggestingtechnology conditions impact the success rate andhence expected rents from R&D. It is difficult toknow whether this is due to variation in value-rele-vance or reliability, or both of these. Measures ofthe technological innovation success rate, cycletime, and links to science in a technology sectorcondition the relations between current earnings andthe market value of equity, suggesting the pervasiveimpact of technology on performance and value.

3.1.5. R&D and IP - output metricsPatents are output metrics which are used as

measures of invention and/or protection arisingfrom IP laws (Lanjouw et al., 1998). The distribu-tion of patents is substantially skewed to low valuepatents." Researchers also find the rate of decline

" Patent protection is reduced by the capacity of imitatorsto 'invent around' a patent, by the difficulties actually secur-ing patents on some innovations, and by the problem thatpatents can disclose information sufficient to facilitate imita-tors' development of variants of the basic technology; andthese problems are typically viewed as greater for process in-novations than for product innovations (see Geroski. 1995).


in the value of patents is also much higher than therate for most physical assets (Cockburn andGriliches, 1988; Schankerman and Pakes, 1986;Pakes, 1986; Schankerman, 1991). Therefore,variation in the use and value of IP renders simplecounts of patents not very useful for valuation pur-poses (Griliches, 1990).

Early studies found patents are value-relevantincrementally to a measure of intangible capitalcomputed from past R&D expenditures (e.g.Griliches, 1981; Pakes, 1985; Cockburn andGriliches, 1988; Megna and Klock, 1993).Greenhalgh and Rogers (2006) find patents are in-crementally value-relevant to R&D for UK firmsfor 1989-2002.

Following the early work using count measures,researchers found patent renewal and patent fami-ly size are useful for computing quality weights forpatent count data (Lanjouw et al., 1998). Ratherthan simply count patents, the patents are parti-tioned into groups according to the age the patentis allowed to lapse, or by the set of countries inwhich patent applications were filed. Serrano(2006) also examines the decision to sell patents asa measure of value. Bessen (2007) develops andtests a more complex model that gives an upperbound estimate of the value-relevance of patents tothe firm. His results are qualitatively similar to thevaluation results using the renewal and sellingmeasures. Bessen (2007), among others, finds thatchemical and pharmaceutical patents are morevaluable than patents in other industries.

Citation analysis is a measurement approachwhich provides a quality weighting to augmentsimple count measures of IP (Narin, 2000). Highcitations to scientific research papers, and fromcurrent to prior issued patents, indicate importantscientific and technological inventions. For exam-ple, a US patent has eight or nine 'ReferencesCited - US patents' on its front page, two refer-ences cited to foreign patents, and one to two non-patent references. These references link the patentto the related prior art (related patented invention)and also limit the claims of the current issuedpatent. Like the patent distribution, the citationdistribution is skewed. For example, Narin (2000)reports that for patents issued in 1988, and cited inthe next seven years, half the patents are cited twoor fewer times, 75% are cited five or fewer times,and only 1 % of the patents are cited 24 or moretimes.

Studies linking citations to the market value ofequity find citations to prior patents, and to scien-tific papers, are value-relevant. For example. Hall,Jaffe and Trajtenberg (2005) find patent citationsare incrementally value-relevant over R&D to as-sets, patents to R&D, and assets, for a sample ofcompanies from 1963 to 1995. They find one ad-ditional citation per patent is associated with a 3%

higher stock price, unanticipated citations havea stronger effect on stock price, and citations tothe companies' own prior related patents aremore valuable than external citations. Hirschey,Richardson and Scholtz (1998) also find patentmetrics computed from patent count and citationdata are incrementally value-relevant to earnings,book value of shareholders' equity, R&D expendi-tures and pro forma R&D capital.

Another factor to consider is the value implica-tions from the strategic use of IP. Cohen, Nelsonand Walsh (2000) provide large sample survey ev-idence that firms patent for more reasons than di-rect protection of profits. Firms use patents toprevent rivals from patenting related inventions(e.g. blocking rivals' patents by chemical firms),the use of patents in negotiations (e.g. by telecom-munications companies) and the prevention of lawsuits. Other strategies used with or instead of IP toprotect profits from invention include secrecy, leadtime advantages, and complementary marketingand manufacturing capabilities. In fact, Cohen et al.find that secrecy and lead time are generally moreimportant than patents for protecting the profits ofmanufacturing firms.

Schankerman and Noel (2006) investigate theproposition that 'strategic patenting' raises thecosts of innovating for rival firms, using two out-put metrics to proxy for strategic patenting activi-ties: patent portfolio size, which they argue affectsbargaining power in patent disputes, and the^ra^-mentation of patent rights, which increases thecosts of enforcement. Consistent with these strate-gies increasing their own inventive activity andown benefits from invention, they find these met-rics are positively associated with innovation ac-tivity and with the market value of equity, for asample of software firms in the period i980-99.

Henkel and Reitzig (2007) study patent blockingwhereby firms patent solely with the intent ofblocking other companies' R&D-related innova-tions. They show that patent blocking is a viablestrategy in competitive, higher technology indus-tries, if the 'blockers' focus on inventions that theycan easily invent around and where the 'blockingpatent infringements' (that the 'blocker' issuesagainst rivals' related patents) are more readily up-held in court.

By contrast, McGahan and Silverman (2006)find, in circumstances where a technologicalbreakthrough creates investment opportunities forall firms in the industry, the positive effects of ad-ditional knowledge and opportunities outweighsthe negative impact of the patent blocking strategy.Hence, the significance of invention appears to in-teract with the strategic use of IP to influence per-formance and firm value.

In summary, patent metrics that are qualityweighted, such as using citations to prior patents.


are reliable enough to be value-relevant and aremore value-relevant than simple patent counts. Itis unclear whether this effect relates to more rele-vance or reliability or more of both. Output meas-ures of strategic patenting are value-relevant.

3.2. Advertising expenditures, brands, and IPBrands and trademark (IP) assets are output from

prior investments in advertising and expendituresassociated with product development and trade-mark registration. They generate value throughmarket power and signalling of product and, possi-bly, seller attributes. These assets can be exchangedand operated independently of specific human cap-ital. For example, newspaper businesses with mast-heads can be sold and operated independently ofthe parties who developed the mastheads.

3.2.1. Advertising expenditures, brands, and IP -management reported assets

Prior to the FRS 10 issue in 1998, when brandrecognition was allowed in the UK, Müller (1999)finds the UK firms capitalised their brands to meetfinancial ratio-based rules set by the London StockExchange (LSE). These LSE rules waive share-holder approval for acquisitions below certain fi-nancial thresholds. For a sample of 33 UKcompanies for 1988-1996, Müller (1999) reportsthe firms had previously written off their pur-chased goodwill, and now with a weak balancesheet, put capitalised brands onto their balancesheets to avoid the costly LSE rules. Hence, themotivation for capitalising the brands is not tocommunicate with investors, at least, as a firstorder effect.

Despite this motivation. Kallapur and Kwan(2004) find the goodwill and identifiable intangi-bles of UK companies, including brands andpublishing titles, are reliable enough to be value-relevant. Their firms' median brand assets are alarge 44% of the book value of shareholders' equi-ty. However, Kallapur and Kwan provide furtherevidence suggesting that the value-relevance ofthe brands was adversely affected by incentives to(1) avoid LSE rules requiring shareholder ap-proval for large acquisition or disposal transac-tions; and (2) reduce leverage. They conclude thebrands are value-relevant but their reliabilityvaries with managements' financial reporting mo-tivations.

In summary, management reported brands priorto 1998 in the UK were reliable enough to bevalue-relevant. However, the brands were lessvalue-relevant and/or less reliable for firms capi-

™ See, for example, Hirschey (1982), Hirschey andWeygandt (1985), and Chauvin and Hirschey (1993).

2' See Peles (1970), Abdel-khalik (1975), Ettredge andBublitz (1988), Bublitz and Ettredge (1989), and Hirschey andWeygandt (1985).

231

talising for agency reasons. It is not clear which ofthese is descriptive.

3.2.2. Advertising expenditures, brands, and IP —researcher estimated assets

Seethamraju (2000) constructs measures ofinternally generated US brand names, from theintensity of advertising expense, and finds theseestimates are value-relevant. Hence, this evidencesuggests brands valued by external parties usingpublicly available expenditures data are reliableenough to be value-relevant.

3.2.3. Advertising expenditures, brands, and IP -annual advertising expenditures

There is a positive relationship between adver-tising expenditures and stock price.^" However,some studies find a significant association only forexpenditures on non-durable goods . ' In tests re-lating advertising to sales, researchers find adver-tising is associated with current rather than futuresales, which suggests the benefits are short-lived(e.g. Boyer, 1974; Clarke, 1976; Grabowski, 1976;Lambin, 1976). Netter (1982) examines whetherfirms spend too much on advertising, leading to aweak relation with stock price. He finds the adver-tising of competitors reduces the effectivenessof non-durable producers' advertising outlays.Hence, advertising generates value conditional onproduct type and competition conditions.

The economics of advertising suggests advertis-ing is linked to value creation through the process-es of new product development and adoption. Assummarised by Nakamura (2005), advertisinghelps consumers to learn more quickly about theexistence and properties of new products, therebyfacilitating the flow of benefits and financial re-wards from innovation to the producers and con-sumers. Since new products have increased ineconomic importance, this suggests the impor-tance of advertising as a continuing long-run in-vestment.

Advertising has another potentially importantbenefit. These costs are often packaged as a jointproduct with entertainment, such as free-to-airradio and television. Borden (1942) shows thatabout half of the total advertising costs in 1937went to fund entertainment including live artistsand leases of phonographs. In relation to the sidebenefits of advertising, Noll et al. (1973) estimatethe value of the rise of television to consumers atabout 5.1% of household income in 1969.

What determines the longevity of advertisingbenefits? Nakamura (2005) suggests longevity isa function of product innovation and adoption.Bemdt et al. (1994) provide insights on this issue.They decompose advertising expenditures toseparate out industry-expanding investment fromrivalry inducing expenditures. They focus on the


market for ulcer drugs (H2-antagonist drugs:Tagamet, Zantac, Pepcid and Axid). For the rival-ry inducing component, they find advertising costsand industry sales are negatively related in cir-cumstances when the number of products on offerin the industry is increasing. Advertising costsunder these conditions depreciate at a fast annualrate of 40%. By contrast, the industry-expandingcomponent appears to have an almost zero rate ofdepreciation.

In summary, advertising expenditures are value-relevant in the short term but the evidence is mixedfor the long term. It is possible the mixed evidenceis due to a lack of value-relevance for companiesspending too much on advertising. Alternatively,the mixed evidence could be due to uncertaintyabout future benefits and hence lack of reliability.It is not clear from the evidence, which is descrip-tive. Another possibility is gaps in the modellingof the costs and benefits of advertising. In particu-lar, the theory suggests insights on value creationmay come from modelling the context of the linksbetween advertising expenditures and marketvalue of equity, focusing on the effects of productinnovation and adoption and joint product costsand benefits. These gaps in the literature call forstudies of the long-term effects of advertising ex-penditures, which have been less common to datedue to problems obtaining data, along with gapsin our understanding of consumer behaviour(Vakratas and Ambler, 1999).

3.2.4. Advertising expenditures, brands, and IP -input metrics

Franses and Vriens (2004) point out the amountof money companies allocate to advertising oftensurpasses the companies' after tax profits, but stillit is not known whether these investments pay offor not. They argue the reason for this gap is in-complete knowledge about what advertising doesin the marketplace and list four factors that are im-portant for understanding the returns to advertisinginputs:

1. The process by which advertising affects con-sumers and leads to brand awareness, brandimage, brand consideration, brand choice, andsales;

2. How the effects of advertising are spread outover time;

3. The role of different advertising media (for ex-ample, TV versus print advertising), how differ-entially efficient these vehicles are, how theirinteraction may lead to synergy effects; and

4. The role and impact of competitive advertising.

These factors suggest the range of input metricsthat are relevant to estimates of value creation andlongevity of benefits from advertising, for exam-

ple, consumer purchasing, repeat business andswitching metrics, firms' advertising strategies, in-cluding the frequency and magnitude of efforts,media use metrics, experience and search attrib-utes of products, and the interaction effects withthe purpose of advertising (e.g. market sharegrowth, and price premium support).

One area of research having some success inmodelling strategy impact on value are studies ofoptimal scheduling of advertising over time. Oneexample by Dube, Hitsch and Manchanda (2004)is a study of a pulsing strategy in which the firmadvertises in sharp, intensive bursts. They developa dynamic programming framework and find thatpulsing is the optimal strategy for the industry sec-tor they study.

Aside from the modelling method employed, thesuccess of the Dube et al. model is due to high-quality input metrics, which contrasts with otherstudies in this area that use the available, patchyadvertising expenditures. The data is Scantraclevel scanner data for frozen entrée foodstuff,comprising weekly sales, prices and advertisinglevels for each brand in 18 cities over 155 weeks.Advertising level is measured using gross ratingpoints rather than dollars, which captures house-hold average exposures to advertising in variousmarkets per week. The long weekly series and thebenchmark rating data allows pulse behaviour tobe tested (i.e. this is not easy to do with annual orquarterly data and raw dollars). Interestingly, theirresults suggest continuous advertising is sub-opti-mal.

In summary, quality input data relating to con-sumer behaviour and firms' advertising strategiesover time are important for measuring and under-standing how advertising creates value for thefirm. The evidence suggests that it is important tostudy advertising expenditures in the context ofstrategy and consumer behaviour to get insights onvalue creation. An example of such an insight re-lates to advertising strategy for which there is evi-dence that continuous advertising rather than apulsing advertising strategy can destroy value.

3.2.5. Advertising expenditures, brands, and IP -output metrics

Trademarks are intermediate output measuresthat are potentially valuable when firms use themto signal desirable product attributes to consumers,thereby reducing information asymmetries be-tween sellers and buyers (Landes and Posner,1987). Trademarks may create value by motivatingthe firm to invest in quality products (Mendonca,Pereira and Godinho, 2004) and engage in innova-tion activity and the building of brand value andbarriers to entry (Schmalensee, 1978).

Consistent with trademarks generating value,Greenhalgh and Rogers (2006) find trademarks are


incrementally value-relevant to R&D and patentsfor UK firms for 1989-2002. They find the trade-marking firms experience 10-30% higher produc-tivity compared with non-trademarking firms.Further, trademarking activity, and the intensity oftrademarking, is associated with larger differencesin the market value of equity and productivityamong firms in the services industries compared tomanufacturing firms.

Barth et al. (1998) test the value-relevance ofoutput measures of brand values from FinancialWorld's (FW) annual brand value surveys.^^ For asample of 595 US firm-years with brands valuedbetween 1991 and 1996, Barth et al. (1998) ñndthe FW brand values (changes in brand values) aresignificantly positively associated with the marketvalue of equity (stock returns). The FW brand val-ues are incrementally value-relevant (for marketvalue of equity) to advertising expense, operatingmargin, growth, market share, recognised brandassets, and analysts' earnings forecasts. The FWsample is not random. It is dominated by large,profitable companies in food and tobacco, chemi-cals and allied products, rubber, plastic, leather,and glass industries, and under-represented by fi-nancial services. Whether brands in general can bevalued as successfully is unclear.

In summary, trademarks are value-relevant andappear to be particularly significant value driversfor firms in services industries with a significantbut lower impact in manufacturing. The evidencedoes not distinguish whether this effect relates tolower relevance or reliability, or both. A possibleimpact of trademarking activity is to motivate thefirm to engage in further value-creating productand brand innovations. While this might be moti-vated by the desire to build the trademark value,there is the potential for feedback effects for thefirm's R&D outlays, which may build additionalvalue in the future via the product pipeline andbrands. While brands do not have ÍP rights at-tached, they are value-relevant for large profitablecompanies, suggesting they are significant indica-tors of market power. It is not known whether theindependent valuations of brands used by Barth etal. (1998) could be successfully undertaken forfirms in general at the same level of reliability.

3.3. Customer loyaltyA number of studies examine whether customer

satisfaction measures relating to the firm's productmarkets are value-relevant. The impetus for thisresearch is marketing studies that propose cus-tomer satisfaction is a key value driver because itreflects information about customer retention,price elasticity reduction, brand and reputation ef-fects (Anderson et al., 1994). Further, there is evi-dence that companies value and track this data(Ross and Georgoff, 1991). There is some overlap

between the value represented by brands and cus-tomer loyalty assets, although no known studiesexamine how this might relate to the propensity tocreate value (e.g. does the interaction betweenthese two constructs generate synergies for thefirm?).

3.3.1. Customer loyalty - management reportedassets

GAAP standards do not provide for the reportingof customer loyalty assets.

3.3.2. Customer loyalty — researcher estimatedasset

A lot of the evidence on the value-relevance ofcustomer loyalty comes from survey-based outputmetrics produced by researchers, research insti-tutes, or the firms themselves (e.g. Ittner andLarcker, 1998). The evidence from the survey met-rics is mixed in part because the survey data is col-lected by different companies and organisationsusing different instruments, subjects, and time pe-riods, which makes the results of the studies hardto interpret (Boyd et al., 2004).

By contrast, Gupta et al. (2004) provide evi-dence on the contribution of customers to valueusing primarily publicly available information.They provide novel evidence that the long-termvalue of the firm's customer base is a good proxyfor the market value of equity. They first develop amodel to value the firm's long-term customer baseusing the following information: a forecast of cus-tomers to be acquired in the future, average cus-tomer acquisition costs, profit margin percustomer, and customer retention rate. These cus-tomer value drivers comprise both financial andnon-financial information, some of which are in-puts to value (e.g. acquisition costs) and some areoutputs to value (e.g. customer retention).Estimates from the customer valuation model areclose to the market value of equity for three(Capital One, Ameritrade, E*Trade) of the fivefirms considered (the measure undervaluesAmazon and eBay, which many analysts believedwere over-valued at the time).

Consistent with other studies, Gupta et al. (2004)find customer retention has the biggest impact oncustomer value, in the order of 3 to 7% for a 1%increase in retention. By contrast, profit margin

• FW reports value estimates, sales, and operating marginsfor individual brands, by industry, as well as the percentagechange in the brand value from the previous year. Brandstrength multiples for each brand, obtained from Interbrand,are a weighted metric computed from seven components:(1) Leadership (maximum 25 points); (2) Stability (maximum15 points); (3) Market (maximum 10 points); (4) Intemationality(maximum 25 points); (5) Trend (maximum 10 points);(6) Support (maximum 10 points); and (7) Protection (maxi-mum 5 points).


per customer has a 1 % impact on customer valuewhile the cost of acquiring customers has a 0.02 to0.3% effect on value. Further, they find a stronginteraction effect between the cost of capital andretention rate. Specifically, the value of customersin the high retention—low cost of capital context is2.5 to 3 times the value in the low retention-highdiscount rate setting. This finding suggests highcost of capital companies would benefit more fromcustomer retention rather than containment of cus-tomer acquisition costs.

In summary, issues relating to the reliability ofcustomer loyalty measures are a possible explana-tion for the mixed value-relevance evidence formeasures obtained from survey data. Estimates ofthe long-term value of the firm's customer base,and the specific drivers of this value (e.g. customerretention), suggest this is an important policy areafor firms which warrants further research.

3.3.3. Customer loyalty - annual expendituresAnnual expenditures on customer loyalty are not

generally reported as separate line items in GAAPfinancial statements.

3.3.4. Customer loyalty — input metricsOne input that is a driver of customer loyalty is

the quality of customer service. Decreasing levelsof customer service have been cited as a cause ofcompetitive decline (Roach, 1991). To provide ev-idence on the information content of the firms' ac-tions to change their quality of customer service,Nayyar (1995) employs an event study. This de-sign has the capacity to indicate whether investorsactually used the information, subject to the re-searcher adequately controlling for competing in-formation. Nayyar argues that if improvingcustomer service leads to improved performance,then actions that improve customer service shouldbe valued positively by the stock market when theactions are announced. Using news reports frombusiness news databases, he identifies actions byfirms relating to changes in customer service over1981-1991. Action (inputs) relate to four customerservice objectives: (I) risk of purchase; (2) pur-chasing cost; (3) ease, convenience, cost of use;and (4) personalisation. Nayyar finds that increas-es (decreases) in customer service are positively(negatively) valued by the stock market, as reflect-ed in cumulative abnormal returns in the event

^ As compared with less valuable actions to change theease, convenience, and cost of use (customer service depart-ment, technical assistance, toll-free numbers, discretionarycomfort features) or the personalisation of products (comput-er to customise products, capacity to meet unique needs).Specific customer service actions viewed most favourably bythe stock market are improved guarantees, increased operatinghours, greater customer service outlets, and better computerlinks to buyers.

window. The strongest effects are for reducing therisk of purchase (appearance of facilities and guar-antees) and purchasing cost (customer service out-lets, credit terms, computer links to buyers, andoperating hours).^^

Like customer satisfaction, consumer switchingcosts (brand loyalty) give the firm market powerover repeat purchasers. These costs make thefirm's current market share an important determi-nant of future profits and value (Klemperer, 2005).Examples of factors that give rise to switchingcosts include frequent flyer programs, computercomponent compatibility, and the cost to learn touse another brand.

The effects of consumer switching cost on firmvalue are ambiguous. Benefits from high cus-tomer retention may be off-set if switching costsincrease product prices over time to the pointwhere it is viable for consumers to switch brands.Switching costs can discourage new entry to theindustry and generate inertia in product andprocess innovation by reducing the firm's incen-tives to differentiate their products, thereby reduc-ing competition. The extent of the value createdfor the firm from consumer switching costs there-fore depends on a careful analysis of the compet-ing effects.

In summary, there is evidence that an input tocustomer loyalty, changes in customer servicequality, is associated with cumulative changes inunexpected stock returns. Firms have incentives tocreate switching costs for consumers to build mar-ket share. The challenge for researchers is to de-fine and collect or construct concrete inputmeasures for tracking and studying customer loy-alty components.

3.3.5. Customer loyalty - output metricsIttner and Larcker (1998) find that a customer

satisfaction output metric for a large telecommuni-cations firm (survey of 2,491 from a total of450,000 customers) is significantly positively andnon-linearly associated with customer retention,revenue and revenue-change measures. Tests forbusiness unit customer satisfaction metrics from73 retail banks (different metric to above) suggestan indirect effect on accounting performance byattracting new customers, consistent with thebank's strategic goal. However, while Ittner andLarcker (1998) find another output metric from theAmerican Customer Satisfaction Index (ACSI)survey is associated with the market value of equi-ty, the relation is not consistent (significant intransport, utility, communication; insignificant indurable and non-durable manufacturing and finan-cial services; significant and negative in retailing).The ACSI metric gives rise to a positive an-nouncement effect (on stock price) suggesting theindex conveys information to investors. The index


is also positively significantly associated withforecasted residual earnings, suggesting some ofthe value is impounded in earnings.

The ACSI is one of the more standardised sur-veys generating output metrics. It is a national eco-nomic indicator managed by the National QualityResearch Center and the American Society ofQuality. The 15 questions in the survey are organ-ised by four variables: perceived quality, customerexpectations, perceived value and customer satis-faction. The customer satisfaction measure is com-puted from three of the questions relating tooverall satisfaction, confirmation of expectationsand comparison with the ideal.

Jacobson and Mizik (2007) find the ACSI meas-ure is only value-relevant for computer and inter-net firms. Jacobson and Mizik (2007) use anannual stock returns specification. The stock re-turns test looks at relevance in terms of whetherthis year's change in customer satisfaction is asso-ciated with this year's changes in stock price,whereas the stock price levels test does not refer toa narrow time frame. These stock return tests sug-gest the customer satisfaction measure may notchange enough to be value-relevant on an annualchanges basis but is value-relevant in a wider timeframe. Alternatively, the stock returns test poten-tially lacks the power to detect value-relevanceand/or the ACSI metric is noisy.

Consistent with a forward-looking value cre-ation effect. Banker et al. (2000) find the non-fi-nancial measures of customer satisfaction,likelihood a customer will return to the hotel andcustomer complaints, are significantly positivelyand negatively, respectively, associated with futurefinancial performance (business unit revenues andoperating profit) for a hotel chain. This data is col-lected by the hotels. Their evidence suggests thatcustomer satisfaction is related more to long-termrather than short-term performance. Further, theyfind that changes in management incentive con-tracts to include these non-financial performance(output) metrics leads to improvements in both fi-nancial and non-financial performance.

In summary, precise component level measuresof customer loyalty such as in Banker at al. (2000)would help us to understand what drives customersatisfaction in different industries and how tomeasure the composite output indexes more pre-cisely.

3.4. Competitive advantageTwo literatures relevant to understanding the

sources of competitive advantage include the in-dustrial organisation model of competitive advan-tage focusing on industry (e.g. Porter, 1980; 1985)and the resource-based model of competitive ad-vantage focusing on firm-specific factors (e.g.Wernerfelt, 1984; Barney, 1991; 2001). Studies

235

from these literatures suggest that both firm-spe-cific and industry effects have a role in explainingvalue creation. However, firm-specific effectsdominate industry-specific effects (Rumelt, 1991).According to Barney (1991) value is created fromfirm-specific endowments when they are valuable,rare, inimitable and difficult to substitute.

Early research focused primarily on financialinformation to study competitive advantage.However, some researchers believe that the strate-gic activities that generate (or destroy) value arenot formally captured in the traditional categoriesof financial information. This has motivated re-searchers to expand their efforts beyond a finan-cial focus to study strategic sources of competitiveadvantage. The evidence reviewed in this sectionsuggests that investors refer to a range of GAAPand non-GAAP information relating to the firm'scapabilities and strategy to value the firm's stock.This literature identifies a range of competenciesthat are possible omitted variables from value-relevance studies.

3.4.1. Competitive advantage — managementreported assets

Consistent with a greater emphasis on non-financial measures of the firm's intangible valuesince the mid-1990s, some managers have em-braced a broader strategic focus on how their firmcreates value. This involves identifying, measur-ing and managing the value drivers of customervalue, organisational innovation, and shareholderwealth (Ittner and Larcker, 2001). Measurementtechniques range from unstructured checklists ofdiverse financial and non-financial measures tostructured methods such as the balanced score-card, economic value measurement (EVA, resid-ual income or abnormal income), causal businessmodelling, and environmental uncertainty mod-els.

Proponents of the unstructured checklist ofmeasures argue that using diverse sets of financialand non-financial measures decreases the risk thatmanagers fail to consider relevant dimensions oftheir firm's performance (Lingle and Schiemann,1996). Proponents of the structured methods arguethat these techniques are useful for identifying suc-cesses and failures relating to the firm's strategyand its fit with the organisation's objectives(Simons, 1991; Stewart, 1991; Stern et al., 1995;Gates, 1999; Kaplan and Norton, 1992; 1996;2001). Structured performance measurement ispredicated on the theory that performance meas-ures must be aligned with and contingent on thefirm's strategy and value drivers to be useful inpromoting value creation (e.g. Langfield-Smith,1997).

Campbell et al. (2002) test and find evidenceconsistent with the proposition that the balanced


scorecard technique can be used to identify prob-lems and highlight causes and solutions relating tothe firm's operating strategy. However, theiranalysis, using data from a convenience storechain, suggests that this result only holds if theanalysis is conditioned on the firm's competencies(i.e. takes into account the fit between the strategyand the skills of the workforce). Their analysis re-veals that a poor fit between strategy and employ-ee capabilities caused the convenience storechain's strategy to be ineffective.

By contrast, Biddle et al. (1997) find traditionalaccounting measures are more value-relevant thanthe structured economic value added measures. Inthe financial services industry, Ittner et al. (2003)fmd that the structured measurement approaches,the balanced scorecard, economic value measure-ment, and causal business modelling are associat-ed with higher measurement system satisfactionby users within the firm. But these three approach-es are not associated with higher accounting orstock price performance. Instead they fmd thefirms using an unstructured approach, comprisinga wide variety of financial and non-financial inputand output measures, earn higher stock returnscompared with firms with similar strategies orvalue drivers that do not use the unstructured ap-proach.

In summary, there is mixed evidence on thevalue-relevance of unstructured and structuredmethods of measuring value added. It is possiblethis is partly due to omitted conditioning variables,such as the firm's competencies (Campbell et al.,2002). Another question is how well the methodscapture the firm's value-creating processes.Aligned to this is a lack of data on the measuresthat are actually used by managers and purpose orobjective of the measures (Ittner and Larcker,2001). The lack of data is due to the proprietarynature of this type of data, along with the diversi-ty of competencies and strategies in use, whichmakes it difficult for researchers to obtain robustresults.

3.4.2. Competitive advantage — researcherestimated assets

Gjerde et al. (2007) examine the value-relevanceof three sources of competitive advantage: indus-try-based competitive advantage and two firm-specific, resources-based competitive advantagesrelating to profitability and risk. Using abnormalstock returns as the valuation variable, and show-ing consistency with other studies, Gjerde et al.(2007) find the firm-specific advantage is three tofour times more value-relevant than the industryspecific advantage. Further, they find these two ef-fects are interdependent.

3.4.3. Competitive advantage — annualexpenditures

Spending that leads to information technology(IT) capabilities has been linked to firm value al-though with mixed results. Dewan and Min (1997)provide evidence that IT capital is a net substitutefor both physical capital and labour in all sectorsof the economy using a Computerworld survey ofspending by US companies on information sys-tems from 1988 to 1992. Earlier, Brynjolfsson(1993) and Wilson (1993) could not find evidencethat IT contributed to firm productivity. Theirresults were rationalised by higher productivitylosing out to lower entry barriers, industryinefficiencies and competition (Hitt andBryjolfsson, 1996). However, later studies usingfirm-level data suggest that IT capabilities are re-lated to positive investment returns (Lichtenberg,1995; Hitt and Brynjolfsson, 1996) and to the stockprice-based measures (Bharadwaj et al., 1999;Brynjolfsson et al., 2002). Anderson et al. (2006)employ financial data from the Y2K spending inIT, in contrast to the prior studies, which primarilyuse survey data. They find opportunistic improve-ments in the firms' IT capabilities (costs were bun-dled with the Y2K spending) are associated withhigher contemporaneous stock price and higher fu-ture profits. What is unclear in this literature is theimpact of omitted correlated intangibles, i.e. manysources of intangible value go unrecognised underGAAP.

3.4.4. Competitive advantage - input metricsDarby et al. (1999) test the value-relevance of

the input measure, ties to star scientists., forbiotechnology firms. Darby et al. (1999) argue thatties to star scientists is valued by investors due tothe investors' ability to observe related indicatorsof the firm's intellectual human capital, such as thenumber of scientists, how many have PhD degrees,where they did their graduate work, and the size ofthe firm's R&D. They develop a valuation equa-tion based on an option-pricing model that embedsa dynamic jump process. This jump process in-volves changes in the firm's assets and valuewhenever the intellectual human capital, in thiscase the ties to star scientists, generate technologi-cal successes. They find that increases in the ties tostar scientists metric is associated with highermarket valuation of the firm but at a decreasingrate. For the average firm (relative to industry),there is a 7.3%, or $16m increase, in the marketvalue of a firm per scholarly article written by, orwith, a star scientist compared with a firm with noarticles in academic journals.

Amir and Lev (1996) study industry-specificmeasures of the firms' value drivers for the wire-less communications industry. In contrast to otherstudies, they find the earnings and book value of


shareholders' equity, the two summary measuresof the firm's financial performance and positionare not value-relevant. However, non-financialinput metrics are highly value-relevant for this in-dustry, including a proxy for the firm's expectedgrowth, the population of potential subscribers,and a proxy for the firm's expected operating per-formance, the penetration ratio of subscribers tothe population of potential subscribers.

Klock and Megna (2000) similarly find that theinput metrics, radio spectrum licenses and thefirm's potential customer base, are incrementallyvalue-relevant over advertising and R&D expendi-tures for firms in the wireless telecommunicationsindustry. In fact, they report that the spectrum li-cense explains over 60% of their market value ofequity measure, the Tobin's q. Tobin's q is the mar-ket valuation of the firm's financial claims dividedby an estimate of the replacement cost of the as-sets.

Ethiraj et al. (2005) study the sources of com-petitive advantage for a large Indian software com-pany with about 90% of revenues from exports.Their dataset includes information on revenues,cost, factor inputs, capability measures, variousproject characteristics, such as size, client industry,and development platform, all measured at theproject level. Ethiraj et al. test and find that twosets of firm-specific capabilities are importantsources of competitive advantage for firms in thesoftware development industry: client specifiic ca-pabilities and project management capabilities.Using non-financial and financial measures, theyfind that the firms develop these capabilitiesthrough leaming-by-doing as well as sustained in-vestment. Further, the two types of capabilitiescontribute heterogeneously to value creation. Thatis, the two capabilities are present in different pro-portions across the software firm's projects, costdifferent amounts of money and provide differentlevels of benefits. If capabilities such as these in-teract with investment to impact firm value, thenvalue-relevance tests might need to understandkey capabilities to generate valid models.

The value-relevant non-fmancial information setrelating to competitive advantage and futureprospects changes over time. Stephan et al. (2007)compare the changing value-relevant informationset for firms in the biotechnology industry overtwo financing windows, 1989-1992 and1996-2000. In the earlier period, 1989-1992,biotechnology firms going public for the first time(IPOs) comprised IP and research capabilities butfew marketable products. The likelihood of suc-cess was a function of the number of products inclinical trials, the reputation of underwriters, intel-

'' The Young and Rubicam brand asset valuator model isdiscussed in Fudge (2005).

237

lectual property, alliances, and linkages with uni-versity-affiliated scientists who won a NobelPrize.

In the later period - 1996-2000, biotechnologyindustry was more established, now comprising alarge stock of IP, as well as research alliances, andproducts in clinical testing. Stephan et al. (2007)find the most striking difference between the twotime periods relates to the value attached to thefirm's association with a Nobel laureate. Thisvalue fell from $20.4m in the 1989-1992 period,when there was little other information to signalthe firm's prospects, to zero in the later 1996-2000period.

In summary, the sources of firm-specifiic com-petitive advantage canvassed in Section 3.4.4 referto the firm's industry, technology, and businessmodel. This is consistent with the Gjerde et al.(2007) evidence that firm-specific and industrysources of competitive advantage are interdepend-ent. The factors reviewed in this section appear tobe robustly value-relevant, consistent with corevalue driver status. The discussion in this sectionhighlights the importance of identifying core valuedrivers to ensure that all important sources ofvalue are included in the tests.

3.4.5. Competitive advantage — output metricsResearchers have studied the contribution of

brands to firm value using non-financial outputmetrics from surveys of customers. While finan-cial measures of brands are usually value-relevant,as discussed earlier, the evidence from these qual-itative measures is not so convincing. For exam-ple, Mizik and Jaeobson (2006) investigatewhether five qualitative attributes of brands -energy, differentiation, relevance, esteem, andknowledge - are value-relevant. Their constructsand measures come from the Young and Rubicambrand survey.^'' Out of the five attributes, only en-ergy (future ability to generate benefits) and rele-vance (relevance of the brand to the customer) arevalue-relevant. Interpreting this study is difficult.For example, it is not clear whether Mizik andJaeobson's lack of results for three of the fivebrand attributes is due to: (1) problems with thefive attributes (e.g. are these really the key factorsand are they sufficiently precise to be useful em-pirically?); (2) problems with the customer re-sponses (e.g. customers have different perceptionsof what the 50 questions are asking them); or (3)problems with the way the five attributes are ere-ated from the 50 survey questions.

Rajgopal et al. (2003) find competitive advan-tage output metrics relating to network advantagesfrom website traffic are incrementally value-rele-vant over earnings and the book value of share-holders' equity. They provide evidence that theinputs that drive the network advantages are the

238 ACCOUNTtNG AND BUSINESS RESEARCH

firm's affiliate referral program and media visibil-ity. Rajgopal et al. (2003) also find the network ad-vantage is positively associated with financialanalysts' one- and two-year ahead earnings fore-casts, providing further support for their value-rel-evance results.

Hand (2000) employs a log-log linear regres-sion, rather than the more common ordinary leastsquares estimator, to examine the value-relevanceof web traffic metrics for internet companies in-crementally to economic variables, including thebook value of shareholders' equity, forecasted one-year ahead earnings and forecasted long-run earn-ings growth (and other supply and demandvariables). This type of regression describes a rela-tionship of diminishing returns - i.e. increases inweb traffic at low levels (high levels) of web traf-fic - are associated with large (much smaller) in-creases in stock price. Hand (2000) finds theeconomic variables dominate the web traffic out-put metrics. Only the number of unique visitors tothe firm's website metric is value-relevant. Themetrics that are not value-relevant include thenumber of page views, hours at the website, andaverage age and income of visitors. Hand's (2000)study suggests that econometric issues relating tothe dispersion of the data and the function used tomodel the relations between web traffic metrics,economic data, and firm value can impact the re-sults.

Firm-specific information advantages arisingfrom networks have been linked to stock price.Aral and Van Alstyne (2007) argue that the net-work metrics are associated with stock price per-formance because the networks provide access tonovel information. They find evidence consistentwith this proposition using output metrics for net-work advantage, comprising a ten-month panel ofemail communication patterns, message content,and performance data from a medium-sized exec-utive recruiting firm. They also find an upper limiton network benefits arising from diminishing mar-ginal productivity returns to novel information,consistent with theories of bounded rationality,and cognitive and information overload.

Aral and Weill (2007) provide evidence that theprior mixed results relating to the benefits from ITcapabilities are due to omitting strategy from theanalysis. They argue that investments in differentIT assets are guided by the firms' strategies (e.g.cost leadership or innovation). That is, the IT as-sets deliver value along dimensions consistentwith the underlying strategy. To provide insightson this hypothesis, they test the association of ITassets, IT capabilities, and strategy inputs, withfour dimensions of performance: market valuation,profitability, cost, and innovation. Aral and Weill's{2001) results suggest that the financial investmentin IT is not value-relevant. However, the combina-

tion of IT investment and IT capabilities drivesdifferences in firm performance and is value-rele-vant.

In summary, non-financial brand measures ofbrands appear on the strength of the results in thissection to be less reliably measured compared withthe financial measures used by Barth et al. (1998)discussed earlier. The studies in this section sug-gest that other information is important, in addi-tion to the fmancial information, for understandingthe contribution of financial information to per-formance and value. Further, the non-informationinteracts with the financial information rather thanentering the model additively. For example, notconditioning the IT investment on the firm's IT ca-pabilities and strategy provides an incomplete pic-ture of the value generated by the IT investmentand capabilities. Two issues of importance for de-signing and interpreting value-relevance studiesare therefore the heterogeneity of firm-specific ca-pabilities and their interdependence with the firm'sstrategic choices (Barney, 1991; Rumelt, 1984;Wernerfelt, 1984).

One final issue that relates generally to the com-petitive advantage literature is the apparent lack ofa conceptual framework for the building blocksand purposes of competitive processes. For exam-ple, the broad approach (e.g. balanced scorecard)seems more relevant to the purpose of strategyevaluation than to the purpose of firm valuation.This lack of structure may explain the proliferationof studies, targeting a wide range of capabilitiesand actions, which have been unable to generaterobust or generalisable evidence.

3.5. Human capitalEmployees create value for the firm by applying

their intellectual inputs and manual efforts in theworkplace. Human capital assets are heteroge-neous and therefore less predictable compared withphysical assets. However, investments in labourassets appreciate rather than depreciate with time(Webster, 1999). The incentive that this creates foremployers to invest in human capital is mitigatedby the employer's inability to own employees andvariation in the employees' commitment and relia-bility. This puts some of the incentives for educa-tion and training onto the employee.

There is a view that human capital is increasingin importance as a factor of production becausenew technologies are now more likely to be em-bodied in intangibles and labour rather than solelyin fixed capital (Kendrick, 1972; Webster, 1999).A number of studies suggest the increasing andoften specialised skill set required to participate insome occupations leads to a division (partitioning)of labour into periodic inputs (current expenses)and long-term assets (e.g. Webster, 1998). Thistrend suggests human capital is important for value


creation, particularly in high-skilled sections.Irrespective of property rights issues relating tohuman capital, firms requiring skilled labour tocompete would be expected to have strong incen-tives to invest in attraction, retention and motiva-tion of their human capital.

3.5.1. Human capital - management reportedassets

Despite anecdotal evidence that a proportion ofhuman capital is an asset, no attempt is made toidentify or report these items under GAAP.

There has been a long-running debate on thequestion of whether to capitalise labour compensa-tion costs as an intangible asset. One suggested ap-proach, similar to capital lease accounting, is toreport the discounted present value of estimatedcompensation costs as a non-current asset and a li-ability (Lev and Schwartz, 1971). The idea is thata going concern investing in plant, property andequipment commits to future compensation costsfor the life of those assets. Expenditures on em-ployees therefore reflect expected benefits and a li-ability to make continuing payments. This idea hasnot been adopted, and human capital assets are notreported under GAAP.

3.5.2. Human capital - researcher estimatedasset

In a series of papers, Rosett (2001, 2003) pro-vides evidence on the value-relevance, financialpolicies, and equity risk implications of imple-menting capitalisation of human capital. Rosett(2001) computes a human capital liability from thepresent value of expected compensation costs inunion labour contracts in the spirit of Lev andSchwartz (1971). In an earlier working paper(Rosett, 1997), he finds the corresponding humancapital asset is significantly positively associatedwith the market value of equity. In Rosett (2001),he finds the increase in leverage from the humancapital liability is positively associated with meas-ures of the firm's equity risk. Tests on industry sec-tors suggest the asset/liability measures are crudefor R&D and knowledge intensive industries,where the proportion of value generated from in-tellectual inputs is greater (i.e. where the partition-ing of labour on the basis of skill levels would begreatest).

Rosett (2003) focuses on the equity investmentrisk and corporate financial policy implication ofthe firm's liability for the human capital intangibleasset. Because labour is cosdy to adjust in theshort run, he argues that the firm has a fixed obli-gation to pay cash to labour, creating an off-bal-ance-sheet intangible liability similar to a lease.This liability creates a form of financial leveragerisk he calls labour leverage risk (total employ-ment deflated by the market value of equity) and

239

labour cost leverage (compensation costs deflatedby the market value of equity). He predicts andfinds the labour leverage measures are positivelycorrelated with equity investment risk, and nega-tively correlated with leverage and dividend pay-out consistent with managers taking human capitalrisk into account when setting financing and divi-dend policies.

Lajili and Zeghal (2006) construct human capi-tal productivity (marginal product of labour esti-mated from labour and training expenditures andproduction function regressions) and an efficiencyindicator (marginal product of labour minus aver-age industry labour costs) and relate these to stockprice performance. Risk-adjusted abnormal returnsare computed for portfolios sorted by size, labour-cost disclosure status, and the human capital indi-cators. They find higher levels of total labourexpenditures, workforce productivity, and efficien-cy is generally associated with higher abnormal re-turns. Hence, labour costs voluntarily disclosed infinancial statements are potentially useful for eval-uating human capital assets and value.

Abde-l-khalik (2003) constructs a measure ofmanagerial skills for executives on the Board ofDirectors. He employs a latent index regressionthat comprises a set of personal variables (experi-ence, risk preference, and value of owned shares)and a set of firm-specific variables (past profit andgrowth, organisational complexity, and operatingrisk). The predicted values from the latent variableregression are the measure of managerial skill.Abdel-khalik finds these predicted values arevalue-relevant.

Disclosures relating to managements' stock op-tion incentives are a human capital-related invest-ment that has been tested for value-relevance. Forexample. Landsman et al. (2004) find the employ-ee stock option (ESO) related costs are value-rele-vant. However, Landsman et al. (2004) provideevidence which suggests that only one of the fouraccounting methods for equity-based incentivesresults in accounting numbers that accurately re-flect the dilution effects of ESOs on shareholdervalue. This method involves the grant date recog-nition of an asset and a liability and subsequentmarking-to-market of the liability. Landsman et al.report that this method is the most value-relevantof the four accounting methods. The other threeless value-relevant methods are non-recognition(APB 25), recognition of only the stock option ex-pense (SFAS 123), and recognition of only a stockoption asset - as in the FASB's Exposure Draft:Share Based Payment (2004).

Measurement of dilutive effects is an issue. Liand Wong (2004) use a warrant-pricing model tojointly allow for the dilutive and shareholder valueimpact of employee stock options and use this inequity valuation. They find the market value of eq-

240

uity is overstated by 6% if the dilutive features ofstock options are ignored. A larger bias exists forheavy users of stock options, small firms, andR&D-intensive firms.

In summary, the evidence suggests human capi-tal assets measured using contract and non-finan-cial and fmancial input data are value-relevant.Managers appear to take the (off-balance sheet) li-abilities into account when setting financing anddividend policies, and these liabilities are positive-ly associated with equity risk. Investments inhuman capital productivity and efficiency are alsovalue-relevant as are measures of managerial skillsand their stock option compensation (although themeasurement of stock options and dilutive effectsis difficult). An area requiring future research ishow to more precisely measure human capital as-sets and liabilities to take skill level into account.

3.5.3. Human capital - annual expendituresGAAP has a limited role in the reporting of

labour costs. Labour costs data is collected byNational Statistical Bureaus in their annual sur-veys. Separate reporting of the expenditures paidto employees is envisaged under IAS 1Presentation of Financial Statements (paragraphs86-95).^' Despite this expectation, there is no evi-dence of widespread reporting of labour expendi-tures under GAAP. Separate reporting of labourexpenditures is voluntary in the US. There is a re-quirement to disclose employee costs and numberof employees under the UK Companies Act.^^Overall, the GAAP data on labour costs appears tobe limited.

In the US setting, Ballester et al. (2002) use amore fully specified regression based on Ohlson(1995), compared to Rosett's (1997) workingpaper discussed above, to examine the proportionof US labour costs that are value-relevant.Separate identification of labour costs in US finan-cial statements is voluntary and they find onlyabout 10% of all US Compustat firms disclosethese costs. Of these disclosed costs, only about16% are value-relevant, with an amortisation rateof 34% per year. Possibly the human capital meas-ures would yield more power in the tests if thelabour costs could be disaggregated to separate outhigher and lower specificity human capital assets(e.g. skilled/unskilled or partitioning according toscientist/engineer/management/sales).

In the spirit of the earlier discussion on the value

^' Paragraph 91 labels these expenditures 'employee bene-fits' which are defined in IAS 19 Employee Benefits to includeall benefits provided to employees in return for employeeservices.

* Thanks to Andy Stark for this information. The UKCompany Act where this provision was made can be viewed athttp://www.opsi.gov.uk/acts/acts2006/ukpga_20060046_en_l,paragraph 411.

ACCOUNTING AND BUSINESS RESEARCH

of skilled labour, Hansson (2004) predicts there isa value premium associated with human capitalthat distinguishes the stock performance of valueand glamour stocks. He finds that the dispersion inwage growth between value and growth stocks ex-plains a large proportion of the differences in stockreturns. The intuition for this result is that thevalue stocks are less exposed to shocks in rents tohuman capital. Hansson also finds that differencesin the labour force characteristics between valueand growth stocks are value-relevant.

Despite considerable corporate opposition, andconcerns about the reliability of fair value esti-mates of stock option expense, accounting stan-dards now require corporations to recogniseexpenses relating to grants of stock options to em-ployees - IFRS 2 Share-Based Payment and FASBSFAS 123 (revised 2004). Equity-based compen-sation is now measured at fair value on grant date,based on the estimated number of awards expect-ed to vest, and allocated as an expense over thevesting period.

Frederickson, Hodge and Pratt (2006) conductan experiment to study how stock option expenserecognition affects the valuation decisions of so-phisticated financial statement users. The subjectsare 220 business school alumni with an average of11 years' experience in financial analysis and16 years' accounting-related work experience.Seventy-nine percent are Certified PublicAccountants. An initial ex ante reliability assess-ment by the subjects is updated based on fourquestions relating to a comparison of stock optionexpense earnings versus no expense earnings.Frederickson et al. predict and find (1) users con-sider stock option expense recognised under aFASB mandate is more reliable than stock optionexpense voluntarily recognised by management;(2) users consider stock option expense voluntari-ly recognised on the income statement is more re-liable than stock option expense disclosed in thefootnotes; and (3) users invest more in a firm thatvoluntarily recognises stock option expense thanin a firm that discloses the expense in the foot-notes, even though voluntary recognition reducesreported net income. The results of this study sug-gest users perceive that stock option grants giverise to expenses and impound this information infirm value accordingly. Further, users behave as ifaccounting regulation sets the ground rules forcredible disclosure, and recognition signals relia-bility (as opposed to disclosure in the notes).

In summary, human capital measures computedfrom labour costs are value-relevant. However, re-searchers find these costs are only very sparselydisclosed and the costs are not sufficiently disag-gregated to provide precise measures of the humancapital assets. Behavioural research suggests ac-counting regulation of stock options has created


value for firms perceived to be appropriately andtransparently applying the standard.

3.5.4. Human capital - input metricsColombo and Grilli (2005) study how non-fi-

nancial input metrics of management quality relateto growth. They examine whether education andprior work experience are key capabilities of thefounders of technology companies that determinedifferences in the firms' growth. For a sample of506 young Italian companies in manufacturing andservices, they find the years of university educa-tion in economic and managerial fields, and to alesser extent in scientific and technical fields, arepositively related to growth but education in otherfields is not. Prior work experience in the same in-dustry of the new firm is positively associated withgrowth while prior work experience in other in-dustries is not. Technical work experience offounders rather than commercial work experiencedetermines growth. There are synergistic gainsfrom having complementary capabilities.

Several studies also report human capital man-agement practices are related to higher firm per-formance (e.g. Huselid, 1995; Ichniowski et al.,1997; Hitt et al., 2001).

Some industries are largely determined by thefirm's endowment of intellectual human capitalspecific to the dominant technology. One of theseis biotechnology (Zucker et al., 1998). Hand(2001) examines whether the human capital re-flected in employees is value-relevant for biotech-nology companies for whom skilled labourcomprising bioscientists and bioengineers is animportant factor of production. Because the rele-vant expenditures on hiring, retaining and incen-tivising employees are not separately reportedunder GAAP, Hand (2001) employs proxies ofhuman capital inputs - the total number of em-ployees - and the quality of human capital - theratio of SGA to the number of employees. It is pos-sible these measures are not precise enough to bevalue-relevant. For example, the number of em-ployees is commonly used to measure firm size ineconomic studies. Further, the number of employ-ees includes all workers, not just the biotechnolo-gy experts who generate the flrm's new scienceand/or technology. The results are consistent withthis conjecture. Using a log-log model, neither ofthese human capital measures are value-relevant.Hand (2001) reports that the GAAP variables(shareholders' equity, retained earnings, treasurystock, revenues, cost of sales, SGA, R&D, anddividends) explain about 70% of stock price.

In summary, some studies employ non-financialinput measures of management skills and find theyare associated with differences in firm growth.This is an important area for future research giventhe lack of GAAP disclosures capable of providing

241

insights on the contribution of human capital in-vestments to value.

3.5.5. Human capital - output metricsEdmans (2007) finds a non-financial output

measure of employee satisfaction, the companies'scores from the Best Companies to Work for inAmerica ranking, is associated with higher stockprice performance. This portfolio of firms alsooutperformed industry, and characteristicsmatched benchmarks.

Several studies find that output measures of thefirm's reputation are value-relevant. The idea isthat reputation increases the probability that value-relevant information is impounded into stock price(Healy and Palepu, 1993). For example. Blacket al. (1999) find reputation rankings based onFortune's America's Most Admired Companies ispositively significantly associated with the differ-ence between the market and book value of equity.Hutton and Stocken (2007) examine the effect offirm reputation for forecast accuracy on investors'reaction to managements' earnings f'orecasts usingsize-adjusted, three-day event window stock re-turns centred on the earnings release and manage-ment forecast. Their measure of forecastingreputation reflects prior forecast accuracy and fre-quency. They find that a forecasting reputationmakes investors more responsive to managementforecast news. A forecasting reputation leads to in-vestors' reaction at the management forecast datelargely pre-empting their earnings announcementstock response. However, the results suggest thatall firms do not build a forecasting reputation be-cause the cost outweighs the benefits when report-ed earnings do not reach management's forecast.

Using a field experiment that is co-linked to alaboratory experiment. List (2006) finds subjectsdrawn from a natural marketplace behave in ac-cordance with social preference models in the lab-oratory experiments. However, in their naturallyoccurring market settings, their behaviour betterapproximates self-interest. List finds the inci-dences of socially orientated behaviours in themarketplace are motivated by reputation concerns.Fisher and Heinkel (2007) study management'smotivation for truth-telling and reputation. In theirmodel, management builds reputation when timesare good and honesty is affordable, and exploitsreputation in times of need. However, competitionappears to constrain this potential managerialagency problem. Relying on the US business com-bination anti-takeover statutes passed between1985 and 1991 to measure variation in corporategovernance states, Giroud and Mueller (2007) flndthe loosening of corporate governance constraintsis associated with negative operating performanceand stock price effects only for firms in less com-petitive industries. Product market competition


thus appears to act as a brake on managerialagency problems, consistent with the view ofAlchian (1950), Friedman (1953) and Stigler(1958) that managerial slack cannot survive incompetitive industries.

In summary, output measures of employee satis-faction and firm reputation are correlated withvalue. Agency conflicts may negatively impact theex post propensity for management to act in accor-dance with good reputation, although competitionmay constrain this tendency.; An area for future re-search is input measures of reputation from theprevious section, as these are important for under-standing the 'causes' of reputation.

3.6. GoodwillThere is a long-lived debate over the conceptual

underpinnings of goodwill and indeed whetherpurchased goodwill is an asset. We have so far en-countered a lot of literature in this paper that canprovide insights on whether or not a firm has valu-able goodwill. It seems much more fruitful to lookat the earnings and value implications of specificand identifiable drivers of value.

3.6.1. Goodwill - management reported assetsGAAP-purchased goodwill is the difference be-

tween the acquisition price of a business or com-pany and the fair value of the identifiable netassets acquired by the acquiring entity (IFRS 3Business Combinations). Most studies find GAAPgoodwill is value-relevant (e.g. Chauvin andHirschey, 1994; McCarthy and Schneider, 1996;Vincent, 1994; Müller, 1994; Jennings, Robinson,Thompson and Duvall, 1996). Vincent (1994)finds the value-relevance relation holds for up tofive years after the acquisition of goodwill.

Chauvin and Hirschey (1994) examine thevalue-relevance of financial information on intan-gibles for a sample of US companies in1989-1991. They find GAAP goodwill, net in-come, advertising, R&D, intangible assets, andtangible assets are value-relevant for non-manu-facturing companies. These variables are all value-relevant for manufacturers, eixcept for the goodwilland intangible assets, possibly because more of themanufacturers' intangible assets are embodied inplant and equipment (Hansen and Serin, 1997).Goodwill is less value-relevant compared with theother intangible assets, suggesting goodwill is lessreliably measured. Müller (1994) also finds evi-dence consistent with this conclusion.

^' Emmanuel et al. (2004) argue that for models, which dis-aggregate an accounting measure like book value of equityinto separate components, a significant association with stockprice does not necessarily mean the market finds the compo-nent is value-relevant. The correct test, they argue, comparesthe coefficients of the component variables and the remainingbook value after the decomposition.

Amir et al. (1993) and Barth and Clinch (1995)examine the value-relevance of the goodwill ad-justment from the reconciliation of UK andAustralian GAAP to US GAAP. Both studies findthe goodwill adjustment is value-relevant. The UKfirms predominandy wrote off all goodwill toshareholders' equity prior to 1996. Hence, thegoodwill adjustment is the entire goodwill assetthat would have been recorded without the write-off to equity. In a separate analysis for the UKfirms, Barth and Clinch (1995) find this goodwilladjustment is value-relevant but less so than theother assets . ^

Goodwill amortisation is not value-relevant(Amir et al., 1993; Vincent, 1994; Muller, 1994;Jennings et al., 1995; Barth and Clinch, 1995).Clinch (1995) suggests this may reflect a percep-tion that some firms' goodwill is not declining invalue. It is also conceivable that the useful lifecannot be estimated (by managers or investors)and/or that the Henning et al. (2000) over-valua-tion component (discussed in Section 3.6.2.) dis-torts the amortisation charge.

Amortisation is now prohibited under interna-tional accounting standards and in other jurisdic-tions. Chambers (2007) examines whether thechange from amortisation of goodwill to an annu-al impairment test under the US standard, SFASNo. 142 Goodwill and Intangible Assets, increasedthe reliability of goodwill and its value-relevance.He finds the annual impairment testing is associat-ed with an increase in value-relevance. However,information appears to be lost as a result of theelimination of systematic amortisation, which issurprising given most studies find amortisationcharges are not value-relevant. Possibly, the im-pairment loss is understated, and this is an omittedvariable being picked up in the tests as a decreasein value-relevance that coincides with the cessa-tion of amortisation.

Consistent with this conjecture, Hayn andHughes (2006) find goodwill write-offs lag behindthe economic impairment of goodwill by an aver-age of three to four years. For one-third of theirsample, the delay can extend up to ten years. Theyfind their results for the period prior to the intro-duction of SFAS No. 142 are also generaiisable tothe goodwill reported under SFAS No. 142.Ramanna and Watts (2007) corroborate this evi-dence for a sample of firms with indications ofimpairment. The frequency of no goodwill impair-ment in their sample is about 71%. The propensitynot to impair is associated with financial charac-teristics that relate to higher management discre-tion from the unverifiable fair-value measures. Theevidence therefore suggests that GAAP goodwillunder the impairment test regime is value-relevantbut not reliably measured.

In summary, purchased goodwill is value-rele-


vant but less so than tangible assets and does notappear to be reliably measured. This conclusion isconsistent with the concepts of relevance and reli-ability, and their roles in value-relevance tests, asdiscussed in Section 2.3 and illustrated in Figure L

3.6.2. Goodwill - researcher estimated assetOne explanation for the lower value-relevance

of goodwill compared with other intangible assetsis that it is over-valued on average, on the balancesheet, relative to investor expectations. Consistentwith this idea, Henning, Lewis and Shaw (2000)find an over-valuation component which is nega-tively associated with stock price. They examinethe value-relevance of four components of good-will: (1) write-up of target firm assets to marketvalue; (2) going-concern value of the target; (3)synergy value created by the acquisition; and (4)over-valuation component. Consistent with theconclusion from the previous section that goodwillis not reliably measured, only the first three com-ponents are value-relevant, with the going concerncomponent dominating. Arguably, the synergycomponent is also difficult to value, and its relia-bility varies.

3.6.3. Goodwill - annual expenditures, inputmetrics and output metrics

Annual expenditures that relate to the firm'sgoodwill are not identified under GAAP standards.Measuring unrecorded intangible assets and theirimplications for firm value was one of the motiva-tions for investment opportunity set (IOS) studies.Smith and Watts (1992), Skinner (1993) and Gaverand Gaver (1993) use various proxies for thefirms' IOS and report that the IOS helps explainthe firms' accounting-based debt, dividend andmanagement compensation policy decisions.

Disaggregating the IOS and identifying andmeasuring constituent components is necessary togain further insights, and the literature reviewed inthis paper demonstrates the innovative ways thatresearchers have tackled this problem. Along theselines, Falk and Gordon (1977) proposed that good-will be defined as the total value of favourablemarket imperfections and related government reg-ulations, with purchased goodwill representing theamount one firm pays another firm for the sum ofthese assets. Their empirical work identifies 21categories of sources of goodwill under fourgroupings: imperfections in financial markets,labour markets, product markets, and governmentregulations. For example, labour market imperfec-tions relate to managerial talent, good labour rela-tions, training programs, and organisationalstructure of the acquired firm. They point out thatgoodwill is unobservable, and it is therefore easierto think about sources and measures directly.

243

4. Conclusions and future researchAt the 1996 SEC Symposium on FinancialReporting and Intangible Assets, Stiglitz (1996)discussed the importance of accounting (and audit-ing) for the workings of a capitalist society andmarkets, and the informational limitations of datafrom the accounting system, in fact data from anymeasurement system. He noted that while there isgoing to be a higher uncertainty associated withvaluing intangible assets than other assets, it is asource of major distortion to incentive systems tovalue intangibles at zero (1996: 17-19).

The main thrust of the value-relevance studiesreviewed in this paper is consistent with the viewsexpressed by Stiglitz (1996). There is a strong per-ception permeating the literature that learningabout the firm's investments in intangibles is im-portant for understanding how firms create (or de-stroy) value. This is consistent with the economicsof intangible investments as a key input to the pro-duction function.

This study compiles a somewhat voluminous re-view of a wide cross-section of studies on intangi-bles information. Such a wide-ranging approach ismotivated by the difficulty of judging whethervalue-relevance is due to relevance or reliabilityand the difficulty of obtaining direct tests of relia-bility.

4.1. Main findingsTable 1 summarises the main findings from the

literature review. The studies are grouped based onthe category of intangibles, the measurement ap-proach, and the value-relevance measure: stockprice level, stock returns or financial performance.The measurement categories reflect (1) the eco-nomics of the value creation processes and the re-searcher and practitioners' interests in theidentification of value drivers and their empiricalmeasures; (2) the influence of GAAP on the re-porting of intangibles and the research problems ofinterest to practitioners and researchers; and (3)the influence of management discretion.

The subtotals show that the literature is concen-trated in the R&D and IP category. Within this cat-egory, the research is concentrated in the annualR&D outlay and the output metrics measurementareas. The Jeast work in the R&D and IP area hasbeen done in the input metrics area, for whichthere is limited data. What goes in R&D is not dis-closed; therefore, it is difficult to know what inputmetrics would be relevant. The annual outlayemphasis reflects the effect of GAAP, which re-quires most R&D to be immediately expensed.The output metric emphasis reflects the interest ofresearchers and practitioners in alternative, non-accounting ways of measuring the success rate ofR&D inputs. It is often argued that the success rateinformation is there in the form of earnings.


However, it is not always feasible to wait untileamings. Further, earnings is an output that is notinformative about how the value was (or is expect-ed to be) created.

Stock price levels studies dominate the R&Dand IP category. Comparing the significance andtotals columns for the R&D and IP category, theinformation on intangibles always has significantcoefficients for the stock level studies. For thestock levels tests, the coefficients on the R&D andIP category intangibles tend to be larger than theother assets in the regression. This result is notwhat one would expect to find. This is an issue forfuture research: to try to design studies that pro-vide insights on the extent to which this result isdue to higher value-relevance versus measurementand research design issues (e.g. omitted variables).

The R&D and IP stock returns studies are con-centrated in the management reported and re-searcher estimated assets areas. The estimatedcoefficients are not always significant for the stockreturns studies. One explanation is that the stockreturns window, which is typically annual, is notwide enough to capture value-relevant informationin the management reported and researcher esti-mated R&D assets. Both these types of R&D as-sets are also subject to potentially significanteconomic uncertainty relating to the success rate.Consistent with this effect, the existing literatureindicates that a dollar of R&D is valued different-ly across time periods, industries and technologies.It would help investors if the causes of this varia-tion could be identified. There are opportunities tostudy the rate that R&D contributes to value andthe cause of changes in the success rate by focus-ing on stock return windows in which there is ashock, such as a significant invention (e.g. an im-portant drug that trials successfully). Providingsystematic evidence on these phenomena will helpto distinguish real effects in value-relevance testsfrom distortions due to research design issues.

For the brands and advertising category, themain focus is the output metrics which are themore readily available data. Output metrics relateto an important area of value creation, includingconstructs such as the long-term customer basevalue and customer retention. There is much lessresearch in the researcher estimated assets andinput metrics areas, presumably because GAAPrules do not require the reporting of brands and ad-vertising. Also, the advertising expenditures mightbe viewed as proprietary information by managers.In terms of value-relevance metrics, the stockprice levels studies dominate the brands and ad-vertising area and the estimated coefficients on in-tangibles in these studies are all significant. Forthe output metrics studies, the coefficients on in-tangibles items tend to be less that the other assetsin the tests, but not for the management reported

brand and advertising assets, which raises the issuefor future research of management reporting in-centives and/or research design issues.

For the customer loyalty category, output met-rics dominate. The output metric coefficients areall significant and the coefficients tend to be larg-er than those on the other assets in the regression.This is consistent with the importance of cus-tomers to a successful business but could also bedue to measurement error or omitted variables.Customer loyalty is an area where there is limitedfinancial accounting data. There is a demand forstudies that can provide insights into accountingregulators on the types of financial information re-lating to customer loyalty that are value-relevant.This type of research requires access to what iscurrently proprietary financial data. Understandinghow customer loyalty is generated and destroyedin different industries is a pre-requisite for identi-fying value-relevant information on customer loy-alty.

For the competitive advantage category, there islimited research in the management reported andresearcher estimated assets areas. Not all the coef-ficients are significant for the management report-ed assets link to financial performance. Thecompetitive advantage studies are concentrated inthe annual outlay, input and output metrics cate-gories. But the most work is in the input metricsarea. The competitive advantage studies are domi-nated by stock level and even more so by financialperformance studies. The coefficients are all sig-nificant and tend to be larger than the coefficientsfor other assets in the regression. This indicatesthat either the competitive advantage factors arevitally important or there is measurement error oromitted variables. The competitive advantage lit-erature is one area that would benefit from a gen-eral theory that defines and articulates the relevantvalue constructs, value creation processes and em-pirical measures. Such a framework is needed toincrease the logical validity and reliability of thebody of evidence and the generalisability of the re-sults.

The human capital category of intangibles hasno management reported assets, reflecting the im-pact of GAAP. The stock price studies are spreadacross the remaining four measurement categories,while the financial performance studies are con-centrated in the input metrics area. The coeffi-cients are always significant for the human capitalinformation. The coefficients tend to be smallerthan other assets in the regression for the stock lev-els but not for the stock returns and financial per-formance tests. This result suggests human capitallevels and changes are both important to firms.The inference is that the human capital contribu-tion to value can change in short time periods.Omitted variables are also a possible cause of the


larger coefficients for the returns and financial per-formance tests. However, this result is consistentwith the theory in this area. There are opportunitiesto design studies that can help accounting regula-tors to understand what financial data on humancapital is value-relevant. Studies of this typewould need access to what is currently proprietarydata.

The goodwill section is fully concentrated in themanagement reported assets, reflecting GAAP,and in the stock level studies for reasons which arenot clear. There are thus opportunities to studystock return and financial performance implica-tions of purchased goodwill. In particular, doespurchased goodwill contribute to future perform-ance and when? Is the purchased goodwill relationwith stock returns and financial performancechanging over time and how rapidly? There arenew research opportunities in the goodwill areaarising from the major change in GAAP fromamortisation to an annual impairment test, particu-larly arising from the evidence so far that the pur-chased goodwill balance is overstated on average.

4.2. Evidence on the reliability offtnancial andnon-financial information

The studies reviewed in this paper taken togeth-er suggest the expenditures on R&D and pur-chased goodwill are value-relevant but are notreliable indicators of the future benefits from theinvestments. For R&D, management knows whatexpenditures are bundled into R&D and how theseexpenditures are expected to create value (refer tothe value construct-to-value creation link inFigure 1). However, this information is not reli-ably reflected in the R&D measure partly becauseR&D bundles successful and unsuccessful efforts,and also because GAAP R&D bundles different,undisclosed types of expenditures that have differ-ent links to the generation of future benefits. Forpurchased goodwill, the link between the goodwillvalue construct and value creation is weak (seeFigure 1). This lack of definition means that theaccounting measure of goodwill cannot be reliable(see Figure 1 and Henning et al., 2000, in Section3.6.2). Hence, the goodwill measure is relevant butvaries in its reliability, a conclusion which is borneout by the empirical evidence (see Section 3.6).The evidence reviewed in this paper indicates thatthere is little point trying to evaluate the value im-plications of R&D by focusing only on the numberof R&D dollars spent. Section 3.1 refers to a rangeof factors that are relevant for evaluating R&D. Todate, Henning et al. (2000) is the only known studyto suggest how to empirically evaluate the value-relevant and reliable components of purchasedgoodwill.

It is difficult to make categorical statementsabout the reliability of most of the other informa-

245

tion items in this paper that researchers have stud-ied. In most cases, differences in value-relevancecould be due to differences in relevance, in relia-bility or differences in both relevance and reliabil-ity. What makes it a bit easier to make thisjudgment for R&D is the triangulation by Healy etal. (2002) and Kothari et al. (2002), using designsthat provide an economic benchmark with a knownvalue against which the R&D can be evaluated.Further, economic theory indicates that R&D isinherently uncertain, providing a strong a prioricase for the unreliability of R&D as a predictor offuture rents.

Given reliable measurement is important to ac-counting regulators and those users relying on fi-nancial accounting information, designing studiesto obtain direct tests of reliability is an importantarea for future research. One approach is to focuson settings where the value of intangibles is knownto be changing and employ stock returns to test forvalue-relevance. Another area for future researchis to identify economic benchmarks other thanstock price for the realisability of the expected fu-ture benefits from intangibles, and incorporatethese benchmarks into the value-relevance tests toprovide direct insights on reliability. For example,Healy et al. (2002) simulate a known firm valuewhich serves as a value-relevance benchmark.Another example is the Matolcsy and Wyatt(2008) study, which examines the value-relevanceof current earnings in the context of three differenttypes of technology conditions that are economicbenchmarks for expected growth and propertyrights effects.

4.3. Research design issuesThe studies canvassed in this paper also suggest

a range of factors that are potentially omitted vari-ables in value-relevance studies. Further, the fi-nancial information links to stock price often varyinteractively (and hence non-linearly) with factorssuch as the firm's resource endowments and strate-gic choices. In some circumstances, the relationbetween the information item of interest and stockprice is increasing (or decreasing) but at a declin-ing rate, a functional form which may be accom-modated by a non-linear function. Non-linearitiescan arise from life-cycle effects, firm specific ef-fects arising from the specialised nature of eachflrm's production function, firms reporting lossesversus profits, and the differing persistence ofearnings components (Das and Lev, 1994;Subramanyam, 1996; Lipe et al., 1998; Call et al.,2007). Thompson et al. (2001), provide an eco-nomic justification for using a log-linear form toestimate stock value based on accounting informa-tion. These model specification issues can be atleast partly addressed if researchers carefully artic-ulate the relevance and reliability links that are


illustrated in Figure 1.Some empirical measures appear to have a large

amount of measurement error (Boyd et al., 2004).Examples include non-financial measures of cus-tomer satisfaction, brands and human capital thatrely on informal survey data, subjective conceptu-al frameworks, and potentially imprecise, bluntmeasures such as the number of employees. Wherepossible, measuring actual iriputs and actual valuecreated is preferable to measuring only the percep-tions of these quantities obtained from surveys.

4.4. Trading-off management discretion andregulation

An issue for future research is the costs and ben-efits of management discretion to voluntarily re-port intangible assets versus regulation. Morefinancial reporting discretion gives managers theopportunity to report their firm's economic reality.However, agency conflicts can arise betweenstakeholders and managers. Financial reportingmay be overly optimistic or in the worst case sce-nario, misleading. Further, voluntary reportingwithout a standardising reporting framework forintangibles adversely impacts the interpretabilityof the resulting information. The adverse effects ofno regulation is evident frorn the history of 'intan-gible capital' reporting which so far has a bewil-dering range of measures but no conceptualframework or clear purpose(s) of measurement(Hunter et al., 2005). Regulation can have eco-nomic consequences if the regulations preventmanagers reporting the firm's economic reality(Anderson and Zimmer, 1992). Regulation canprovide benefits. For example, there is evidencethat investors perceive the regulated stock optioncompensation reporting is rnore reliable than theunregulated stock option reporting (Section 3.5.3).

There is evidence from the Australian settingthat management discretion to report intangible as-sets is associated with the financial reporting ofvalue-relevant identifiable intangible assets.Acquired and internally generated intangibles (butnot basic research) could be reported in theAustralian setting until the adoption of IFRS in2005. The evidence suggests the most discre-tionary items (the least regulated intangibles) arethe most value-relevant, which suggests that dis-cretion is associated with a balance of relevanceand reliability (Wyatt, 2005). Wyatt finds thatR&D assets and purchased goodwill are not value-relevant in this setting, where investors know thatmanagers have discretion to report identifiable in-tangible assets that are more informative about thesource of future benefits.

Contrast these results with the evidence (Section3.1 and 3.6) that R&D expenditures and purchasedgoodwill are value-relevant in countries wheremanagement has limited discretion to report more

precise indicators of future benefits. For example,the expected source of benefits from brands and li-cences relates to market power and the source ofexpected benefits from patents and trademarks aremonopolies over a specific invention or mark. Buthow will purchased goodwill benefit the acquirer?Would this goodwill be value-relevant if managershad more accounting discretion to report on intan-gibles?

What if regulators give firms discretion to reportintangible assets on the balance sheet constrainedonly by the definition and recognition criteria forassets and the statutory audit? The Australian ex-perience prior to the IFRS adoption in 2005 sug-gests that market efficiency might continue asbefore. In the Australian setting, the recognised in-tangible assets that are the least regulated are theassets that are associated with the firms' underly-ing economic reality (Wyatt, 2005); recognised in-tangible assets are associated with the generationof future earnings (Ritter and Wells, 2005); finan-cial analyst following is higher, and earnings fore-cast errors are lower for firms that recogniseintangible assets and have growth opportunitiesbut not for the extremely high or low growth op-portunities firms (Matolcsy and Wyatt, 2007). Thisevidence does not suggest that discretion to reportintangibles seriously impacts market efficiency.

Does the asymmetric treatment of acquired andinternally generated intangible assets achieve thedesired aim of increasing the reliability of report-ed intangibles? The international standard, IAS 38Intangible Assets assumes acquired intangibles arethe most relevant and reliable measures due toa market transaction. However, the conceptualdiscussion summarised in Figure 1 suggests rele-vance and reliability are determined jointly, not bymode of acquisition, but by the level of definitionof the value construct and value creation process,and the ability of the accounting measure to reflectexpectations about value creation. This conceptu-alisation (discussed in Section 2.3) and the empir-ical evidence (Sections 3.1 and 3.6) suggest theacquired goodwill and the internal R&D are not re-liable indicators of future benefits. Reliability isimportant when payoffs are specified in terms ofaccounting numbers (e.g. the measure of earningsavailable for distribution to shareholders as divi-dends). This is one reason for the pervasiveness ofaccounting conservatism. What, if any, are the eco-nomic consequences of recording these unreliableassets? How much discretion do managers have inapplying these asymmetric rules? What affectstheir judgment in implementing the IAS 38 asym-metric standards?

Some studies suggest recognition in the finan-cial statements is a more reliable signal comparedwith the alternative of disclosing in the notes to theaccounts (e.g. Ahmed et al., 2006). Consistent with


this idea, external auditors appear to permit moremisstatement in footnotes compared with recog-nised amounts (Libby et al., 2006). There are op-portunities to exploit the differences in GAAPacross countries to obtain insights on the effects ofaccounting conservatism, and recognition versusdisclosure, on the value-relevance of financial in-formation relating to intangibles.

Barth et al. (2003) find that recognition of ahighly unreliable accounting amount, rather thandisclosure, increases the information in stock priceif the recognised amount is relevant information.From Figure 1, when information is relevant themanner of value creation is reasonably well de-fined and the lack of reliability relates to the in-ability of the measure (e.g. R&D asset) toprecisely reflect the expected value creation. Theirstudy suggests that because of the imprecision ef-fects of aggregation (e.g. bundling expenditures toobtain an R&D number) that basing recognitiondecisions on reliability alone is too simplistic.Reliability relative to relevance is the key, not re-liability on its own. The evidence reviewed in thispaper is consistent with the Barth et al. (2003)findings in the sense that the information on intan-gibles is value-relevant in spite of the obviousproblems with reliable measurement.

4.5. Regulatory implicationsThe general problem of incomplete information

discussed in Section 2 suggests that more informa-tion is better even if it is uncertain. Accountingregulators have gone the other way, increasinglymoving to prevent firms measuring and reportinginternally generated intangible assets. However,even unreliable numbers can be useful signals that(unobservable) assets exist, pointing investors inthe direction of additional relevant informationsources. For example, a patent measured andrecorded at £1 on the balance sheet is informativeif it signals the existence of a patent for which de-tailed information is publicly available to anyonewho cares to search the public patent office onlinedatabases.

One gap in financial reporting that is evidentfrom the review in this paper is the reporting ofseparate line items of expenditures on intangiblesin the income statement. The review suggests thereare deficiencies in reporting labour expenditures,advertising and marketing, the components ofR&D, and expenditures relating to the generationof customer loyalty, IP, and competitive advan-tage. If all companies were required to disclosebroad categories of expenditures on intangibles(more comprehensively than the narrow R&D se-ries), this might level the playing field and allevi-ate the risk of unilateral information spillovers.

Financial accounting is only one source of infor-mation about intangibles. This paper highlights a

247

range of other non-financial sources of informa-tion that are value-relevant. This evidence is con-sistent with Whisenant (1998), who providesevidence that value-relevance is not solely a func-tion of GAAP. Instead, investors use financialstatement analysis techniques and recognised datain the annual report, including the notes to the ac-counts, to adjust the financial statement informa-tion before using it in their valuation models.Investors do not expect financial information tostand alone. A question for future research is towhat extent the gaps in the financial reporting onintangibles are already addressed by non-financialsources of information.

Finally, an implication of the evidence reviewedin this paper is that accounting regulators mightbetter facilitate value-relevant disclosures on in-tangibles if they give discretion to management toreport their firm's economic reality (as in theAustralian experience). To be interpretable, ac-counting standards are needed as guidance formanagers. To be relevant, the standards need to bebenchmarked to the economics of the intangibleinvestments so that compliance means the firmsreport in accordance with their firm's economics(e.g. the technical feasibility test in SFAS No. 86Accounting for the Costs of Computer Softwareto Be Sold, Leased, or Otherwise Marketed).Research that identifies which of the firms' expen-ditures create value, and how, is important for as-sisting regulators to promulgate economicallyrelevant accounting standards.

Regulators can efficiently oversee the exerciseof management's financial reporting discretionusing an electronic financial reporting surveil-lance. An example of a reportedly effective systemsuggested by Bayley and Taylor (2007) uses fairlysimple financial statement analysis techniques thatpinpoint firms engaged in the management ofGAAP financial statements outside an acceptablebound.

ReferencesAbdel-khalik, A. R. (1975). 'Advertising Effectiveness and

Accounting Policy'. The Accounting Review, October:657-70.

Abdel-khalik, A. R. (2003). 'Self-Sorting, IncentiveCompensation, and Human-Capital Assets'. EuropeanAccounting Review, 12 (4): 661-697.

Abernathy, W. J. and Clarke, K. B. (1985). 'Innovation:Mapping the Winds of Creative Destruction'. ResearchPolicy, 14, 3-22.

Abernethy, M., Bianchi, P., Labory, S., Lev, S., Wyatt, A.,Zambón, S. (2003). 'Study on the Measurement andReporting of Intangible Assets and Associated ReportingPractices', prepared for the Commission of the EuropeanCommunities - Enterprise Directorate General Contract,contract n.FIF.20010720.

Aboody, D. and Lev, B. (1998). 'The Value Relevance ofIntangibles: The Case of Software Capitalisation'.Journal of Accounting Research, 36, Supplement:


161-191.Abramovitz, M. (1993). 'The Search for the Sources ofGrowth: Areas of Ignorance, Old and New'. The Journalof Economic History, 53 (2): 217-243.

Akbar, S. and Stark, A. W. (2003). 'Deflators, NetShareholder Cash Flows, Dividends, CapitalContributions and Estimated Models of CorporateValuation'. Journal of Business Finance & Accounting,30(9-10): 1211-1233.

Accounting Standards Board (1989). SSAP 13 Accountingfor Research and Development.

Accounting Standards Board (1991). 'The Objective ofFinancial Statements and the Qualitative Characteristicsof Financial Information', Exposure Draft (SP), July.

Accounting Standards Board (1997). FRSIO Goodwill andIntangible Assets.

Accounting Standards Board (1999). Statement ofPrinciples.

Accounting Standards Board, Foreningen ofStatsautoriserede Revisor, Deutches RecchnungslegungsStandards Committee, Komitet StandardowRachunkowoci and EFRAG (2007). Stewardship/Accountability as an Objective of Financial Reporting,Pro-Active Accounting Activities in Europe.

Agarwal, R. and Gort, M. (2001). First Mover Advantageand the Speed of Competitive Entry, 1887-1986. Journalof Law and Economics, 44 (\): 161-177.

Ahmed, A. S., Kilic, E. and Lobo, G. J. (2006). 'DoesRecognition versus Disclosure Matter? Evidence fromValue-Relevance of Banks' Recognized and DisclosedDerivative Financial Instruments' The AccountingReview, S\ (3): 567-588.

Alchian, A. (1950). 'Uncertainty, Evolution, and EconomicTheory'. Journal of Political Economy, 58: 211-221.

American Accounting Association Financial AccountingStandards Committee (2001). Comment Letter on theFASB proposal 'Disclosure of Information aboutIntangible Assets'.

American Institute of Certified Public Accountants(AICPA) (2004). http://accounting.rutgers.edu/raw/aicpa/index.htm.

Amir, E., Harris, T. and Venuti, E. (1993). 'A Comparisonof the Value-Relevance of U.S. versus Non-U.S. GAAPAccounting Measures Using Form 20-F Reconciliations.Journal of Accounting Research, 31: 230-264.

Amir, E., Guan, Y. and Livne, G. (2007). 'The Associationof R&D and Capital Expenditures with SubsequentEarnings Variability, Journal of Business'. Finance &Accounting, 34 (1 and 2): 222-246.

Amir, E. and Lev, B. (1996). 'Value-Relevance ofNonfinancial Information: The Wireless CommunicationsIndustry'. Journal of Accounting and Economics, 22:3-30.

Anderson, E. W., Fomell, W. and Lehmann, D. R. (1994).'Customer Satisfaction, Market Share and Profitability:Findings from Sweden'. Journal of Marketing Research,58 (3): 53-66.

Anderson, M. C , Banker, R. D., and Ravindran, S. (2006).'Value Implications of Investments in InformationTechnology'. Management Science, 52 (9): 1359-1376.

Anderson, D. and Zimmer, I. (1992). 'Reactions toRegulation of Accounting for Goodw'úV. Accounting andFinance, November: 27-50.

Anthony, J. H. and Ramesh, K. (1992). 'Association be-tween Accounting Performance Measures and StockPrices - A Test of the Life Cycle Hypothesis'. Journal ofAccounting and Economics, 15 (2-3): 203-227.

Aral, S., and Van Alstyne, M. W. (2007). 'NetworkStructure and Information Advantage', Working Paper,

NYU Stem School of Business; Massachusetts Instituteof Technology (MIT) - Sloan School of Management andBoston University - Department of ManagementInformation Systems; MIT - Center for E-Business.

Aral, S. and Weill, P. (2007). 'IT Assets, OrganizationalCapabilities and Firm Performance: Do ResourceAllocations and Organizational Differences ExplainPerformance Variation?' Working Paper, NYU SternSchool of Business and Massachusetts Institute ofTechnology (MIT) - Sloan School of Management.

Ashton, R. H. (2005). 'Intellectual Capital and ValueCreation: A Review'. Journal of Accounting Literature,24: 53-134.

Audretsch, D. (1995). Innovation and Industry Evolution,Massachusetts Institute of Technology (MIT), USA.

Austin, D. (1993). 'An Event Study Approach toMeasuring Innovative Output: T'he Case ofBiotechnology'. American Economic Review, 83:253-258.

Ball, R. and Brown, P. (1968). 'An Empirical Evaluation ofAccounting Income Numbers'. Journal of AccountingResearch, 6 (2): 159-178.

Ballester, M., Livnat, J. and Seethamraju, C. (1998).'Individual-Firm Style Loadings, Unrecorded EconomicAssets, and Systematic Risk'. Journal of Accounting,Auditing and Finance, 13 (3): 275-296.

Ballester, M., Livnat, J. and Sinha, N. (2002). 'Labor Costsand Investments in Human Capital'. Journal ofAccounting, Auditing and Finance, 17 (4): 351-373.

Banker, R. D., Chang, H. and Majumdar, S. K. (1995).'The Consequences of Evolving Competition on theComponents of Firms' Profits: Recent Evidence from theU.S. Telecommunications Industry'. InformationEconomics and Policy, 7: 37-56.

Banker, R. D., Potter, G. and Srinivasan, D. (2000). 'AnEmpirical Investigation of an Incentive Plan that IncludesNonfinancial Performance Measures'. The AccountingReview,15 (1): 65-92.

Barney, J. (1991). 'Firm Resources and SustainedCompetitive Advantage'. Journal of Management, 17:99-120.

Barney, J. (2001). 'Resource-Based Theories ofCompetitive Advantage: A Ten-Year Retrospective on theResource-Based View'. Strategic Management Journal,27: 643-650.

Barth, M. E., Clement, M. B., Foster, G. and Kasznik, R.(1998). 'Brand Values and Capital Market Valuation'.Review of Accounting Studies, 3: 41-68.

Barth, M. and Clinch, G. (1995). 'International accountingdifferences and their relation to share prices: evidencefrom U.K., Australian, and Canadian firms', unpublishedmanuscript.

Barth M. E. and Clinch, G. (1998). 'Revalued Financial,Tangible, and Intangible Assets: Association with SharePrices and Non-Market-Based Value Estimates'. Journalof Accounting Research, 36: 199-233.

Barth, M. E., Beaver, W. H. and Landsman, W. R. (2001).'The Relevance of the Value Relevance Literature forFinancial Accounting Standard Setting: Another View'.Working Paper, Stanford University.

Barth, M. E., Clinch, G. and Shibano, T. (2003). 'MarketEffects of Recognition and Disclosure'. Journal ofAccounting Research, 41 (4): 581-609.

Bayley, L. and Taylor, S. L., 2007. 'Identifying EarningsOverstatements: A Practical Test'. Working Paper,University of New South Wales.

Ben Zion, U. (1978). 'The Investment Aspect ofNonproduction Expenditures: An Empirical Test'.Journal of Economics and Business, (Spring-Summer):


224-229.Ben Zioti, U. (1984). 'The R&D and Investment Decisionand Its Relationship to the Firm's Market Value: SomePreliminary Results', in: Z. Griliches (ed.), R&D Patentsand Productivity. Chicago, IL: University of ChicagoPress.

Bemdt, E., Bui, L. T., Reiley, D. and Urban, G. L. (1994).'The Roles of Marketing, Product, Quality and PriceCompetition in the Growth and Composition of the U.S.Anti-Ulcer Drug Industry'. NBER Working Paper No.W4904.

Bernstein, J. I. and Nadiri, M. I. (1982). 'Financing andInvestment in Plant and Equipment and Research andDevelopment'. NBER Working Paper No. W1017.

Bessen, J. E. (2006). 'Estimates of Firms' Patent Rentsfrom Firm Market Value'. Working Paper, BostonUniversity School of Law.

Bharadwaj, A. S., Bharadwaj, S. G. and Konsynski, B. R.(1999). 'Information Technology Effects on FirmPerformance as Measured by Tobin's q'. ManagementScience, 45(1): 1008-1024.

Biddle, G.,Bowen,R. and Wallace, J. (1997).'Does EVA®Beat Eamings? Evidence on the Associations with StockReturns and Firm Values'. Journal of Accounting andEconomics, 24: 301-336.

Black, E., L., Cames, T. A. and Richardson, V. J. (1999).'The Market Valuation of Firm Reputation'. WorkingPaper, Brigham Young University, Berry College andUniversity of Arkansas at Fayetteville.

Boone, J. P. and Raman, K. K. (2001). 'Off-Balance SheetR&D Assets and Market Liquidity'. Journal of Accountingand Public Policy, 20 (2): 97-128.

Borden, N. (1942). The Economic Effects of Advertising,Chicago, Irwin.

Boyd, B. K., Gove, S. and Hitt, M. A. (2005). 'ConstructMeasurement in Strategic Management Research: Illusionor Reality?'. Strategic Management Journal, 26 (3):239-257.

Boyer, K. D. (1974). 'Informative and GoodwillAdvertising'. The Review of Economics and Statistics, 56(4): 541-548.

Brynjolfsson, E. (1993). 'The Productivity Paradox ofInformation Technology', Communications of the ACM,36 (12): 66-77.

Brynjolfsson, E., Hitt, L. and Yang, S. (2002). 'IntangibleAssets: Computers and Organizational Capital'.Brookings Papers on Economic Activity, 1: 137-199.

Bublitz, B. and Ettredge, M. (1989). 'The Information inDiscretionary Outlays: Advertising, Research, andDevelopment'. The Accounting Review, 64 (1): 108-124.

Call, A. C , Hewitt, M. and Shevlin, T. (2007), 'Investors'Mispricing of Eamings: A Model of Persistence and theStock Price Consequences'. Working Paper, University ofWashington.

Campbell, D., Datar, S., Kulp, S. C. and Narayanan, V. G.(2002). Using the Balanced Scorecard as a ControlSystem for Monitoring and Revising Corporate Strategy,Harvard Business School and George WashingtonUniversity.

Cantwell, J. (1989). Technological Innovation andMultinational Corporations. Oxford: Basil Blackwell.

Caves, R. (1982). Multinational Enterprise and EconomieAnalysis. New York, NY: Cambridge University Press.

Caves, R. E. and Murphy, W. F. (1976). 'Franchising:Firms, Markets, and Intangible Assets', SouthernEconomic Journal, 42 (4): 572-586.

Ceccagnoli, M., Arora, A., Cohen, W. and Vogt, W. (1998).'R&D, Knowledge Spillovers, and Competition amongFirms with Asymmetric Technological Capabilities'.

Working Paper, Camegie Mellon University.Chambers, D. J. (2007). 'Has Goodwill Accounting under

SFAS 142 Improved Financial Reporting?'. WorkingPaper, University of Kentucky.

Chambers, D. J., Jennings, R. and Thompson, R. B. (2002).'Excess Retums to R&D-lntensive Firms'. Review ofAccounting Studies, 1 (2-3), 133-158.

Chan, S-H. Martin, J. and Kensinger, J. (1990). 'CorporateResearch and Development Expenditures and ShareValue'. Journal of Einancial Economics, 26: 255-276.

Chauvin, K. W. and Hirschey, M. (1993). 'Advertising,R&D Expenditures and the Market Value of the Firm'.Einancial Management, 22: 128-140.

Chauvin, K. W. and Hirschey, M. (1994). 'Goodwill,Profitability, and the Market Value of the Firm'. Journalof Accounting and Public Policy, 13: 159-180.

Cheng, Q. (2005). 'What Determines Residual Income?'.The Accounting Review, 80 (1), 85-112.

Cheng, A., Hopwood, W. S. and McKeown, J. C. (1992).'Non-Linearity and Specification Problems inUnexpected Eamings Response Regression Model'. TheAccounting Review, 67 (3): 579-598.

Glarke, D.G. (1976). 'Econometric Measurement of theDuration of Advertising Effects on Sales'. Journal ofMarketing, 13 (4): 345-357.

Clinch, G. (1995). 'Capital markets research and the good-will debate'. Australian Accounting Review 5: 22-30.

Cockbum, I. and Griliches, Z. (1988). 'Industry Effects andAppropriability Measures in the Stock Market's Valuationof R&D and Patents'. American Economic Review, 78(2): 419-423.

Cohen, W. (1995). 'Empirical Studies of InnovativeActivity', in P. Stoneman (ed.). Handbook of theEconomics Of Innovation and Technological Change.Cambridge, Massachusetts: Basil Blackwell Ltd.

Cohen, W. M., and Klepper, S. (1992). 'The Anatomy ofIndustry R&D Intensity Distributions'. AmericanEconomic Review, 82: 773-788.

Cohen, W. M. and Levinthal, D. A. (1989). 'Innovation andLeaming: The Two Faces of R&D - Implications for theAnalysis of R&D Investment'. Economic Journal, 99:569-596.

Cohen, W. M., Nelson, R. R. and Walsh, J. P. (2000).'Protecting Their Intellectual Assets: AppropriabilityConditions and Why U.S. Manufacturing Firms Patent(Or Not)'. NBER Working Paper 7552, Cambridge,Massachusetts.

Colombo, M. G. and Grilli, L. (2006). 'Founders' HumanCapital and the Growth of New Technology-Based Firms:A Competence-Based View'. Research Policy, 34 (6):795-816.

Connolly, R. A., Hirsch, B. T. and Hirschey, M. (1986).'Union Rent Seeking, Intangible Gapital, and MarketValue of the Firm'. The Review of Economics andStatistics, 68 (4): 567-577.

Connolly, R.A. and Hirschey, M. (1990). 'Firm Size andR&D Effectiveness: A Value-Based Test'. EconomicsUtters, 32 (3): 2n-2&\.

Darby, M. R., Liu, Q. and Zucker, L. G. ( 1999). ' Stakes andStars: The Effect of Intellectual Human Capital on theLevel and Variability of High-Tech Firms' MarketValues'. NBER Working Paper No. W7201.

Das, S. and Lev, B. (1994). 'Nonlinearity in the Retums-Earnings Relation: Tests of Altemative Specifications andExplanations'. Contemporary Accounting Research, 11:353-379.

Deng, Z. and Lev, B. (1998). 'The Valuation of AcquiredR&D'. Working Paper, Stem School of Business, NewYork University.


Dewan, S. and Min, C. (1997). 'The Substitution ofInformation Technology for Other Factors of Production:A Firm Level Analysis'. Management Science, 43 (12):1660-1675.

Dosi, G. (1988). 'Sources. Procedures and MicroeconomicEffects of Innovation'. Journal of Economic Literature,36: 1126-1171.

Dube, J., Hitsch, G. J. and Manchanda, P. (2004). 'AnEmpirical Model of Advertising Dynamics'. WorkingPaper, University of Chicago.

Edmans, A. (2007). 'Does the Stock Market Fully ValueIntangibles? Employee Satisfaction and Equity Prices'.Working Paper, University of Pennsylvania.

Ely, K. and Waymire, G. (1999). 'Intangible Assets andStock Prices in the Pre-SEC Era'. Journal of AccountingResearch, 37, Studies on Credible Financial Reporting:17-44.

Emanuel, D., Wong, N. and Hunt, A. (2004). Testing for aCross-Sectional Relationship Between Market andAccounting Numbers, University of Auckland andCharles River Associates.

Ethiraj, S. E., Kale, P., Krishnan, M. S. and Singh, J.(2005). 'Where Do Gapabilities Come from and How DoThey Matter? A Study in the Software Services Industry'.Strategic Management Journal, 26: 25-45.

Ettredge, M. and Bublitz, B. (1988). 'A Reexamination ofMarket-Assessed Intangible Assets: Advertising andR&D'. Working Paper, Washington State University andUniversity of Kansas.

Falk, H., and Gordon, L. (1977). 'Imperfect Markets andthe Nature of Goodwill', Journal of Business Finance andAccounting, 4 (4):, 443^62.

Financial Accounting Standards Board (1972). AccountingPrinciples Board Opinion No. 25, Accounting for StockIssued to Employees.

Financial Accounting Standards Board (1974). Statementof Financial Accounting Standards 2, Accounting forResearch and Development Costs.

Financial Accounting Standards Board (1985). Statementof Financial Accounting Standards 86, Accounting for theCosts of Computer Software to Be Sold, Leased orOtherwise Marketed.

Financial Accounting Standards Board (1995). Statementof Financial Accounting Standards 123, Accounting forStock-Based Compensation.

Financial Accounting Standards Board (2001). Proposalfor a New Agenda Project: Disclosure of Informationabout Intangible Assets Not Recognized in EinancialStatements, 17 August.

Financial Accounting Standards Board (2001). Statementof Financial Accounting Standards 142, Goodwill andOther Intangible Assets.

Financial Accounting Standards Board (2004). Statementof Financial Accounting Standards 123, Share-BasedPayment.

Financial Accounting Standards Board (2004), ExposureDraft, Share Based Payment. FASB, Norwalk, CT.

Fisher, I. (1930). The Theory of Interest. New York, NY:Macmillan.

Fisher, A. J. and Heinkel, R. L. (2007). Reputation andManagerial Truth-Telling as Self-Insurance, University ofBritish Columbia.

Ford, J. L. (1994). G.L.S. Shackle: The DissentingEconomist's Economist, Edward Elgar, Aldershot.

Franses, P. H. and Vriens, M. (2004). 'Advertising Effectson Awareness, Consideration and Brand Choice UsingTracking Data'. Working Paper, Erasmus ResearchInstitute of Management, Erasmus University ofRotterdam.

Frederickson, J., Hodge, F. and Pratt, J. (2006). 'TheEvolution of Stock Option Accounting: Disclosure,Voluntary Recognition, Mandated Recognition andManagement Disavowals'. Working Paper, MelbourneBusiness School, University of Washington, IndianaUniversity, Bloomington.

Freeman, R. N. and Tse, S. Y. (1992). 'A Nonlinear Modelof Security Price Responses to Unexpected Earnings'.Journal of Accounting Research, 30 (2): 185-209.

Friedman, M. (1953). The Methodology of PositiveEconomic, in Essays on Positive Economics. Chicago:University of Chicago Press.

Fudge, A. (2005). 'Brand Energy: A Compass forNavigating Troubled Waters'. Marketing Science InstituteConference on Connecting with Customers in a ComplexWorld.

Gates, S. (1999). Aligning Strategic Performance Measuresand Results. New York, NY: The Conference Board.

Gaver, J., and Gaver, M. (1993). 'Additional Evidence onthe Association Between the Investment Opportunity Setand Corporate Financing, Dividend, and CompensationPolicies'. Journal of Accounting and Economics, 16:125-160.

Geroski, P. A. (1995). Chapter 4 'Markets for technology:knowledge, innovation and appropriability', in: P.Stoneman, (ed.). Handbook of the economics of innova-tion and technological change. Great Britain: BlackwellPublishers Ltd: 90-131.

Geroski, P. A. (2000). 'Models of Technology Diffusion'.Research Policy, 29: 603-625.

Giroud, X. and Mueller, H. M. (2007). 'Does CorporateGovernance Matter in Competitive Industries?'. WorkingPaper, New York University.

Givoly, D. and Shi, C. (2007). 'Accounting for SoftwareDevelopment Gosts and the Gost of Gapital: Evidencefrom IPO Underpricing in the Software Industry'.Working Paper, Pennsylvania State University andUniversity of California, Irvine.

Gjerde, O., Knivsfla, K. and Saettem, F. (2007). 'Evidenceon Competitive Advantage and Superior Stock MarketPerformance'. Working Paper, Norwegian School ofEconomics and Business Administration.

Grabowski, H. G. (1976). 'The Effect of Advertising on theInterindustry Distribution of Demand'. Explorations inEconomic Research, 3: 21-75.

Grabowski, H. G., and Mueller, D. G. (1978). 'IndustrialResearch and Development, Intangible Capital Stocks,and Firm Profit Rates'. The Bell Journal of Economics:328-342.

Graves, S. B. (1989). 'Why Gosts Increase When ProjectsAccelerate'. Research Technology Management, 32 (2):16-18.

Green, J. P., Stark, A. W. and Thomas, H. M. (1996). 'UKEvidence on the Market Valuation of Research andDevelopment Expenditures'. yoiir«a/ of Business Financeand Accounting, ITi (2): 191-216.

Greenhalgh, G. and Rogers, M. (2006). 'The Value ofInnovation: The Interaction of Gompetition, R&D andir. Research Policy, 35: 562-580.

Griliches, Z. (1981). 'Market Value, R&D, and Patents'.Economic Letters,!: 183-187.

Griliches, Z. (1990), 'Patent Statistics as EconomicIndicators: A Survey'. Journal of Economic Literature,28: 1661-1707.

Griliches, Z. (1994). 'Productivity, R&D, and the DataConstraint, Presidential Address delivered at the one-hun-dred sixth meeting of the American EconomicAssociation', The American Economic Review, 84, 1,1-23.


Griliches, Z., Hall, B. H. and Pakes, A. (1991). 'R&D,Patents, and Market Value Revisited: Is There Evidenceof A Second Technological Opportunity Related Factor?'.NBER Working Paper 2624, National Bureau ofEconomic Research.

Griliches, Z., Pakes, A. and Hall, B. (1987). 'The Valueof Patents as Indicators of Inventive Activity', inP. Dasgupta and P. Stoneman (eds.). Economic Policyand Technological Performance. Cambridge: CambridgeUniversity Press.

Guay, W. and Verrecchia, R. E. (2007). 'ConservativeDisclosure'. Working Paper, The Wharton School,University of Pennsylvania.

Gupta, S., Lehman, D. R. and Stuart, J. A. (2004). 'ValuingCustomers'. Journal of Marketing Research, XLI,February: 7-18.

Hall, B. (2000). 'Innovation and Market Value', in Barrell,R., G. Mason and M. O'Mahoney (eds.). Productivity,Innovation and Economic Performance. Cambridge UK:Cambridge University Press.

Hall,B. H. (1991). 'The Impact of Corporate Restructuringon Industrial Research and Development'. NBERWorking Paper No. W3216.

Hall, B. H. (1993a). 'The Stock Market's Valuation ofR&D Investment During the 1980s'. American EconomicReview, 83 (2): 259-264.

Hall, B. H. (1993b). 'Industrial Research During the 1980s:Did the Rate of Return Fall?'. Brookings Papers onEconomic Activity: Microeconomics, 2: 289-393.

Hall, B. H. and Kim, D. (1999) Valuing Intangible Assets:tiie Stocic Market Value of R&D Revisited. UC Berkeley,Harvard University, and NBER.

Hall, B. H., Jaffe, A. B. and Trajtenberg, M. (2001). 'TheNBER Patent Citation Data File: Lessons, Insights andMethodological Tools. NBER Working Paper No. 8498.

Hall, B. H., Jaffe, A., and Trajtenberg, M. (2005). 'MarketValue and Patent Citations'. RAND Journal of Economics,36(1): 16-38.

Hand, J. (2000). 'The Role of Economic Fundamentals,Web Traffic, and Supply and Demand in the Pricing ofU.S. Internet Stocks'. Working Paper, University of NorthCarolina at Chapel Hill.

Hand, J. (2001). 'The Market Valuation of BiotechnologyFirms and Biotechnology R&D'. Working Paper,University of North Carolina at Chapel Hill.

Hand, J. (2003). 'The Value Relevance of FinancialStatements in Private Equity Markets'. Working Paper,University of North Carolina at Chapel Hill.

Hansen, P. A., and Serin, G. (1997). 'Will Low TechnologyProducts Disappear? The Hidden Innovation Processes inLow Technology Industries'. Technological Eorecastingand Social Change, 55: 179-191.

Hansson, B. (2004). 'Human Capital and Stock Returns: Isthe Value Premium an Approximation for Return onHuman Capital?'. Journal of Business, Einance andAccounting, 3\ (3-4): 333-358.

Hayek, F. A. (1945). 'The Use of Knowledge in Society'.The American Economic Review, XXXV (4): 519-530.

Hayn, C. and Hughes, P. J. (2006). 'Leading Indicators ofGoodwill Impairment'. Journal of Accounting, Auditing,and Einance, 2\ (3): 223-265.

Healy, P. M. and Palepu, K. G. (1993). 'The Effect ofFirms' Financial Disclosure Strategies on Stock Prices'.Accounting Horizons, 1: 1-11.

Healy, P. M., Myers, S. C. and Howe, C. (2002). 'R&DAccounting and the Tradeoff Between Relevance andObjectivity'. Journal of Accounting Research, 40 (3):677-710.

Henkel, J., and Reitzig, M. (2007). 'Patent Sharks and the

Sustainability of Value Destruction Strategies'. WorkingPaper, Munich University of Technology, Faculty ofEconomics and Business Administration, and LondonBusiness School.

Henning, S. L., Lewis, B. L., and Shaw, W. H. (2000).'Valuation of the Components of Purchased Goodwill'.Journal of Accounting Research, 38 (2): 375-386.

Hirschey, M. (1982). 'Intangible Capital Aspects ofAdvertising and R&D Expenditures'. Journal ofIndustrial Economics, 4: 375-389.

Hirschey, M., Richardson, J., and Scholtz, S. (1998).'Value Relevance of Nonfinancial Information: The Caseof Patent Data'. University of Arkansas and University ofKansas.

Hirschey, M., and Weygandt, J. J. (1985). 'AmortisationPolicy for Advertising and Research and Development'.Journal of Accounting Research, 23: 326-335.

Hitt, L. M., and Brynjolfsson, E. (1996). 'Productivity,Business Profitability, and Consumer Surplus: ThreeDifferent Measures of Information Technology Value'.MIS Quarterly, 20 {ly 121-142.

Hitt, M. A., Bierman, L., Shimizu, K., and Kochhar, R.(2001). 'Direct and Moderating Effects of Human Capitalon Strategy and Performance in Professional ServiceFirms. A Resource Based Perspective'. Academy ofManagement Journal, AA\ 13-28.

Holthausen, R., and Watts, R. (2001). 'The Relevance ofthe Value relevance Literature for Financial AccountingStandard Setting'. Journal of Accounting and Economics,31: 3-75.

Hunter, L. C , Webster, E., and Wyatt, A. (2005).'Techniques for Measuring Intangible Capital: A Reviewof Current Practice'. The Australian Accounting Review,15 (36): 4-21.

Hunter, L. C , Webster, E., and Wyatt, A. (2007). 'AnAccounting Approach to Estimating the Return onIntangible Investments'. Working Paper, University ofTechnology, Sydney and the Intellectual PropertyResearch Institute of Australia, University of Melbourne.

Huselid, M. A. (1995). 'The Impact of Human ResourceManagement Practices on Turnover, Productivity, andCorporate Financial Performance', Academy ofManagement Journal, 38: 635-672.

Hutton, A., and Stocken, P. (2007). 'Effect of Reputationon the Credibility of Management Forecasts'. WorkingPaper, Boston College and Dartmouth College.

Ichniowski, C , Shaw, K. and Prennushi, G. (1997). 'TheEffects of Human Resource Management Practices onProductivity: A Study of Steel Finishing Lines'. TheAmerican Economic Review, 87: 291-313.

International Accounting Standards Board (2004) (re-vised). IAS 19 Accounting for Retirement Benefits inEinancial Statements of Employers.

International Accounting Standards Board (2004) (re-vised). IAS 38 Intangible Assets.

International Accounting Standards Board (2004). IFRS 2Share-Based Payment.

International Accounting Standards Board (2004). IFRS 3Business Combinations.

International Accounting Standards Board (2007) (re-vised). IAS I Presentation of Einancial Statements.

Ittner, C. D. and Larcker, D. F. (1997). 'ProductDevelopment Cycle Time and OrganizationalPerformance'. Journal of Marketing Research, XXXIV:13-23.

Ittner, C. D. and Larcker, D. F. (1998). 'Are Non-FinancialMeasures Leading Indicators of Financial Performance?An Analysis of Customer Satisfaction'. Journal ofAccounting Research, 36, Supplement: 1-46.


Ittner, C. D. and Larcker, D. F. (2001). 'AssessingEmpirical Research in Managerial Accounting: A Value-Based Management Perspective'. Journal of Accountingand Economics, 32 (1-3): 349-410.

Ittner, C. D., Larcker, D. F. and Randall, T. (2003).'Performance Implications of Strategic PerformanceMeasurement in Financial Services Firms'. Accounting,Organizations and Society, 28: 715-741.

Jaeobson, R., and Mizik, N. (2007). 'The FinancialMarkets and Customer Satisfaction: Re-examining theValue Relevance of Customer Satisfaction from theEfficient Markets Perspective'. Working Paper,University of Washington and Columbia University.

Jaffe, A.B. (1986). 'Technological Opportunity andSpillovers of R&D: Evidence from Firms' Patents,Profits, and Market Value'. American Economic Review,76 (5): 984-1,001.

Jennings, R., Robinson, J., Thompson, R. and Duvall, L.(1996). 'The Relationship between Accounting GoodwillNumbers and Equity Values'. Journal of BusinessEinance and Accounting, 23 (4); 513-533.

Johnson, L. D., and Pazderka, B. (1993). 'Firm Value andInvestment in R&D'. Journal of Managerial andDecision Economics, 14 (1): 15-24.

Jorgenson, D. (1989). 'Production Functions', in The NewPalgrave - A Dictionary of Economics, J. Eatwell, M.Milgate, and P. Newman (eds.). United Kingdom: TheMacmillan Press Limited: 1002-1007.

Kallapur, S. and Kwan, S. (2004). 'The Value Relevanceand Reliability of Brand Assets Recognized by U.K.Firms'. The Accounting Review, 78 (1): 151-172.

Kaplan, R. and Norton, D. (1992). 'The BalancedScorecard: Measures that Drive Performance'. HarvardBusiness Review, 70 (I) : 71-80.

Kaplan, R. and Norton, D. (1996). The BalancedScorecard: Translating Strategy into Action. Boston, MA:Harvard Business School Press.

Kaplan, R. and Norton, D. (2001). The Strategy FocusedOrganization: How Balanced Scorecard CompaniesThrive in the New Business Erivironment. Boston, MA:Harvard Business School Press.

Kendrick, J. (1972). 'The Treatment of IntangibleResources as Capital'. Review of Income and Wealth,18(1): 109-125.

Klemperer, P. (2005). 'Competition When Consumers haveSwitching Costs: An Overview with Applications toIndustrial Organization, Macroeconomic, and InternationalTrade'. Review of Economics Studies, 62: 515-539.

Klock, M. and Megna, P. (2000). 'Measuring and ValuingIntangible Capital in the Wireless CommunicationsIndustry'. The Quarterly Review of Economics andEinance,'iO: 519-532.

Knight, F. (1921). Risk, Uncertainty, and Profit. Chicago,IL: University of Chicago Press.

Kothari, S. P., Laguerre, T. E. and Leone, A. J. (2002).'Capitalization versus Expensing: Evidence on theUncertainty of Future Earnings from CapitalExpenditures versus R&D Outlays'. Review ofAccounting Studies, 1 (4): 355-382.

Kothari, S. P., Shu, S., and Wysocki, P. (2007). 'DoManagers Withhold Bad News?'. Working Paper,Massachusetts Institute of Technology.

Kwon, S. (2001). 'Value Relevance of FinancialInformation and Conservatism: High-Tech versus Low-Tech Stocks'. Working Paper, Rutgers Business School,Camden.

Lach, S. and Rob, R. (1996). 'R&D, Investment, andIndustry Dynamics'. Journal of Economics andManagement Strategy, 5 (2): 217-249.

Lach, S. and Schankerman, M. (1989). 'Dynamics of R&Dand Investment in the Scientific Sector'. Journal ofPolitical Economy, 97 (4): 880-904.

Lajili, K., and Zéghal, D. (2006). 'Market PerformanceImpacts of Human Capital Disclosures'. Journal ofAccounting and Public Policy, 25 (2): 171-194.

Lambin, Jean J. (1976). 'Advertising, Competition andMarket Conduct in Oligopoly Over Time'. Amsterdam:North Holland Publishing Co.

Landes, W. M., and Posner, R. A. (1987). 'Trademark law:An economic perspective'. Journal of Law andEconomics, 30 (2): 265-309.

Landsman, W. R., Peasnell, K., Pope, P. and Yeh, S. (2004).'The Value Relevance of Alternative Methods ofAccounting for Employee Stock Options'. WorkingPaper, University of North Carolina at Chapel Hill,Lancaster University and National Taiwan University.

Langfield-Smith, K. (1997). 'Management ControlSystems and Strategy: A Critical Review'. Accounting,Organizations and Society, 22 (2): 207-232.

Lanjouw, J. O., Pakes, A. and Putnam, J. (1998). 'How toCount Patents and Value Intellectual Property: The Usesof Patent and Renewal and Application Data'. TheJournal of Industrial Economics, XLVI, 4: 405^32.

Lev, B. and Ohison, J. (1989). 'Market-Based EmpiricalResearch in Accounting: A Review, Interpretation, andExtension'. Journal of Accounting Research, 20,Supplement: 249-322,

Lev, B., Sarath, B. and Sougiannis, T. (2005). 'R&DReporting Biases and their Consequences'. ContemporaryAccounting Research, 22 (4): 977-1026.

Lev, B. and Schwartz, A. (1971). 'On the Use of EconomicConcepts of Human Capital in Financial Statements'. TheAccounting Review, 46 (1): 103-112.

Lev, B., and Sougiannis, T. (1996). 'The Capitalisation,Amortisation and Value Relevance of R&D'. Journal ofAccounting and Economics, 21: 107-138.

Li, F. and Wong, M. H. F. (2004). 'Employee StockOptions, Equity Valuation, and the Valuation of OptionGrants using a Warrant-Pricing Model'. Working Paper,University of Michigan at Ann Arbor and University ofToronto.

Libby, R., Nelson, M. W. and Hunton, J. E. (2006).'Recognition v. Disclosure, Auditor Tolerance forMisstatement, and the Reliability of Stock-Compensationand Lease Information'. Journal of Accounting Research,44 (3): 533-560.

Lichtenberg, F. R. (1995). 'The Output Contributions ofComputer Equipment and Personnel: A Firm LevelAnalysis'. Economic Innovations and New Technology, 3:201-217.

Lingle, J. and Schiemann, W. (1996). 'From BalancedScorecard to Strategic Gauges: Is Measurement WorthIt?'. Management Review, 85: 56-61.

Lipe, R., Bryant, L. and Widener, S. K. (1998). 'Do Non-Linearity, Firm-Specific Coefficients, and LossesRepresent Distinct Factors in the Relation Between StockReturns and Accounting Earnings?'. Journal ofAccounting and Economics, 25: 195-214.

List, J. A. (2006). 'The Behavioralist Meets the Market:Measuring Social Preferences and Reputation Effects inActual Transactions'. Journal of Political Economy, 114:1-37.

Liu, M. (2007). 'Accruals and Managerial OperatingDecisions over the Firm Life Cycle'. Working Paper,Pennsylvania State University.

Lo, K. and Lys, T. (2000). 'Bridging the Gap BetweenValue Relevance and Information Content'. WorkingPaper, University of British Columbia and Northwestern


University.Magee, S. P. (1977). 'Multinational Corporations, the

Industry Technology Cycle and Development', yowraa/ ofWorld Trade Law, 2 (4): 297-321.

Matolcsy, Z. and Wyatt, A. (2006). 'Financial Analysts andIntangible Assets'. Accounting and Finance, 46 (3):457^79.

Matolcsy, Z., and Wyatt, A. (2008). 'The AssociationBetween Technological Conditions and Market Value ofEquity'. Forthcoming in The Accounting Review.

McCarthy, M., and Schneider, D. (1996). 'Evidence fromthe US Market of the Association of Capitalized Non-Goodwill Intangibles to Firm Equity Value'. Advances inInternational Accounting, 9: 111-127.

McGahan, A., and Silverman, B. (2006). 'Profiting fromTechnological Innovation by Others: The Effect ofCompetitor Patenting on Firm Value'. Research Policy,35: 1222-1242.

Megna, P., and Klock, M. (1993). 'The Impact ofIntangible Capital on Tobin's Q in the SemiconductorIndustry'. American Economic Association Papers andProceedings, 83 (2): 265-269.

Mendonca, S., Pereira, T., and Godinho, M. (2004).'Trademarks as an Indicator of Innovation and IndustrialChmgt.\ Research Policy, 33: 1385-1404.

Metcalfe, J. S. (1998). Evolutionary Economics andCreative Destruction. London: Routledge.

Mizik, N. and Jacobson, R. (2006). 'The InformationContent of Brand Attributes'. Working Paper, ColumbiaUniversity Graduate School of Business and University ofWashington School of Business.

Müller, K. (1994). 'The Valuation Implications of theAccounting Measurement of Brand Names, PublishingTitles, and Goodwill - Evidence from the Debate in theUnited Kingdom'. Unpublished Manuscript.

Müller, K. (1999). 'An Examination of the VoluntaryRecognition of Acquired Brand Names in the UnitedKingdom'. Journal of Accounting and Economics,26: 179-191.

Nakamura, L. (2005). 'Underestimating Advertising:Innovation and Unpriced Entertainment'. Federal ReserveBank of Philadelphia Business Review, Q4, 21-29.

Narin, F. (2000). Techline Background Paper, in J. Tidd,(ed.). From Knowledge Management to StrategicCompetence. Imperial College Press. Series onTechnology Management, 3: 155-195.

Nayyar, P. R. (1995). 'Stock Market Reactions to CustomerService Changes'. Strategic Management Journal, 16:39-53.

Netter, J. (1982). 'Excessive Advertising: An EmpiricalAnalysis', yourna/ of Industrial Economics, 4: 360-373.

Noll, R. G., Peck, M. J. and McGowan, J. J. (1973).Economic Aspects of Television Regulation. BrookingsInstitution, Washington D. C.

Ohlson,J. A.(I995). 'Earnings, Book Value and Dividendsin Security Valuation'. Contemporary AccountingResearch, 11,661-668.

Pakes, A. (1985). 'On Patents, R&D, and the Stock MarketRate of Return'. Journal of Political Economy, 21:390^99.

Pakes, A. (1986). 'Patents as Options: Some Estimatesof the Value of Holding European Patent Stocks'.Econometrica, 54: 755-784.

Pavitt, K. (1984). 'Sectoral patterns of technical change'.Research Policy, 13: 343-373.

Peles, Y. (1970). 'Amortization of AdvertisingExpenditures in the Financial in the FinancialStatements'. Journal of Accounting Research, 8 (1):128-137.

Penman, S. H. (2004). Financial Statement Analysis andSecurity Valuation. New York, NY: McGraw-Hill.

Porter, M. (1980). Competitive Strategy: Techniques forAnalyzing Industries and Competitors. New York, NY:Free Press.

Porter, M. (1985). Competitive Advantage: Creating andSustaining Superior Performance. New York, NY: FreePress.

Rajgopal, S., Venkatachalam, M. and Kotha, S. (2003).,'The Value-Relevance of Network Advantages: The Caseof E-Commerce Firms', yowrna/ of Accounting Research,40(4): 135-162.

Ramanna, K. and Watts, R. L. (2007). 'Evidence on theEffects of Unverifiable Fair-Value Accounting'. HarvardBusiness School Working Paper No. 08-014.

Reekie, W.D. and Bhoyrub, P. (1981). 'Profitability andIntangible Assets: Another Look at Advertising and EntryBarriers'. Applied Economics, 13: 99-107.

Ritter, A. and Wells, P. (2006). 'Identifiable IntangibleAsset Disclosures, Stock Prices and Future Earnings'.Accounting and Finance, 46 (5): 843-863.

Roach, S. S. (1991). 'Services Under Siege: TheRestructuring Imperative'. Harvard Business Review, 69(5): 82-91.

Rosett, J. G. (1997). 'Capitalized Labor Contracts'.Working Paper, Graduate School of Business, Universityof Chicago.

Rosett, J. G. (2001). 'Equity Risk and the Labor Stocks:The Case of Union Contracts'. Journal of AccountingResearch, 39 (2): 337-364.

Rosett, J. G. (2003). 'Labour Leverage, Equity Risk andCorporate Policy Choice'. European Accounting Review,12 (4): 699-732.

Ross, J., and Georgeoff, D. (1991). 'A Survey ofProductivity and Quality Issues in Manufacturing: TheState of the Industry'. Industrial Management,January/February, 33 (1): 3-25.

Rugman, A. M. (1981). Inside the Multinationals: TheEconomics of Internal Markets. New York, NY: ColumbiaUniversity Press.

Rumelt, R. R. (1991). 'How Much Does Industry Matter?'Strategic Management Journal, 12: 167-185.

Schankerman, M. (1991). 'The Private Value of PatentRights in France, 1969-87: An Empirical Study of PatentRenewal Data'. NBER Working Paper No. 3780.

Schankerman, M. and Noel, M. D. (2006). 'StrategicPatenting and Software Innovation'. London School ofEconomics and Political Science and University ofCalifornia, San Diego.

Schankerman, M. and Pakes, A. (1986). 'Estimates of theValue of Patent Rights in European Countries Duringthe Post-1950 Period'. Economic Journal, 96 (384):1052-1077.

Scherer, F. M. (1966). 'Time-Cost Tradeoffs in UncertainEmpirical Research Projects'. Naval Research LogisticsQuarterly, 13 (2): 71-82.

Schmalensee, R. (1978). 'Entry Deterrence in the Ready-to-Eat Breakfast Cereal Industry'. Bell Journal ofEconomics, 9: 305-327.

Schumpeter, J. (1934). The Theory of EconomicDevelopment. Cambridge, MA: Harvard University Press.

Schumpeter, J. (1942). Capitalism, Socialism andDemocracy. New York, NY: Harper and Brothers.

Seethamraju, C. (2000). 'The Value-Relevance ofTrademarks'. Working Paper, New York University.

Serrano, C. J. (2005). 'The Market for Intellectual Property:Evidence from the Transfer of Patents'. Working paper.

Shah, S., Stark, A. W. and Akbar, S. (2007). 'Firm Size,Sector and market valuation of R&D Expenditures'.


Applied Einancial Economic Letters, Forthcoming.Shevlin, T. (1991). 'The Valuation of R&D Firms with

R&D Limited Partnerships', The Accounting Review, 66:1-21.

Shevlin, T. (1996). 'The Value Relevance of Non-FinancialInformation: A Discussion'. Journal of Accounting andEconomics, 22: 31-42.

Shortage, R. T. (2004). 'Market Valuation of Successfulversus Non-Successful R&D Efforts in thePharmaceutical Industry'. Journal of Business, Einanceand Accounting, 3 \ (9-10): 1301-1325.

Simons, R. (1991). 'Strategic Orientation and TopManagement Attention to Control Systems'. StrategicManagement Journal, 12: 49-62.

Skinner, D. (1993). 'The Investment Opportunity Set andAccounting Procedure Choice'. Journal of Accountingand Economics, 16: 407—445.

Smith, W. Jr., and Watts, R. (1992). 'The InvestmentOpportunity Set and Corporate Financing, Dividend, andCompensation Policies'. Journal of Einancial Economics,32: 263-292.

Sougiannis, T. (1994). 'The Accounting Based Valuation ofCorporate R&D'. The Accounting Review, 1: 44-68.

Stark, A. W. and Thomas, H. M. (1998). 'On the EmpiricalRelationship between Market Value and Residual Incomein the U.K.'. Management Accounting Research, 9:445-460.

Stephan, P. E. (1996). 'The Economics of Science', Journalof Economic Literature, XXXIV, September, 1119-1235.

Stephan, P. E., Higgins, M. J., and Thursby, J. G. (2007).'Changing Regimes in Biotechnology: Nobel Affiliatesand Valuation'. Working Paper, Georgia State University,Georgia Institute of Technology, and Emory University.

Stem, J., Stewart, G. and Chew, D. (1995). 'The EVAFinancial Management System'. Journal of AppliedCorporate Einance, 8: 32-46.

Stewart, G. (1991). The Quest for Value, New York, HarperBusiness.

Stigler, G. (1958). 'The Economics of Scale'. Journal ofLaw and Economics, 1: 54—71.

Stiglitz, J. E. (1996). 'Plenary Remarks on IntangibleAssets'. SEC Symposium on Financial Reporting andIntangible Assets, April 11-12, U. S. Securities andExchange Commission, Washington D. C.

Stoneman, P. and Toivanen, O. (1997). 'Innovation andMarket Value in the United Kingdom: Some PreliminaryResults'. Working Paper, Warwick Business School.

Subramanyam, K. R. (1996). 'Uncertain Precision andPrice Reactions to Information'. The Accounting Review,71:207-220.

Thompson, R. W., Beatty, R. P. and Riffe, S. (2001). 'Log-Linear Stock Valuation Based on AccountingInformation'. Working Paper, Southern MethodistUniversity and University of Southern California.

Trajtenberg, M. (1990). 'A Penny for your Quotes: PatentCitations and the Value of Innovations'. RAND Journal ofEconomics, 21 (1): 172-187.

Vakratsas, D., and Ambler, T. (1999). 'How AdvertingWorks: What Do We Really Know?'. Journal ofMarketing, 63: 26—43.

Vincent, L. (1994). 'The Equity Valuation Implications ofAccounting Acquisition Premiums'. UnpublishedManuscript.

Webster, E. (1998). 'Labour as an Asset: Can it Partitionthe Labour Market?'. Working Paper, Melbourne Instituteof Applied Economic and Social Research, University ofMelbourne.

Webster, E. (1999). The Economics of IntangibleInvestment. UK: Edward Elgar Publishing Limited.

Wernerfelt, B. (1984). 'A Resource-Based View of theFirm'. Strategic Management Journal, 5: 171-180.

Whisenant, S. (1998). 'Does Fundamental AnalysisProduce More Value-Relevant Summary Measures?'.Working Paper, University of Houston and MassachusettsInstitute of Technology (MIT).

Wilson, D. (1993). 'Assessing the Impact of InformationTechnology on Organizational Performance' in R. Banker,R. Kauffman, and M. A. Mahmood (eds.). StrategicInformation Technology Management. Hamsberg, PA:Idea Group.

Wyatt, A. (2005). 'Accounting Recognition of IntangibleAssets: Theory and Evidence on EconomicDeterminants'. The Accounting Review, 80 (3):967-1003.

Zucker, L. G., Darby, M. R. and Brewer, M. B. (1998).'Intellectual Human Capital and the Birth of U.S.Biotechnology Enterprises'. The American EconomicReview, is (1): 290-306.


App

endí

60

"3B

BC9

ifie

itb

sist

ud

ies

'S

Tab

le 1

Su

mm

ary

1

avan

ce

;3

Inta

ngib

les

•rfo

rman

cea.

eta

s53Ktí

g

:k r

etui

gCO

-Sä

pric

eoe

k

CO

trie

1

inta

ngib

le t

Tota

lsN

on-

V

;^

O

•£f

Tota

ls

Í

IA.<

!..O

0ÓCO

Tota

lsN

on-

O

.00Co

Sig.

O.A

.

•2

Sig.

O.A

.

1 R

&D

and

- H

O

o

^

- H

m

- H

(N

(N

(N

o

-

( N

in

asse

ts

•S

Mgt

. re

porl

es

o

o

(N

o

O

es

-

o

o

o

.-(

;set

s

cd

tim

ated

<u

Res

earc

her

en

- H

O

(N

es

o

o

o

o

o

o

- H

o

o

o

o

o

o

o

o

o

o

eN

o

- H

es

es

entis

i

'o

outl

ay(e

.g.

# s

Q

Ann

ual

R<& u

Inpu

t m

etri

es

o

o

(N

es

o

- H

O

es

aten

t

a.

s (e

.g.#

rie

Out

put

met

cita

tion

s)

90

O

VO

- H

Vi

ve

90

o

90

Su

bto

tal

B

idve

rtis

i

eg

Bra

nd

s an

-

o

o

o

o

o

o

o

o

o

asse

ts

pa:

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

sets

tim

ated

isa

Mgt

. re

port

Res

earc

her

en

O

o

o

e .

O

o

o

o

o

o

o

o

o

o

outla

yA

nnua

l ad

\

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

Inpu

t m

etri

o

o

o

o

o

-

o

o

o

-

es

o

es

<N

CN

o3

s (e

.g.

cri

e1

Out

put

met

rank

ings

)

o

o

O

o

o

o

Su

bto

tal

alty

S'

Cu

stom

er

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

asse

ts

pa:

Mgt

. re

port

-

o

o

o

-

o

o

o

o

o

o

o

o

o

o

sets

en

tim

ated

C/3

U

Res

earc

her

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

lay

Ann

ual

out

- H

O

o

o

- H

o

o

o

o

o

o

o

o

o

o

Inpu

t m

etri

es

o

o

o

es

es

O

o

O

o

C/5

rie;

Out

put

met

o

o

o

o

o

o

o

Su

bto

tal


a

'con

tin.

wecs

nt f

mdi

ifi

ca

•g.£f

wit

h s

studie

sel m

ary

of

1 I

scS

relé

Valu

tjK

rfor

m

Ci.

[gs:tni.

Co

c

k re

tuSt

oc,

"S

lev

ice

tock

pr

'-n

trie

igib

lesl

igib

le m

e.

as

Tota

UO

/sl

;>-So

•SP

btal

s

(-<

§

'A.<

O

•S?

: |

V

' '

o

sig.

set

ä

ass(

' Í

eu

adv

anta

ipet

itiv

e 1

oU

—

O

O

- H

o

o

o

o

o

o

o

o

—,

asse

ts

o

o

o

o

o

o

o

o

o

o

o

o

o

asse

tstim

ated

ire

port

edar

eher

es

ob <uOí

O

o

o

o

o

o

o

o

o

t N

O

13ce<

o

o

o

o

o

o

o

o

es

ustr

y)(e

.g.

ind

t met

ries

3a .c

O

o

^^

o

o

o

o

'S"S3C/J

lark

et

faobó

o'C

3fi"3O

©

©

©

©

©

,^

ota

l

.s

o

o

o

o

o

o

o

°

o

o

o

o

idin

gal

(in

clu

qual

ity]

asse

ts

lan

capit

agem

ent

repo

rted

3 « oí)HH es ^MH B i¿

O

O

<N

O

O

O

—

<N

O

^^

CN

asse

tsti

mat

ed ;

areh

er e

s

u

o

o

o

o

o

o

—

o

íal

outl

ay

g

""

o

o

o

o

o

o

o

o

t met

ries

g.

o

o

o

CM

O

o

(N

- H

O

_

ut m

etri

e

3

ve

©

©

©

©

• > »

©

•o

V )

Ota

l

. f i

o

o

o

t N

O

o

es

0 0

o

vo

00

asse

tsIw

ill

repo

rted

o -gj

O S

o

o

o

o

°

o

o

o

o

o

o

o

o

o

isse

tstim

ated

;ar

eher

esi

OS

-^

o

o

-

o

o

o

o

o

o

o

o

o

, in

put

aie

sla

l ou

tlay

tput

met

ri

c o

©

©

©

©

0 0

©

0 0

ota

l

, f i

©

so

0 0

©

ve

r m

etri

es

o

3

U

eVica

herw

ise

zero

ot

• o

=«

o

¡u

regí

the

B

inta

ngil

elud

ed

.c

ient

s on

;ffie

ou

iñea

nS

ign

rary

.s

asse

ts a

ther

o

eof

1V

alu

r as

sets

Co Ü

.a ob£O

1/3

1Cd

iblf

0 0

inta

nth

er

o

íeni

coef

fie

u

ntl

C3. S

J

sets

c/:cS

angi

ble

S

ieie

nt f

o

8u"o

valu

eT

he<

O.I

.A.

asse

t:th

er

o

a

íeni

coef

fie

cu

ntl

•It/3(U

sets

c / ]

angi

ble

E

ieie

nt f

o

seuoo

"o

valu

eT

he<

0.A

.

u•c

3

sEbl

e

'5bB

he i

nta

o

ffie

ient

seo

e:

CS

'B00

No

sS

lg.

BOZ

what financial and non-fínancial information on intangibles is...

Documents