Related Diversification, Core Competences and Corporate Performance.pdf

Strategic Management Journal, Vol. 15, 149-165 (1994)

RELATED DIVERSIFICATION, CORE COMPETENCESAND CORPORATE PERFORMANCECONSTANTINOS C. MARKIDESLondon Business School, London, U.K.

PETER J. WILLIAMSONLondon Business School, London, U.K.

Despite nearly 30 years of academic research on the benefits of related diversification, thereis still considerable disagreement about precisely how and when diversification can be usedto build long-run competitive advantage. In this paper we argue that the disagreement existsfor two main reasons: (a) the traditional way of measuring relatedness between twobusinesses is incomplete because it ignores the 'strategic importance' and similarity of theunderlying assets residing in these businesses, and (b) the way researchers have traditionallythought of relatedness is limited, primarily because it has tended to equate the benefits ofrelatedness with the static exploitation of" economies of scope (asset amortization), thusignoring the main contribution of related diversification to long-run, competitive advantage;namely the potential for the firm to expand its stock of strategic assets and create new onesmore rapidly and at lower cost than rivals who are not diversified across related businesses.An empirical test supports our view that 'strategic' relatedness is superior to marketrelatedness in predicting when related diversifiers outperform unrelated ones.

A fundamental part of any firm's corporatestrategy is its choice of what portfolio ofbusinesses to compete in. According to theacademic literature, this decision should reflectthe 'superiority' of related diversification overunrelated diversification (e.g., Ansoff, 1965;Bettis, 1981; Lecraw, 1984; Palepu, 1985; Rumelt,1974; Singh and Montgomery, 1987). This isbecause related diversification presumably allowsthe corporate center to exploit the interrelation-ships that exist among its different businesses(SBUs) and so achieve cost and/or differentiationcompetitive advantages over its rivals. But despite30 years of research on the benefits of relateddiversification, there is still considerable disagree-ment about precisely how and when diversifi-cation can be used to build long-run competitiveadvantage (e.g., Hoskisson and Hitt, 1990;

Ramanujam and Varadarajan, 1989; Reed andLuffman, 1986). In this paper we argue thisdisagreement exists for two main reasons:

1. Traditional measures of relatedness providean incomplete and potentially exaggeratedpicture of the scope for a corporation toexploit interrelationships between its SBUs.This is because traditional measures look atrelatedness only at the industry or marketlevel. But as we explain below, the relatednessthat really matters is that between 'strategicassets' (i.e., those that cannot be accessedquickly and cheaply by nondiversified competi-tors.' Therefore, to accurately measure whether

Key words: Related diversification, core com-petences, corporate performance

' It is important here to clarify the difference between'strategic assets' and "core competences." Strategic assets areassets that underpin a firm's cost or differentiation advantagein a particular market and that are imperfectly imitable,imperfectly substitutable and imperfectly tradeable. Theseassets also tend to be market-specific. An example wouldbe Honda's dealer network distributing and servicing its

CCC 0143-2095/94/090149-17© 1994 by John Wiley & Sons, Ltd.

150 C. C. Markides and P. J. Williamson

two businesses are related, we need to gobeyond broad definitions of relatedness thatfocus on market similarity; we need to look atthe similarities between the underlying strategicassets of the various businesses that a companyis operating in (see also Hill, 1994).

2. The way researchers have traditionally thoughtof relatedness is limited. This is because ithas tended to equate the benefits of relatednesswith the static exploitation of economies ofscope. While we would not deny that econo-mies of scope are an important short-termbenefit of related diversification, we believethe real leverage comes from exploitingrelatedness to create and accumulate newstrategic assets more quickly and cheaply thancompetitors (rather than simply amortizingexisting assets—i.e., reaping economies ofscope). To predict how much a strategyof related diversification will contribute tosuperior, long-run returns it is necessary todistinguish between four types of potentialadvantages of related diversification.

a. the potential to reap economies of scopeacross SBUs that can share the samestrategic asset (such as a common distri-bution system);

b. the potential to use a core competenceamassed in the course of building ormaintaining an existing strategic asset inone SBU to help improve the quality ofan existing strategic asset in another ofthe corporation's SBUs (for example, whatHonda learns as it gains more experiencemanaging its existing dealer network forsmall cars may help it improve themanagement of its largely separate net-work for motorbikes);

c. the potential to utilize a core competencedeveloped through the experience of build-

motorbikes. On the other hand, core competences are thepool of experience, knowledge and systems, etc. that existelsewhere in the same corporation and can be deployed toreduce the cost or time required either to create a new,strategic asset or expand the stock of an existing one. ThusHonda's experience in building competitive dealer networksfor a particular class of consumer durables would be anexample of a core competence. Each of these networks (onefor motorbikes and another for lawn mowers, for example)would be a separate strategic asset: 'different trees, sharingthe same (core competence) root stock.'

ing Strategic assets in existing businesses,to create a new strategic asset in a newbusiness faster, or at lower cost (such asusing the experience of building motorbikedistribution to build a new, parallel distri-bution system for lawn mowers—whichare generally sold through a different typeof outlet);

d. the potential for the process of relateddiversification to expand a corporation'sexisting pool of core competences because,as it builds strategic assets in a newbusiness, it will learn new skills. These,in turn, that will allow it to improve thequality of its stocks of strategic assets inits existing businesses (in the course ofbuilding a new distribution system forlawn mowers, Honda may learn new skillsthat allow it to improve its existingdistribution system for motorbikes).

We term these four potential advantages ofrelated diversification 'asset amortization,' 'assetimprovement,' 'asset creation' and 'asset fission'respectively.

We will argue that the long-run value of arelated diversification lies not so much in theexploitation of economies of scope (assetamortization)—where the benefit is primarilyshort-term—but in allowing corporations to morecost efficiently expand their stocks of strategicassets. Relatedness, which opens the way forasset improvement, asset creation and assetfission, holds the key to the long-run competitiveadvantages of diversification.

This means that in most cases, similarities inthe processes by which strategic assets areexpanded and new strategic assets are createdare more important than static similaritiesbetween the strategic assets that are the outcomeof those processes. Firms that are diversifiedacross a set of 'related markets' where thestrategic assets are either few, or the processesrequired to improve and create them are context-specific cannot be expected to out-performunrelated diversifiers.

THE MEASURE OF RELATEDNESS

The strategy of related diversification is con-sidered superior to unrelated diversification

Related Diversification and Corporate Performance 151

because it allows the firm to exploit interrelation-ships among its different business units. Specifi-cally, the corporate center in related diversifiersis expected to identify important assets residingin any one of its SBUs and then transfer theseassets and utilize them in another SBU. Canon'sdeployment of technology from its camera SBUin developing its photocopier business is a goodexample.^

Even though the advantages of the strategy ofrelated diversification are usually cast in termsof the cost of differentiation benefits that arisefrom the cross-utilization of the firm's underlyingassets, the actual measurement of relatednessbetween two businesses often does not evenconsider the underlying assets residing in thesebusinesses. Relatedness has been traditionallymeasured in two basic ways (e.g., Montgomery,1982; Pitts and Hopkins, 1982): (i) using anobjective index like the entropy index of SICcount (e.g.. Caves et al., 1980; Jacquemin andBerry, 1979, Palepu, 1985) which assumes thatif two businesses share the same SIC they musthave common input requirements and similarproduction/technology functions; and/or (ii) usinga more subjective measure such as Rumelt's(1974) diversification categories which considerbusinesses as related '... when a common skill,resource, market, or purpose applies to each.'(Rumelt, 1974: 29).

We do not doubt that the traditional measurescould be acceptable proxies for what they aretrying to measure. In fact, if these measuresdid not suffer from any systematic bias, onewould consider them as a 'good enough' wayto substitute for a costly and time consumingideal measure. However, they do suffer fromone systematic bias. Consider a firm using thestrategy of related diversification so as toexploit the relatedness of its SBU-level assets.Suppose, however, that the SBU-level assetsthat the corporate center is trying to exploitare not 'strategically important' (as definedbelow). For example, suppose that the assetservices that Firm X provides to an SBU bycross-utilizing the assets of a sister subsidiary

^ An extension of this argument has been proposed by Hill(1988): the corporation will be in a better position to exploitthe interrelationships among its businesses if it is structuredappropriately. Hill finds that related diversifiers are betterserved by the CM-form organizational structure than the M-form structure.

are such that any other firm can easily purchaseon the open market at close to marginal cost.In that case, even if Firm X achieves short-termcompetitive advantage through exploitation ofeconomies of scope, it will not really achieveany sustainable competitive advantage overtime; other firms will quickly achieve similarpositions by purchasing similar asset services.The opportunity for a diversified firm toamortize the costs of running a trucking fleetby sharing it across two SBUs is often a casein point. If nondiversified firms could buysimilar trucking services from a common carrier(which itself achieves the economies of scopeacross customers) at close to marginal cost,then there would be no competitive advantageto diversification even though the two marketswere closely 'related' according to traditionalmeasures like SIC similarity.

This implies that any measure of relatednessshould take into consideration not only whetherthe underlying SBU-level assets of a firm arerelated, but also consider whether these assetsare a potential source of competitive advantage.Even if the traditional measures of relatednessdo a good job in capturing the relatedness ofthe underlying assets, they consistently ignorethe evaluation of whether these assets are'strategic' assets; and they do so because inmeasuring relatedness, they do not explicitlyconsider the underlying assets.

Strategic assets

To win competitive advantage in any market,a firm needs to be able to deliver a given setof customer benefits at lower costs thancompetitors, or provide customers with a bundleof benefits its rivals cannot match (Porter,1980). It can do so by harnessing the driversof cost and differentiation in its specific industry.For example, if scale is an important driver ofcost leadership then those firms that operatelarge-scale plants will outperform their subscalecompetitors. However, to effectively exploitthese cost and differentiation drivers, the firmneeds to access and utilize a complex set oftangible and intangible assets. For example,to reap the benefits of scale economies inproduction, it may require the services oftangible assets like a large-scale plant andintangible assets like the skills to manage this


scale facility effectively and distributor loyaltyto support a constant high volume of sales.^

Given that a particular set of asset stocks isnecessary to allow a firm to exploit cost anddifferentiation advantages, the crucial questionfor a firm is: 'How can I access these assets?'A firm can secure these required asset servicesin a number of ways. It may obtain them withthe endowment which establishes the business.A company established to exploit a proprietarytechnology, for example, often receives avaluable patent asset from its founder. It mayacquire the assets on the open market, orcontract directly for the services of an asset (asin the case of an equipment lease). It mightaccess the required asset services by sharingthe asset with a sister SBU or an alliancepartner. Finally, it may accumulate the requiredasset through a process of combining tradeableinputs with existing asset stocks and learningby doing (Dierickx and Cool, 1989).

Firms that possess assets which underpincompetitive advantage will earn rents (Rumelt,1987). To the extent that competitors canidentify these rent producing assets, theycan decide between two alternative ways inreplicating this competitive advantage: theymay seek to imitate the assets through one ofthe four mechanisms above, or they may try tosubstitute them with other assets which canearn similar rents by producing equivalent orsuperior customer benefits. The assets onwhich long-term competitive advantage criticallydepends (strategic assets) are, therefore, thosethat are imperfectly imitable and imperfectlysubstitutable (Barney, 1986; Dierickx and Cool,1989).

The importance of asset accumulationprocesses

The conditions above imply that assets whichare readily tradeable cannot act as sources oflong-term competitive advantage (Williamson,1975). Similarly, assets which can be quicklyand/or cheaply accessed through endowment,acquisition or sharing can only provide competi-

•* See Verdin and Williamson (1994) for a fuller discussionof the link between Porter's cost and differentiation driversand the assets on which exploitation of these driversdepend.

tive advantage which is short-lived. In the longrun, internal accumulation is likely to be themost significant source of imperfectly imitableand imperfectly substitutable assets. This isbecause most assets will be subject to erosionover time (see e.g., Eaton and Lipsey, 1980).Customer assets like brands, for example, willdecay as new customers enter the market orformer customers forget past experience or exitthe market. The value of a stock of technicalknow-how will tend to erode in the face ofinnovation by competitors. Patents will expire.Thus, assets accessed through initial endowmentor an initial asset base shared with anotherSBU will tend to lose their potency as sourcesof competitive advantage over time unlessthey are replenished by internal accumulationprocesses.

Moreover, even when an asset can be accessedthrough acquisition, alliance, or sharing, it isquite likely that the existing assets availablewill not perfectly fit the requirements of themarket they will be used to serve. Existingassets generally need some adaptation to aspecific market context and integration withexisting asset bundles. Internal asset accumu-lation processes therefore play a role in moldingassets which an SBU accesses externally into acompetitive, market-specific bundle.

Regardless of whether the initial stock ofstrategic assets within an SBU is obtained byendowment or acquisition, or accessed throughsharing, therefore, the long-term competitiveadvantage of a firm will largely depend on itsability to continuously adapt and improve itsstrategic assets to meet market-specific demandsand to create new strategic assets that it canexploit in existing or new markets.

If these asset accumulation processes werefrictionless and firms could speed them up atlittle cost, then it would be difficult for a firmthat gained an initial advantage in respect of aset of assets (e.g., through endowment, sharingor first mover experience in a new, growingsegment of the market) to maintain this lead. Inpractice, however, there are many impedimentswhich prevent laggards from replicating orsurpassing the asset positions of the leaders.Dierickx and Cool (1989) identify four separatecategories of these impediments to assetaccumulation: time compression diseconomies,asset mass efficiencies, asset interconnectedness


and causal ambiguity.'* These impediments alsolie behind the concept of barriers to mobility(Caves and Porter, 1977) and Rumelt's 'isolatingmechanisms' which include property rights onscarce resources, lags, information asymmetriesand other sources of friction in processes ofasset imitation (Rumelt, 1987).

When the process necesary to accumulatean asset suffers from one or more of theseimpediments, all firms will face higher costs andtime delays in building it. This will restrict theirability to satisfy their market by offering thedifferentiation or cost advantages that the elusiveasset would underpin. Impediments like timecompression diseconomies, asset mass efficienciesand asset interconnectedness, however, willimpose higher costs on later entrants to abusiness, making it more difficult for them tocatch up with first movers and established firmswho have had longer to accumulate nontradeableassets. Diversifiers entering a market for the firsttime against established firms would thereforesuffer a handicap from late arrival, other thingsbeing equal.

It may be, however, that by deploying itsexisting core competences a diversifier canovercome some of these frictions. By drawingon its existing competence pool, such a corpor-ation may be able to imitate valuable, nontrade-able assets, or accumulate new, substitute ones,or create entirely new strategic assets morecheaply and quickly than competitors who lackedaccess to similar core competences: to grow newtrees more rapidly and more cheaply by drawingon a common, existing root stock. Likewise, byproperly deploying core competences betweenbusiness units, a diversified corporation may also

•* Time compression diseconomies are the extra cost associatedwith accumulating the required assets under time pressure(the cost of compressing an activity in time). For example,it may take more than twice the amount of marketing toachieve in 1 year the same level of brand awareness as anestablished competitor may have been able to develop overa period of 2 years (other things equal). Asset mass efficienciesrefer to the fact that some types of assets are more costlyto accumulate when the firm's existing stock of that asset issmall. It is more difficult, for example, to build the customerbase of a credit card when it has few existing users.Asset interconnectedness refers to the fact that a lackof complementary assets can often impede a firm fromaccumulating an asset which it needs to successfully serve itsmarket. Causal ambiguity refers to the impediment associatedwith the uncertainty of pinpointing which specific factors orprocesses are required to accumulate a required asset (theprecise chain of causality is ambiguous).

be able to maintain or extend its competitiveadvantage in its existing businesses through itsability to augment its nontradeable, market-specific assets more quickly and cheaply than itscompetitors. This is especially important inmarket environments that are undergoing signifi-cant change. Even firms with massive asset baseswill lose their competitive advantage if they areunable to develop the new, strategic assetsnecessary to serve a changing market.

Core competences as catalysts in the 'productionfunction' of strategic assets

If strategic assets are the imperfectly imitable,imperfectly substitutable and imperfectly trade-able assets necessary to underpin an SBU's costor differentiation advantage in a particularmarket, then core competences can be viewedas the pool of experience, knowledge and systems,etc. that exist elsewhere in the same corporationwhich can be deployed to reduce the cost ortime required either to create a new, strategicasset or expand the stock of an existing one.Competences are potential catalysts to the processof accumulating strategic assets. If the firm knowsfrom past experience how to efficiently build thetype of distribution network which will improvethe competitiveness of its product (i.e., the'competence' in building a suitable type ofdistribution network exists), then it will be ableto put the necessary asset in place more quicklyand cheaply than a firm which lacks thiscompetence. Competences may also act ascatalysts to the processes of adapting andintegrating assets that an SBU has accessedthrough acquisition, alliances or sharing. Prahaladand Hamel (1990), for example, cite the case ofNEC's competency in managing collaborativearrangements as an important factor in theirability to access and then internalize technologicalassets and skills from their alliance partners.

This catalytic role of competences in the'production function' for building assets whichare nontradeable, nonsubstitutabie and difficultto accumulate is illustrated in Figure 1. Inputsinclude time, readily tradeable assets, existingnontradeable assets and the catalyst to theconstruction process: competences.

The obvious next question is: where can afirm get hold of the competences that wouldallow it to speed up its rates of asset accumulation.


Non-Tradeable,SBU Assets

CoreCompetences

(Catalysts)

ISBU Asset

Accumulation Processes• time compression disecotiomies

• asset inter-relatedness

• asset mass efficiencies

• causal ambiguity

SBU's ExistingNon-Tradeable

Assets

Cash

Time/Experience

TradeableAssets/Inputs

Figure 1. Core competences and the 'production function' for assets

adaptation and integration? The first place tolook is the open market. But competencesthemselves often have characteristics which ren-der markets inefficient as a mechanism forexchange. Characteristics such as informationimpactedness and scope for opportunism makecompetences, like other intangible assets, difficultto sell at arms-length (Williamson, 1973; Caves,1982, Ch.l). This leads to excess capacity incompetences which cannot be easily utilized byseeking buyers in the open market. Uniquecompetences developed by an SBU throughlearning by doing therefore risk becoming 'impris-oned' in that unit, even though they could bepotentially valuable catalysts to the process ofasset accumulation in other businesses (Prahaladand Hamel, 1990).

Compared with the problems associated withtrading competences in the open market, it isoften more efficient to transfer competencesbetween businesses using conduits internal to asingle organization (Williamson, 1975). Suchinternal mechanisms include posting staff fromone business unit to another, bringing togethera corporate task force with individuals from anumber of businesses to help solve a problemfor one of them, and passing market intelligenceor other information between SBUs which couldact as catalysts to asset accumulation.

Not all of the competences of a corporationwhich can act as catalysts in expanding the assetbase of a new or existing SBU, however, willmake an equal contribution to improving thecompetitive advantage of an SBU. Honda'scompetence in building networks of dealers forconsumer durables may speed up the rate andimprove the cost at which it can build aneffective, specialized distribution network for itsnew lawn mower product. But if a competitorcould effectively substitute this by a distributionagreement with one or two national retail chains,the Honda Corporation's competence may affordits lawn-mower SBU little or no competitiveedge. Likewise, if a rival could acquire a suitablenetwork at a competitive price, or obtain accessto one through a strategic alliance, access toHonda Corporation's competence might provideits related SBU with little or no competitiveadvantage. In both of these cases, while thecompetence is both available and transferable, itdoes not lead to the creation of a strategic assetthat is both hard to substitute and difficult toimitate.

By contrast, Honda's competence in smallpetrol engines may enable its lawn mower SBUto quickly and cost effectively bring a superiorproduct to market, backed by a superior pro-duction process. If competitors had no way of


matching the resulting buyer benefits, except byspending a great deal of money over a long periodof time, Honda's engine design organization andthe combination of its manufacturing hardwareand software would represent extremely potentstrategic assets for the lawn mower SBU oncethey were in place. So access to Honda's enginecompetence would be a very significant sourceof competitive advantage for its lawn mowerSBU.

We therefore have two conditions which mustbe satisfied for internal transfer of competencesbetween SBUs to create advantage for thecorporation:

1. it must be more efficient to transfer thecompetence internally between businesses inthe same group than via an external market;

2. the competence must be capable of acting asa catalyst to the creation of market-specificassets which are nontradeable, nonsubstitut-abie and slow or costly to accumulate, therebyacting as a source of competitive advantagefor the recipient SBU.

The larger the efficiency advantage of internaltransfer, and the more costly the resulting assetis to accumulate, the greater the advantage tobe gained from shifting a competence from onebusiness unit to another existing or new SBU.̂

A DYNAMIC VIEW OF RELATEDNESS

So far we have established that an SBU'scompetitive advantage depends importantly onits access to strategic assets. We have alsodiscussed how core competences can be used ascatalysts in the processes of expanding an SBU'sstock of strategic assets. The real, long-runbenefits of relatedness should therefore lie inopening up opportunities to quickly and cheaplycreate and accumulate these strategic assets. Itis then possible to distinguish five different typesof relatedness.* These distinctions help pinpointexactly when and how related diversification will

' The role of organizational structure in allowing a firm toexploit the benefits of related diversification is explored inmore detail in Markides and Williamson (1993).^ We would like to thank Gary Hamel for his contributionin the formulation of these ideas.

lead to competitive advantage for a corporation(and when it will not).

The first category, we term 'exaggeratedrelatedness.' This is where the markets servedby two SBUs share many similarities, but thereis little potential to exploit these similaritiesfor competitive advantage. The relatedness is'exaggerated' in the sense that looking at theoverall similarity (which traditional measures ofmarket-relatedness tend to do, as we explainbelow), overstates the likelihood that a corpor-ation will achieve superior performance bydiversifying across both markets. This exagger-ation may arise under any of a number ofdifferent conditions. It may be that while thediversified firm can quickly and cheaply buildthe asset stocks necessary to supply the market,so can any other firm, because most of these assetsare easily imitable. Even if other, nondiversifiedfirms cannot replicate the assets built by thediversifier, they may be able to substitute someother, readily available asset at no disadvantageto their competitiveness. In short, the assets thatthat relatedness helps a diversifier build may benon-strategic. A manufacturer of fashion knitwearin Europe or North America, for example, mayhave the competence to bring a local productionfacility for knitting standard, men's socks on-stream quickly and efficiently. But such afacility may prove a nonstrategic asset againstcompetitors who rely on off-shore sourcingfor this type of nontime sensitive, nonfashionproduct. This type of relatedness, therefore,would not create an opportunity for profitablediversification.

Similarly, exaggerated relatedness may arisewhen the market-specificity of the strategic assetsand the competences that can help build them,are underestimated by the indicators a diversifierchooses to consider. Diversification by Levisfrom jeans into men's suits, for example, wasrecognized as a failure. The two businesses mayappear highly related on many dimensions fromproduction through to marketing and distribution,but the strategic assets and competences requiredto build competitive advantage turned out to bevery different.

The second type of relatedness arises wherethe strategic assets in one SBU can be sharedwith another to achieve economies of scope(e.g.. Porter, 1987; Teece, 1982). This type ofrelatedness underpins what we term 'amortization


advantage,' by allowing related diversifiers toamortize the cost of an existing asset by using itto serve multiple markets. This type of relatednesscan offer important, short-term advantages inthe form of reduced costs and improved differen-tiation. But, for most corporations, diversificationis a long-term step that could be costly to reverse.And simply exploiting its existing stock of assets(even if they are the 'right' assets) cannot beenough to create long-term competitive advan-tages (e.g., Prahalad and Hamel, 1990). Thetruly successful firms over the long term will bethe ones that continuously create new strategicassets.

The third category of relatedness is wherethe strategic asset itself cannot be shared ortransferred between two SBUs (because it ismarket-specific), but the competence gained inthe process of building or maintaining an existingstrategic asset in one SBU can be used as acatalyst to help improve the quality of an existingstrategic asset in another SBU. This role ofcompetences in asset accumulation is illustratedwith the example of Canon's camera, photocopierand laser printer divisions in Figure 2.

Consider the position at Canon at the pointwhere the company has successfully establisheditself in both the camera and photocopierbusinesses. Many of the strategic assets whichunderpin these respective SBUs cannot be shareddirectly. The dealer networks and componentmanufacturing plans are largely specific to eachSBU. But in the course of its operationsproducing and marketing cameras, the camera

division has extended this initial asset stock bya mix of learning-by-doing and further purchasesof assets in the market. As a by-product of thisasset accumulation process, the camera businessalso developed a series of competences likeknowledge of how to increase the effectivenessof a dealer network, how to develop new productscombining optics and electronics and how tosqueeze better productivity out of high-volumeassembly lines.

Because Canon is in two businesses, camerasand photocopiers, where the processes of improv-ing dealer effectiveness, speeding up productdevelopment or improving assembly-line pro-ductivity are similar, it can improve the qualityof the strategic assets in its photocopier business,by transferring competences learned in its camerabusiness and vice versa. This type of relatedness,similarities in the processes required to improvethe effectiveness and efficiency of separate,market-specific stocks of strategic assets in twobusinesses, opens up opportunities for what wecall 'asset improvement' advantages for relateddiversifiers.

The fourth type of relatedness emerges wherethere is potential to utilize a core competencedeveloped through the experience of buildingstrategic assets in existing businesses, to createa new strategic asset in a new business faster, orat lower cost. For example, in the course ofoperating in the photocopier market, and buildingthe asset base required to out-compete rivals,this SBU also accumulated its own, additionalcompetences that the camera SBU had not

Competences <—

Asset Accumulation:Cameras

MarketExperience

-| Asset Stock [•t1

EndowmentAcquisition

Sharing

— • Competences

t iAsset Accumulation:

Photocopiers

[Market .

Experience

-| Asset Stock |-t1


Sharing

— ^ Competences

t iAsset Accumulation:

Laser Printers

MarketExperience

•j Asset Stockt1


Sharing

Figure 2. Core competences and asset accumulation at Canon


developed. These may have included how tobuild a marketing organization targeted to busi-ness, rather than personal buyers, and how todevelop and manufacture a reliable electrostaticprinting engine.

When Canon diversified into laser printers,this new SBU started out with an endowment ofassets, additional assets acquired in the marketand arrangement to share facilities and corecomponents. But even more important for itslong-term competitiveness, the new laser printerSBU was able to draw on the competences builtup by its sister businesses in cameras andphotocopiers to create new, market-specific stra-tegic assets faster and more efficiently than itscompetitors (illustrated by the arrows pointingto the right in Figure 2). This kind of relatedness,where the competences amassed by existingSBUs can be deployed to speed up and reducethe cost of creating new market-specific strategicassets for the use of a new SBU, we term the'asset creation' advantage of related diversifiers.Again, only where the processes required tobuild the particular strategic assets needed bythe new SBU are 'related' in the sense that theycan benefit from existing core competences, willthis type of diversification advantage be available.

The fifth, and final, type of relatedness iswhere in the process of creating the new strategicassets required to support diversification into anew business (like laser printers), the corporationlearns new competences that can then be usedto enhance its existing SBUs. For example, increating the assets required to support the design,manufacture and service of the more sophisticatedelectronics demanded by the laser printer busi-ness. Canon may have developed new com-petences that could be used to improve itsphotocopier business. Alternatively, by combin-ing the competences developed in its photocopierand laser printer businesses, may have helped itto quickly and cheaply build the strategic assetsrequired to succeed in a fourth market: that forplain paper facsimiles. This kind of advantageover single-business firms or unrelated diversifi-ers, we term 'asset fission' advantage.

It is these last three types of relatedness thatare likely to offer the greatest advantages fromrelated diversification over the long-run. As thelabel suggests, exaggerated relatedness offerslittle or no scope for a strategy of relateddiversification to deliver superior performance.

despite what may be a high degree of similaritybetween two markets. Related diversificationdesigned to reap economies of scope, helping toamortize existing assets, is likely to provide onlyephemeral advantage. Only relatedness thatallows a corporation to access asset improvement,asset creation and asset fission promises long-run competitive advantage to related diversifiers.The problem is that traditional measures ofrelatedness have not been designed to distinguishbetween these profitable and unprofitable typesof diversification.

Extending traditional measures of marketrelatedness: Towards strategic reiatedness

As we have seen above, it is not broad market-relatedness that matters. Two markets may beclosely related, but if the opportunity to rapidlybuild assets using competences from elsewherein the corporation does no more than generateasset stocks which others can buy or contract inat similar cost, then no competitive advantagewill ensue from a strategy of diversificationbetween them. 'Strategic relatedness' betweentwo markets, in the sense that they valuenontradeable, nonsubstitutabie assets with similarproduction functions, is a requirement for diversi-fication to yield super-normal profits in the long-run. By failing to take into account differencesin the opportunities to build strategic assetsoffered by different market environments, thetraditional measures suffer from the 'exagger-ation' problem described above. They willwrongly impute a benefit to related diversificationacross markets where the relatedness is primarilyamong nonstrategic assets.

In order to operationalize the concept ofstrategic relatedness, we need to develop indi-cators of the importance of similar types ofnontradeable, nonsubstitutabie and hard-to-accumulate assets in different market environ-ments. These types of assets may be divided intofive broad classes (Verdin and Williamson, 1994):

• customer assets, such as brand recognition,customer loyalty and installed base;

• channel assets, such as established channelaccess, distributor loyalty and pipelinestock;

• input assets, such as knowledge of imper-fect factor markets, loyalty of suppliersand financial capacity;


• process assets, such as proprietary tech-nology, product or market-specific func-tional experience (e.g., in marketing orproduction) and organizational systems;

• market knowledge assets, such as accumu-lated information on the goals andbehavior of competitors, price elasticityof demand or market response to thebusiness cycle.

Thus, if our indicator suggested that channelaccess and distributor relationships were likelyto be very important to competitive advantagein each of two markets, we would identify themas 'strategically related' on this dimension.We would then be more confident that corecompetences in building networks of channelrelationships would be applicable to both. If, onthe other hand, the second market involved aproduct that was most effectively sold directly toa small number of buyers, we would class themarkets as having lower strategic relatedness onthe channel dimension. Although these marketsmay be closely related in some other way, suchas use of similar raw materials, the opportunityto benefit from transfer of competences inbuilding a third-party distribution network wouldnot be available. Meanwhile, if all competitorscould buy raw material inputs at a similar price,relatedness on the input dimension would notoffer a source of competitive advantage. Therelatedness between this second pair of marketswould be 'non-strategic'

We could then develop an overall picture ofthe degree of strategic relatedness between pairsof markets by using a portfolio of indicators,each one seeking to measure the extent to whichcompetence in building the same class of strategicasset could add to competitive advantage in bothenvironments. The higher the level of strategicrelatedness between two markets, other thingsequal, the larger would be the expected gainsfrom diversification of firms from one to theother.

We are the first to acknowledge that thestructural indicators we use in the empiricalinvestigation that follows are not direct measuresof how similar the processes required to buildstrategic assets are between two markets. Weare not able to develop a direct, quantitativemeasure of the degree to which core competencesare transferable across SBUs serving each setof markets where our sample firms operate.

However, we would contend that by goingbeyond the standard variables used to captureindustry structure (like advertising and R&Dintensity), to include indicators such as whetherthe product lines in each market are made-to-order or sold from stock, or how many of theproduct lines require after-sales service, we havetaken an important step towards this goal.Specifically, we believe that our measures comecloser to capturing the extent to which twomarkets share similar nontradeable, nonsubstitut-abie and imperfectly imitable assets (oftenintangibles), with potential to draw on thesame root-stock of competences, than traditionalmeasures (especially since these have largelyignored the distribution, marketing and servicerequirements as well as the need for customizationduring manufacture, that our measuresemphasize).

In what follows, we discuss the structuralindicators used for three of the main classes ofstrategic assets on which we have data: customerassets, channel assets, and process experienceassets.

Customer asset indicators

The first set of indicators seeks to capture thefact that the nature of interactions with thecustomer is an important determinant of thetypes of assets necessary to effectively serve amarket.I. Customer concentrationifragmentation. Thisindicator, FEWCUST, measures the degree towhich each manufacturer deals with few, largecustomers rather than interfacing with a frag-mented base of accounts. FEWCUST, is definedas the percentage of product lines for whichthere were less than 1000 customer accounts atthe manufacturer level. If the manufacturer dealsthrough resellers, it will reflect the number ofdistributor accounts which it must manage. Tothe extent that the manufacturer also deals directwith users, these additional accounts are alsoincluded in computing this measure.

Our objective is to capture the extent to whichthere is scope for manufacturers to build deepand sustained relationships with their accountbase: dedicating a member of sales staff toindividual customers; responding to customer'sspecific commercial and information needs;developing proprietary IT interfaces; using con-


signment stocks and so on. To the extent that acorporation's portfolio of markets shares thesesimilar sales management requirments, we wouldexpect there to be greater opportunities forbuilding and sharing core competences in theseareas. Conversely, if a corporation's portfoliowas strategically unrelated on this dimension,including a mix of businesses some with aconcentrated account base and others facingaccount fragmentation, we would expect feweropportunities for building and sharing corecompetences in sales and customer service.2. Service requirement. Good customer relation-ships (service reputation) and organizationalcapital (to provide quality service), both largelynontradeabie assets, are likely to be moreimportant in industries characterized by a highlevel of service requirement. Access to corecompetences in rapidly establishing an organiza-tion capable of providing quality service to thecustomer will offer greater advantage in suchmarkets. To the extent that a corporation'sportfolio of markets shares these similar servicemanagement requirements, we would expectthere to be greater opportunities for buildingand sharing core competences in these areas.The service requirement in the industry ismeasured by the percentage of product linesrequiring a 'moderate to high' degree of aftersales or technical service (HSERV).

Channel asset indicators

A second class of indicators refers to theimportance of imperfectly tradeable, channel-related assets as a basis for competitive advantage.Such assets are likely to be more important inindustries where a large portion of the productsare sold through intermediaries: opportunitiesfor a related diversifier to share competences indelivering products, information and service tothe customer will then be high.3. Channel dependence. Our indicator of thedegree of third-party channel dependence,CHANNEL, is the percentage of products whichpass through an intermediary before reachingthe final user, rather than being sold direct tousers by the manufacturer.

Distribution relationships are a critical assetin many businesses dependent on third-partychannels. They are also difficult to trade on a freestanding basis. Skills in building and managing

distribution and dealer networks form the basisof a potentially important core competence.Where market economics dictate that that depen-dence on third-party channels is high, com-petences in dealer recruitment and overcoming'shelf space' restrictions (Porter, 1976) will bevaluable in assisting an SBU to accumulate theassets it requires to compete effectively. Strategicrelatedness will also tend to be higher amongmarkets that share similar levels of channeldependence.

Process experience asset indicators

In most industries it is important to build uprelevant process experience and associated assetsin order to underpin competitive advantage. Weuse two measures to indicate the degree ofsimilarity between the types of process experienceassets that SBUs require: the proportion ofproducts which are made to order and theaverage skill level of employees.4. Product customization (products made to order)vs. standardization. Successful made-to-ordersupply depends on a variety of asset bases tofacilitate two-way communication with customers,the design and management of flexible manufac-turing systems, and the reduction of lead times.Efficient supply of products from stock requiresa different set of tangible and intangible assetswhich underpin effective stock, control, demandforecasting, and batching and run-lengthefficiencies, etc. In both cases, nontradeableassets play a potentially important role, but thenature of the assets and the competences requiredto accumulate them differ. Strategic relatednessbetween businesses will therefore tend to be highwhen they share a common focus on eithermade-to-order production or supply from stock.

Our measure of relatedness on this dimensionis the percentage of product lines supplied bythe business which are made to order based oncustomer specifications {TORDER).5. The average skill level of the labor force. Inbusinesses where groups of skilled staff are animportant source of advantage, human capitaland the associated systems to generate andmanage will be even more critical to advantagethan in businesses with high labor intensity, butlow skill levels. Again, businesses which sharethe need to develop an effective base of skilledstaff with experience working together will have


higher strategic relatedness than a pair ofbusinesses, one requiring highly skilled staff andthe other, a base of cost effective, low skilledworkers.

Our indicator, SKILL, measures the proportionof 'high-skilled' jobs in the industry as apercentage of total employment.

HYPOTHESES, DATA ANDMETHODOLOGY

Hypotheses

So far we have argued that the traditional wayof measuring relatedness between two businessesis incomplete; to be meaningful, relatednessneeds to consider: (i) the 'strategic importance'of the underlying assets of these two businesses(i.e., are these assets nontradeable andnonsubstitutabie?); and (ii) whether these assetsare related. Only firms that exhibit this type of'strategic relatedness' will perform well in thelong term. This implies that if we were tomeasure the performance of firms classified as'related' in the traditional (Rumelt) way andagain according to whether their underlyingstrategic assets are related, we should be able toshow that the latter way of looking at relatednessis superior. Therefore:

Hypothesis 1: Related diversifiers will outper-form unrelated firms only where they competeacross a portfolio of markets where similartypes of accumulated assets are important.

In addition, the above discussion suggests thateven within the population of related-diversifiedfirms (defined as such in the traditional way),some related firms will do better than otherrelated firms. Specifically, those related firmsthat compete across a portfolio of markets wheresimilar types of accumulated assets are importantshould outperform the related firms that competein a portfolio of markets where accumulatedassets are less important or the types of strategicassets required differ widely across the firm'sportfolio. Therefore:

Hypothesis 2: Related firms that compete in aportfolio of markets where similar types ofaccumulated assets are important will outper-form other related firms.

Data and methodoiogy

To test these hypotheses, a sample of 200 firmswas randomly selected from the 1981 Fortune500 list. The population of Fortune 500 firmswas selected for study because it contains manydiversified firms. The sample firms were classifiedaccording to Rumelt's (1974) diversification cate-gories (i.e.. Single-business; Dominant-business;Related-business; and Unrelated-business), usingdata from the TRINET tapes as well as theirannual reports. Since this is a study on diversifi-cation, the Single-business firms were excludedfrom the sample. Hence, the final sample consistsof 164 diversified firms.

To test Hypothesis 1 we first measurerelatedness in the traditional (Rumelt) way as a(0,1) dummy (RELATED): those firms classifiedas Related or Dominant take the value of 1,while firms classified as Unrelated take the valueof 0. The following equation is then estimated

i = a + Pi (RELATED) +i = 2

+ e

(1)

where ROS is the profitability of the samplefirms, measured as return on sales, and IND areindustry control variables. We use three standardvariables to control for industry effects: R&Dintensity (RDSLS), measured as the total expen-diture on R&D as a percentage of sales in agiven industry; advertising intensity (WXAD),measured as expenditure on media advertisingas a percentage of sales in an industry; andcapital intensity (CAPX), measured as capitalexpenditures as a percentage of sales in anindustry.^ We decided to use return on sales(ROS) rather than return on assets (ROA) as

' Given the multiindustry nature of the sample firms, thesethree variables were industry-weighted. As an example ofhow this was achieved, consider how we industry-weightedWXAD (a firm's industry advertising intensity): First, aCompustat program was used to identify all the firms assignedto each 2-digit SIC, and to calculate their advertisingintensity. Then, using their sales as weights, the averageadvertising intensity of every SIC was estimated. Next, eachsampte firm's sales by SIC were obtained from the TRINETtapes, and the percentage of the firm's sales in each SICwas calculated. The firm's industry-weighted advertisingintensity was then calculated by multiplying the share of thefirm's sales in each SIC by the corresponding advertisingintensity of that SIC, and adding the results.


our profitability variable for a specific reason:As explained by Ravenscraft and Scherer (1987),depending on the accounting method that a firmuses to account for an acquisition (pooling ofinterest vs. purchase accounting), profitabilitymeasures such as ROA end up systematicallysmaller under purchase accounting than underpooling of interest accounting. To avoid thispotential bias we use ROS which is not affected bythe accounting policies of acquisition recording.Finally, even though it is possible that profitabilityaffects the industry variables as much as it isaffected by them, we decided to treat the systemas recursive rather than endogenous because webelieve that feedbacks in the system occur atsufficiently long lags to allow us to pull outindividual equations for separate treatment (e.g..Cowling, 1976).

After estimating Equation (1), we replace theRELATED variable by our structural indicatorsof relatedness to reestimate the above equation,i.e.

ROS, = a e(2)i = 1

where Kj are the five structural indicatorsdescribed in the previous section. In order totest our hypotheses about strategic relatedness,we express each Ki as the weighted average ofeach structural indicator 'i' divided by theweighted variance of the indicator across thebusinesses in each diversified firm's portfolio (A/V).** The reason for this transformation is bestexplained by example. Suppose we have twodiversified firms. Firm X has 70 percent of itssales in a business requiring a high level of after-sales service, but has diversified the remainderof its sales into businesses where the productsdo not require any after-sales service. Theindicator HSERV will therefore have a highvariance for Firm X, suggesting little scopefor sharing competences in building a servicenetwork. By dividing our average indicator bythis variance, we are effectively discounting forthe fact that any competence Firm X has inbuilding an after-sales service network cannotbe exploited in its other businesses (it is

" We'd like to thank an anonymous reviewer for suggestingthis to us.

'imprisoned'). Compare this with Firm Y whichhas all of its sales in two businesses, both ofwhich require high after-sales service. Firm Ywill score in two ways on our indicator Kj(compared with Firm X). It will start out with ahigh weighted average on HSERV. And the factthat both its businesses share a requirement forhigh after-sales service means that the varianceof HSERV is very low for Firm Y, so the valueof its competence in building effective servicenetworks will not be discounted as it was for theunrelated diversifier. Firm X.

To calculate the weighted average of eachstructural indicator we first obtained an industrybreakdown of the indicator from Bailey (1975).Next, each sample firm's sales by SIC wereobtained from the TRINET tapes and thepercentage of the firm's sales in each SIC wascalculated. The industry-weighted average ofeach indicator was then calculated by multiplyingthe share of a firm's sales in each SIC by thecorresponding value of the indicator in that SIC,and adding the results.

To calculate the weighted variance of eachindicator, we used the following formula:

variance (x) = X (̂ ̂ )̂ ~ ^^

where x̂ = weighted average of the indicator x;and s = percentage of each industry (SIC) intotal sales.

A priori we would expect that the coefficientof RELATED in Equation (1) is positive andsignificant, consistent with previous studies onrelated diversification. In addition, for Hypothesis1 to be supported we should find: (a) the R^ ofEquation (2) significantly higher than the R^obtained from Equation (1), implying that Equa-tion (2) does a better job in explaining theprofitability differences between Related andUnrelated firms; and (b) the coefficients of thestructural indicators in Equation (2) to be positiveand significant.

To test Hypothesis 2 we simply estimateEquation (2) again, but only on the subsampleof firms classified as 'related' in the traditionalway (A' = 109). The hypothesis will be supportedif Equation (2) does a good job (as measuredby its R^) in explaining the variability in theprofitability of related firms.

Data for calculating the five structural indi-cators and the three industry-control variables


were derived from the following sources: Thevariables FEWCUST, CHANNEL, HSERV andTORDER, as defined above, were drawn froma U.S. survey of marketing expenditures (Bailey,1975); the variables ROS, CAPX and WXADeome from Compustat; RDSLS was taken fromthe National Seienee Foundation (1978); andSKILL was computed from job classificationscontained in the Census of Population (1980).

RESULTS

Table 1 presents descriptive statistics and corre-lations for all the variables in the study. The lowintercorrelations among these variables suggestno problems with multicollinearity. The lowcorrelations imply that there is sufficient indepen-dent variation among the variables used in thisstudy to allow discrete effects to be estimated.

The estimated coefficients from Equation (1)are reported in Table 2. Consistent with previousstudies, we find that Related diversification ispositively correlated with profitability. Consistentwith IO theory, we also find that the proxies forindustry structure (advertising, capital and R&Dintensity) are also positively correlated withprofitability. The equation is statistically signifi-cant at the 99 percent level and explains about26 percent of the variation in the dependentvariable.

The comparative results obtained from Equa-tion (2) when relatedness is measured by thefive strategic indicators are presented in column(A) of Table 3. Again, the equation is statisticallysignificant at the 99 percent level. Two results

Table 2. Relatedness measured by Rumelt categories"

Variable

Constant

RELATED

WXAD

RDSLS

CAPX

Dependent

(A)

11.828(26.98)***

1.239(1.52)—

—

—

N=160

F=2.30

variable=R0S81

(B)

5.674(6.06)***1.370

(1.94)**0.378

(2.55)***0.788

(2.72)***0.510

(6.33)***N=159

F= 15.05

"t-statistics reported in parentheses'significant at the 10% level (two-tail test)"significant at the 5% level (two-tail test)•"significant at the 1% level (two-tail test)

Stand out: First, the adjusted R^ of this equation(0.36) is significantly (at the 1% level) biggerthan the adjusted R^ obtained from Equation(1). This is strong support of Hypothesis 1 inthat compared with the traditional definition ofrelatedness, our measures of relatedness do a farsuperior job in explaining the variation in thedependent variable. Second, the five structuralindicators that we used to capture strategicrelatedness exhibit the expected sign and arestatistically significant, again strongly supportingHypothesis 1.

Our results with respect to the structuralindicators imply the following:

Table 1. Means, standard deviations and intercorrelations"

Variable

1. ROS2. A/V CHANNEL3. A/V TORDER4. A/V FEWCUST5. A/V HSERV6. A/V SKILL7. CAPX8. WXAD9. RDSLS

Mean

11.9040.7324.3210.660.58

220.66.5433.2291.851

S.D.

4.692.34

21.651.421.11

375.43.932.191.14

1

0.0140.1280.145

-0.0150.180.4160.2150.204

2

-0.8190.245

-0.309-0.336-0.051

0.037-0.057

3

-0.0290.1260.4950.064

-0.0190.023

4

-0.528-0.051-0.197

0.139-0.031

00

- 0- 0

5

.217

.038

.096

.108

6 7 8 9

0.101 —-0.133 0.0002 —-0.119 -0.046 0.240 —

"^=164; Correlation coefficients greater than 0.19 are significant at p < 0.05, those greater than 0.25 are significant at p< 0.01, and those greater than 0.32 are significant at p < 0.001.


Table 3. Relatedness measured by strategic indicators

Variable

Constant

A/V FEWCUST

A/V HSERV

A/V CHANNEL

A/V TORDER

A/V SKILL

WXAD

RDSLS

CAPX

Dependent variable=ROS81

(A)

5.437(3.55)***0.812

(3.13)***0.712

(2.17)**0.549

(2.24)**0.054

(2.01)**0.002

(2.07)0.286**(2.05)**1.04

(3.75)***0.556(7.24)***

N=16A

F= 12.25

(B)

5.773(1.69)*0.906

(3.14)***0.676

(1.89)*0.542

(2.12)**0.048

(1.72)*0.002

(2.12)**0.165

(0.70)1.178

(3.56)***0.572

(6.31)***N=109

F=10.13

'significant at the 10% level (two-tail test)."significant at the 5% level (two-tail test),"'significant at the 1% level (two-tail test).

operating in a portfolio of markets withhigh serviee requirement (HSERV) allowsthe firm to benefit from customer relation-ship skills and accumulated competencesin working with the customer and providingquality service. Other marketing com-petences can be most effectively developedand exploited when the firm operates ina portfolio of markets where a few largecustomers exist.when a manufacturer can operate byserving a few large accounts—FEWCUST—(i.e., its direct customers arenot fragmented), it will be able to buildup deep and sustained relationships withthese customers, and so develop its corecompetences in managing sophisticatedcustomer interfaces.operating in markets where products haveto pass through intermediaries beforereaching the final user (CHANNEL)allows the firm to build up its dealerrecruitment skills and channel manage-ment competences.

operating in markets where products arecustom-made (TORDER) enables the firmto build up and exploit competences inthe exchange of complex informationwith its customers and/or develop flexiblemanufacturing.operating in markets where labor skill ishigh (SKILL) allows the firm to buildup competences in managing knowledge-based activities which can be transferredto other businesses. It may also help thefirm improve its competences in humanresource management.

These results are in general conformity withHypothesis 1. Similarly supporting results wereobtained for the second hypothesis. These resultsare presented in column (B) of Table 3 and arethe estimates of Equation (2) when only thesubsample of Related firms is used. The fivestrategic indicators are able to explain more than41 percent of profitability differences among thepopulation of related firms. This suggests thateven within the population of related - diversifiedfirms (defined as such in the traditional way),some related firms will do better than otherrelated firms. Specifically, since all five ofthe strategic indicators come out positive andsignificant, we can argue that those related firmsthat compete across markets where certain typesof assets (nontradeable, nonsubstitutabie andslow and costly to accumulate) are important,outperform the related firms that compete inmarkets where these accumulated assets are lessimportant.

SUMMARY AND CONCLUSIONS

In this paper, we argued that the traditional wayof measuring relatedness between two businessesis incomplete because it ignores the 'strategicimportance' and similarity of the underlyingassets residing in these businesses. A firm maybe in a set of related businesses without derivinga significant advantage from the potential linksbetween its SBUs. Relatedness will be of littleadvantage when it does not assist the firmin accumulating nontradeable, nonsubstitutabieassets efficiently. This, in turn, implies that it isrelatedness of strategic assets between SBUs that


is important, not the market-level relatednessbetween businesses.

But simply exploiting existing strategic assetswill not create long-term competitive advantage.In a dynamic world, only firms who are able tocontinually build new strategic assets faster andmore cheaply than their competitors will earnsuperior returns over the long term. Corecompetences have a pivotal role to play inthis process. By transferring core competencesbetween its SBUs, a corporation is able toaccelerate the rate and lower the cost atwhich it accumulates new strategic assets. Theseopportunities for benefitting from core com-petences underpin the dynamic advantage ofrelated diversification and define the types ofrelatedness that a firm should seek to exploit(asset amortization, asset improvement, assetcreation and asset fission).

This analysis led us to hypothesize that'strategic' relatedness is superior to marketrelatedness and that related firms outperformunrelated ones only in markets where accumu-lated assets are important. These hypotheses weresupported by our empirical tests. Specifically, wefound that firms operating in portfolios ofbusinesses which shared similar opportunities toexploit brand building: marketing and channelmanagement; and process skills in customizationand management of skilled teams gained signifi-cant benefit from related diversification.

ACKNOWLEDGEMENTS

We would like to thank Harbir Singh, JuliaLiebeskind, Robert Hoskisson, two anonymousreviewers and especially Gary Hamel for manyhelpful suggestions on earlier drafts.

REFERENCES

Ansoff, H. I. (1965). Corporate Strategy. McGraw-Hill, New York.

Bailey, E. L. (1975). Marketing Cost Ratios of U.S.Manufacturers. Conference Board Report No. 662.Conference Board, New York.

Barney, J. B. (October 1986). 'Strategic factormarkets: Expectations, luck and business strategy'.Management Science, 32, pp. 1231-1241.

Bettis, R. A. (1981). 'Performance differences inrelated and unrelated diversified firms'. StrategicManagement Journal, 2(4), pp. 379-393.

Caves, R. E. (1982). Multinational Enterprise andEconomic Analysis. Cambridge University Press,Cambridge, MA.

Caves, R. E. and M. E. Porter (1977). 'From entrybarriers to mobility barriers: Conjectural variationsand contrived deterrence to new competition'.Quarterly Journal of Economics, 91, pp. 241-262.

Caves, R. E., M. E. Porter, M. A. Spence and J. T.Scott (1980). Competition in the Open Economy:A Model Applied to Canada. Harvard UniversityPress, Cambridge, MA.

Cowling, K. (1976). 'On the theoretical specificationof industrial structure-performance relationships',European Economic Review, 8(1), pp. 1-14.

Dierickx, I. and K. Cool (December 1989). 'Assetstock accumulation and sustainability of competitiveadvantage'. Management Science, 35,pp. 1504-1514.

Eaton, B. and R. Lipsey (1980). 'Exit barriers areentry barriers: The durability of capital as abarrier to entry'. Bell Journal of Economics, 11,pp. 721-729.

Hill, C. W. L. (1988).'Internal capital market controlsand financial performance in multidivisional firms'.Journal of Industrial Economics, XXXVII(l),pp. 67-83.

Hill, C. W. L. (1994). 'Diversification and economicperformance: Bring structure and corporate man-agement back into the picture'. In R. Rumelt, D.Schendel and D. Teece (eds.). Fundamental Issuesin Strategy. Harvard Business School Press, Boston,MA, pp. 297-321.

Hoskisson, R. E. and M. A. Hitt (1990). 'Antecedentsand performance outcomes of diversification:Review and critique of theoretical perspectives'.Journal of Managment, 16, pp. 461-509.

Jacquemin, A. P. and C. H. Berry (1979). 'Entropymeasure of diversification and corporate growth'.Journal of Industrial Economics, XXVII(4),pp. 359-369.

Lecraw, D.J. (1984). 'Diversification strategy andperformance'. Journal of Industrial Economics,XXXIII (2), pp. 179^198.

Markides, C. C. and P. J. Williamson (1993).'Corporate diversification and organizational struc-ture: A resource-based view'. Working paper,London Business School.

Montgomery, C. A. (1982). 'The measurement of firmdiversification: Some new empirical evidence'.Academy of Management Journal, 25(2),pp. 299-307.

National Science Foundation (1978). Research andDevelopment in Industry, Technical Notes andDetailed Statistical Tables, Washington, DC.

Palepu, K. (1985). 'Diversification strategy, profitperformance, and the entropy measure'. StrategicManagement Journal, 6(3), pp. 239-255.

Pitts, R. A. and H. D. Hopkins (1982). 'Firm diversity:Conceptualization and measurement'. Academy ofManagement Review, 7(4), pp. 620-629.

Porter, M. E. (1976). Interbrand Choice, Strategy andBilateral Market Power, Harvard University i'ress.


Cambridge, MA.Porter, M. E. (1980). Competitive Strategy: Techniques

for Analyzing Industries and Competitors, FreePress, New York.

Porter, M. E. (May-June 1987). 'From competitiveadvantage to corporate strategy'. Harvard BusinessReview, pp. 43-59.

Prahalad, C. K. and G. Hamel (May-June 1990). 'Thecore competence of the corporation'. HarvardBusiness Review, pp. 71-91.

Ramanujam, V. and P. Varadarajan (1989). 'Researchon corporate diversification: A synthesis'. StrategicManagement Journal, 10(6), pp. 523-551.

Ravenscraft, D. J. and F. M. Scherer (1987). Mergers,Sell-offs and Economic Efficiency. The BrookingsInstitution, Washington, DC.

Reed, R. and G. A. Luffman (1986). 'Diversification:The growing confusion'. Strategic ManagementJournal, 7(1), pp. 29-35.

Rumelt, R. (1974). Strategy, Structure and EconomicPerformance. Division of Research, Harvard Busi-ness School, Cambridge, MA.

Rumelt, R. P. (1987). 'Theory, strategy and

entrepreneurship'. In D. Teece (ed.). The Competi-tive Challenge. Cambridge, Ballinger, MApp. 137-158.

Singh, H. and C. A. Montgomery (1987). 'Corporateacquisition strategies and economic performance'.Strategic Management Journal, 8(4), pp. 377-386.

Teece, D. J. (1982). 'Towards an economic theoryof the multiproduct firm'. Journal of EconomicBehavior and Organization, 3, pp. 39--i63.

U. S. Bureau of Census (1980). 'Job classificationstatistics'. Census of Population, U.S. GovernmentPrinting Office, Washington, DC.

Verdin, P. J. and P. J. Williamson (1994). 'Corecompetences, market analysis and competitiveadvantage: Forging the links'. In G. Hamel andA. Heene (eds.). Sustainable Competitive Advantagethrough Core Competence. Wiley, New York,forthcoming.

Williamson, O. E. (May 1973). 'Markets and hier-archies: Some elementary considerations', /ImericflwEconomic Review, 63, pp. 316-325.

Williamson, O. E. (1975). Markets and Hierarchies.Free Press, New York.

Documents

Related Diversification, Core Competences and Corporate Performance.pdf