environment, market share, and market power

Upload: aharyandini

Post on 06-Apr-2018

230 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/3/2019 Environment, Market Share, and Market Power

    1/19

    Environment, Market Share, and Market Power

    Author(s): William Boulding and Richard StaelinSource: Management Science, Vol. 36, No. 10, Focussed Issue on the State of the Art in Theoryand Method in Strategy Research (Oct., 1990), pp. 1160-1177Published by: INFORMSStable URL: http://www.jstor.org/stable/2632658 .

    Accessed: 12/10/2011 03:58

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

    JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

    content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

    INFORMS is collaborating with JSTOR to digitize, preserve and extend access toManagement Science.

    http://www.jstor.org

    http://www.jstor.org/action/showPublisher?publisherCode=informshttp://www.jstor.org/stable/2632658?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/2632658?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=informs
  • 8/3/2019 Environment, Market Share, and Market Power

    2/19

    MANAGEMENT SCIENCEVol. 36, No. 10, October 1990

    Printled n U.S.A.

    ENVIRONMENT, MARKET SHARE, AND MARKET POWER*WILLIAMBOULDINGANDRICHARDSTAELIN

    Fuqla School QfBiusiness, Dutke University, Durh1am,North Carolinca 7706In this paper we develop a process model relating market share to firm profits. In particular,we specify averageprice and averagecost equations as a function of previous year market shareposition, changes in market share,environmental conditions, and interactions of environmentalconditions with the laggedmarketposition and market sharechange variables. We estimate theseequations using PIMS data after controlling for unobservable factors. Our results suggest thatfirms with high market shares derive no extra market power benefits except if they operate inenvironments with little buyer power. Instead,environmental factors and changes in marketsharemost strongly influence price and cost. An analysis of the marketshare associated with maximalfirmprofits indicates that for the majorityof firmsin our sample steadily increasingmarket sharedoes not always associate with increasing profits, giving credence to the proposition that "moremarket share is not always better."(MARKETING STRATEGY; MARKET SHARE; MARKET POWER; FIXED EFFECT ES-TIMATION; PIMS)

    1. IntroductionExtensive documentation exists for the stylized fact of a positive relationship betweenprofitability (as measured by ROI) and market share (see Buzzell and Gale 1987, for areview of this relationship). However, researchershotly contest the reasons for this re-lationship. One group of strategy researchers argues that "third" unobserved variables

    such as managerial expertise (e.g., Jacobsen and Aaker 1985), or the firm's ability todeliver superiorvalue (e.g., Anterasian and Phillips 1988) correlatewith both profitabilityand market share. These strategists posit that better run firms not only have lower costsper unit of sales and therefore higher profit margins,but also are more skilled marketers,and therefore attain higher market shares.A second group of strategyresearchers maintains that a firm's market share positionenables it directly to increase its profitability (e.g., Buzzell and Gale 1987) through bothincreasedvolume that provides higherrevenues and lower averagecosts due to experiencecurve effects(Day and Montgomery 1983), and greatermarketpower (e.g., Martin 1988,Schroeter 1988, Staten et al. 1988). This market power allows the higher market sharefirm to obtain lower input costs and get more functions from its channel members at agiven price as well as set prices ratherthan acting as a price taker.Adherence to the belief that more market share is better leads a number of firms toset corporate policy guidelines that precludethe firm from competing within an industryunless it has a high probabilityof being the number one or two player (e.g., see GeneralElectric Company 1981). However, this belief does not receive universal acceptance.Some question the premise that high market share leads to market power (e.g., Fisheret al. 1983). Others point to the need for considering the costs of obtaining and/ormaintaining market share as well as the benefits associated with the increased share(Fruhan 1974, Wernerfelt 1985, Montgomery and Wernerfelt 1989).Given the uncertainty concerning the influence of market share on profitability,andthe strategic importance of the question, we revisit the question of how market shareand profitability relate. We do this by drawing on the literatures of business strategy,

    * Accepted by Diana L. Day, John U. Farley,and JerryWind.1160 0025- 1909/90/36 10/1 160$01.25

    Copyright?C1990, The Institute of ManagementSciences

  • 8/3/2019 Environment, Market Share, and Market Power

    3/19

    ENVIRONMENT, MARKET SHARE AND MARKET POWER 1161marketing, industrial organization, and economics to develop.a model of how marketshare affects profits after controlling for the effects of industry structure.We use the PIMS database to estimate this model, controlling for unobserved factorsthat create a potential bias in estimating market share effects. Specifically, we removefrom our analysis both the effects of firm-specific factors that remain unchanged overthe period of analysis (such as managerial expertise or providing the customer withsuperior value), and the effects of unobserved contemporaneous factors, such as luck(e.g., a serendipitous product innovation). After controlling for these unobservables, wecan assess the validity of the argument that the observed market share-profitability re-lationship is entirely due to unobserved factors.In addition, our method of analysis provides four enhancements over prior PIMSanalyses. First, instead of measuring profitability with return ratios such as return oninvestment, sales, or assets, we investigate the influence of market share on the firm'stwo components of profit-annual total revenues and annual total costs. Decomposingprofits into these components enables us to develop and test a process model of how agood market share position might affect a firm's ability to obtain higher average pricesand/or lower average costs, thereby leading to higher profits.Second, our model explicitly acknowledges that firm actions and the environment inwhich the firm operates affect a firm's ability to earn increased profits. We represent thefirm's environment via a set of external forces previously postulated to affect a firm'sprofitability, i.e., supplier power, buyer power, competitive rivalry within the industry,and potential threat of entry into the industry (Porter 1980).' We postulate that theseforces directly affect prices and costs, as well as modify the direct effect of initial marketposition on prices and costs.Third, our model incorporates dynamic market share effects, i.e., what happens whena firm experiences shifts in market share? Again, we postulate that these shifts yielddifferent benefits and costs depending on environmental conditions facing the firm.Finally, decomposing profits into total revenues and total costs allows us to estimateboth the marginalrevenueproductand marginalfactor cost of market share. This providesan equilibrium frameworkand allows us to calculate the optimal market share for eachbusiness unit in our sample.

    2. Process Model of Firm ProfitabilityWe study firm profitability by modeling the firm's average price and averagecost. Ourbasic premise is that a firm can increase its profits by effectively shielding itself fromexternal forces that adversely affect its ability to get higher prices or lower costs. Thisshieldingcould occur via marketshareposition or other mechanisms such as highbarriersto entry, power over the actions of others (i.e., buyers or suppliers), or competing in anindustrywith little rivalry. Higher market share might enable the firmto dictate policiesto its suppliers and/or its distribution network, or give the firm monopoly power insetting prices to buyers. Alternatively, nonmarket share factors such as having access to

    many sources of supply (thereby allowing the firm to negotiate for and obtain a lowinput cost), or operating in an industrywith high plant utilization (thereby reducingtheneed for competitors to cut price), might influence average price or averagecost.Table 1 provides a list of previously suggested factors that enable a firm to get lowerunit costs or higher unit prices. We organize these factors into six general categories:power over suppliers, power over buyers, lack of competitive rivalry,lack of competitiveentry threats, market position, and firm factors.' Anotherapproach o capture he firm'senvironmentusesmeasuresof the firm'sown and cross-price lasticities,

    e.g., see Hanssens ( 1980),;and Carpenter et al. ( 1988).

  • 8/3/2019 Environment, Market Share, and Market Power

    4/19

    1162 WILLIAM BOULDING AND RICHARD STAELINTABLE 1

    Factors that Might In/lience a Firm's Abilitvto Get Highlcer rices or Lo4'er CostsPow,erover Suppliers (POS)

    Firm has alternativesources of supplySuppliersrely heavily on firm for businessPowver ver Buyer~(POB)Firm has largerthan averagebuyer baseProduct is unimportant to buyerFirm has few new productsFirm was early entrantFirm has customized productsLack of CompetitiveRivalr) (LCR)Industry has little brandswitchingIndustry has high real market growthIndustry has variance in size of competitorsIndustry has high plant utilizationIndustry has no regularnew productintroductionsProduct is early in product life cycleIndustry has recent entrant and/or no recentexitsIndustry has many competitorsFirms have low exit costsLack of Threat of Competitive Entrif (LTE)Industry has high capital barriersFirm has patent protectionFirm has high shared costsMarket Position (MP)Firm has large market shareMajor competitors have small market share

    Firtm Factors (FF)Firm delivers superiorproduct-qualityFirm delivers high quality advertisingFirm has a high quality sales forceFirm uses high quality promotionsFirm delivers superior valueFirm has managerialexpertiseFirm has good luck

    In order to develop a theoretical model of how these six categories affect firm perfor-mance we proceed as follows. First, we paitition firm profits into its two components,total revenues (TR) and total costs (TC). Using accounting identities, we recognize thattotal revenue and total cost for year t equalTR,-Q, *P,-IS, *MS, *P,, (l1a)

    andTC, Q,*C,- IS,*MS,*C, (lb)

    respectively, whereQ, quantity in units for year t,P= firm's average unit price,IS, = industry sales in units,MS, = firm's market share,Cl = firm's average unit cost, andwe drop the firm subscripts for notational simplicity.

  • 8/3/2019 Environment, Market Share, and Market Power

    5/19

    ENVIRONMENT, MARKET SHARE AND MARKET POWER 1163From ( la) and ( Ib), firm profits equal

    II, IS,*MS,*(P, - C,). (2)Examination of (2) reveals that if a firm increases its profits, it must do so via volumeincreases and/or by increasing the difference between unit price and average unit cost,i.e., P, - C,. Since proponents of the share-causes-profitabilityhypothesis attribute thesize of this difference to the firm's market share position, we test this hypothesis byformulating a model of how a firm's average unit price and average unit cost vary withmarket share. We start by assuming that P, consists of a component that varies over timedue to observable changes in the firm'sstrategyor environment, a fixedfirmand industry-specific component, and a random unobservable component. Formally,

    P - ( 3)where X,, is a vector of observable time varying factors that influence firm price, d is theassociated parameter vector, Z, is a vector of time-invariant influences such as, perhaps,managerial expertise and certain industryconditions faced by the firm, y is the associatedparametervector, and c,, representsa random errorterm.2Similarly, we specify that the firm's average unit costs consist of a component thatvaries over time according to changes in the firm's observable strategyand environment,a fixed, firm and industry-specific component, and a random errorterm. Specifically,

    C, e-X2,+aZ2+Q., (4)where X2, is a vector of observable time varying factors hypothesized to affect C,, Z2 isa vector of time invariantfactors that influence C,, a and Xaretheirassociatedparametervectors, and (2, represents the random errorcomponent.Equations (3) and (4) specify the general form of the price and cost functions. Table1 identifies the types of variables we want to include in the X and Z vectors above. Wesummarize our beliefs on how these factors influence price and cost through a series ofpropositions and conjectures.We base our first proposition on the belief that firms in less competitive environmentsobtain higher prices since they do not have to worry about competitive reactions and/or having buyers challenge their offered price. More formally:

    Pl. Afirmn'saverage annutalu1it price increases (all else eqlual) n environments tlatshield thefirm.from adversecomnpetitiveonditions, i.e., in environmnents ith little corn-petitive rivalry, low threat of comnpetitive ntry, and/or high power over the buyer.Our second proposition is a variant of the statements of Buzzell and Gale ( 1987), andreflects the belief that a high market share position gives a firm market power and thusthe ability to set higher prices:P2. Afirmns average annutal nit price increases(all else equlal with thefirmns mnarketposition, and this increase occurs at a decreasing rate.The next two propositions concern the effects of changes in dollar market share onprice. Holding fixed factors that shift a firm's downwardsloping demand curve, increases(decreases) in market share over the last year should associate with decreases(increases)in price. Moreover, firms in good environments (i.e., have low own price elasticities)should experience smaller price reductions for a given gain in market share than thosein adverse environments. This leads to the following two propositions:2 We specify a nonlinear function for two reasons. First, as will become clear subsequently, we take the logof P,and this form is linear in log space. Second, depending on the specific form of variablesused, this specificationallows us to capture a wide variety of functional relationships.

  • 8/3/2019 Environment, Market Share, and Market Power

    6/19

    1164 WILLIAM BOULDING AND RICHARD STAELINP3. A firm 's average annutal utnitprice decreases (all else equtal) n the vear it gainsdollar mnarkethare.P4. The cf&cis postutlatedin P3 are less in environments that shield the firm fiomncompetition.Finally, we recognize that the effect of market position on a firm's ability to get higherprices may vary according to the type of external environment in which the firm operates.One view holds that the effects of market position have greater value for firms in adverseenvironments, since only then is market position needed to shield the firms from thereactions of others. Alternatively, one could argue that firms in good environments usetheir marketpositionto geteven higherprices. We take the former position in the followingconjecture:Cl. The effectspostutlated n P2 generalIv decrease in environments that alreadYshield

    thefirm from adverse competitiveconditions.Our next two propositions concern averagecost. The firstderives from the belief thatfirms with better market share position can extract larger price concessions from outsideagents, relative to firms with lower market share position. Consequently, high marketshare firms should have lower average costs. The second follows from the belief thatincreases in market share allow a firm to spread its fixed costs over a larger number ofunits. More formally,P5. A firm 's average annutalutnitcost decreases (all else eqlual) with high initialmarket share position, and decreases at a decreasing r-ate.P6. A firm s average annutal utnitcost decreases (all else equial) in the Yeara firmincreases in mnarkethare.The effect of environmental conditions on the firm's average cost could be seen intwo ways: either firms in easy environments do not spend as much money to get thesame results, or because of easy conditions these firms operate less efficiently, resultingin higher average costs. In the following three conjectures we opt for the belief thatinefficiencies caused by easy conditions outweigh cost advantages:C2. The firm's average annutaliunitcost increases when thefirm shields itself fromnadverse competitiveconditions.C3. The effects of P5 decrease when a firmnshields itself fron adverse competitiveconditions (i.e., costs go down less).C4. The effects of P6 decrease when a firm shields itself from adverse competitiveconditions (i.e., costs go down less).Table 1specifiesthe types of variableswe might use to measurecompetitive conditions,firm actions, and market position, while propositions PI-P6 and conjectures C1-C4more

    clearly identify the functional forms and anticipated signs for the parameters : and cafrom equations (3) and (4). Specifically,we include linear and quadraticterms for laggedmarket position (where we define market position as a firm's market share divided bythe market share of its three biggest competitors) and change in market share.3Thesevariablesallow us to test P2, P3, P5 and P6. We also include variablesfelt to measuretheenvironmental forces, thereby allowing us to test P1and C2. Where relevant, we interactthese external forces with the lagged market position and market share change variables.3Although not specified in our propositions referring to changes in market share, we include a quadraticterm to capture any nonlinearities.

  • 8/3/2019 Environment, Market Share, and Market Power

    7/19

    ENVIRONMENT, MARKET SHARE AND MARKET POWER 1165This specification enables tests of P4, Cl, C3, and C4. Finally, we include a series ofvariablessuch as relative product quality, salesforceexpenditures,advertisingexpenditures,and promotional expenditures to control for activities that might shift the firm's demandor cost curves.

    Although we develop a method of analysis that controls for unobservable factors suchas managerial expertise, specialized assets, or luck, we do not estimate the magnitude ordirection of their effects. We discuss this more fully when we describe our analysis pro-cedure.3. Database and Independent Measures

    Since much has been written about the PIMS database (e.g., Anderson and Paine1978) we limit our discussion to those elements of the database essential to our study.Our sample consists of a longitudinal history of all business units for which PIMS hasat least 9 yearsof history.4Our sample contains 3,250 annualobservations for 340 differentbusiness units.We test our propositions and conjectures by estimating the parametersd and ca n (3)and (4). To do this we first recognize that PIMS does not contain measures of manyfixed firm-specific variables postulated to influence price or cost, nor does it containdirect measures of P, and C,. However, as we show in ?4, one can obtain measures ofthese latter two variables for estimation purposes. We do this by recognizing from ( la)and ( Ib) that P, and C, are ftinctions of TR,, TC,, MS,, and IS,. Although the PIMSdatabase does not contain measures for TR,, TC,, or IS,, it does contain measures thatallow us to deriveestimating equations for P,and C,. For example, disguisedtotal revenues(DTR,) and disguised total costs (DTC,) appearin the PIMS database. Researchersusuallyavoid these disguised measures since each firm i uses its own disguise factor Ki, whereDTR, = Ki*TR, and DTC, = Ki*TC,. However, as shown by Moore and Boulding(1987), we can remove the unknown disguise factor (along with unknown IS,) from theanalysis by taking logs and then differencing.5We latergive full exposition to this process.Table 2 provides a brief description of the measures we use to represent the factorslisted in Table 1, whether each factor is measured annually or only once (i.e., is eithertime variant or fixed), and whether it is at the business unit level or industry level.Although most names communicate the variablecontent, a few require explanation.

    First, we derive our industry capital barriersvariable by dividing the business unit'sassets by its market share.Thus, we assume the business unit's asset base, afteradjustingfor its proportion of industry business, represents a good proxy for the industry levelbarrier.Also, since the business unit's assets are disguised we remove the disguise factorvia the previously noted transformations. Second, we base the brand switching variableon annual changes in the top four firms' market share levels. We recognize that thisaggregatevariable may not measure the true extent of brand switching. Third, ratherthan using an empirical model containing each of the measures listed in Table 1 tocapture the environmental forces, we use one aggregatemeasure per force. We createthese single measuresby standardizing(mean zero, standard deviation one) the measuresof each factor and then summing all the standardizedmeasuresassociated with a specificforce. Thus, the aggregatemeasure Power over Buyers (POB) representsthe sum of fivestandardized variables, Power over Suppliers (POS) two, Lack of Competitive Rivalry(LCR) nine, and Lack of Threat of Entry (LTE) three. Finally, we additively rescale4 This screen may result in slightly more successful PIMS business units (e.g., the average market position inour sample (.779) is higher than that reported elsewhere (0.633 in the Strategic Planning Institute ( 1988)).Since we control for firm-specific effects, this reduces the potential for selection bias to adversely affect ouranalysis.5See Hagerty et al. ( 1988) for a different approach to removing the disguise factor.

  • 8/3/2019 Environment, Market Share, and Market Power

    8/19

    1166 WILLIAM BOULDING AND RICHARD STAELINTABLE 2

    VariableDefinitionsaTable 1 Time Variant Business UnitVariable Category or Invariantb or Industryc Definition

    Lagged Market Position MP V BU Business unit dollar market share,(LMP) divided by the sum of its threelargest competitors dollar marketshare, lagged one year.Change in Dollar MP V BU Business unit change in share pointsMarket Share ($MS) from the previous year.Relative Product FF V BU The percentage of this business' salesQuality (PQ) volume accounted for by productsand services that, from theperspective of the customer, areassessed as "superior,"to thoseavailable from the three leadingcompetitors less those assessed"inferior."

    Relative Advertising FF V BU Relative to leading competitors, how(AD) much did this business spend, as apercentageof sales, on advertising?1 = much less2 = somewhat less3 = about the same4 = somewhat more5 = much more.Relative Promotion FF V BU Relative to leading competitors, how(PROMO) much did this business spend, as apercentageof sales, on promotions?Same coding as AD.Relative Sales Force FF V BU Relative to leading competitors, how(SF) much did this business spend, as apercentageof sales, on salesforce?Same coding as AD.Low Supplier Power POS F BU Multiply the percentage of your totalpurchases from your three largestvendors times the variable "are

    there alternatesources of supply,"which is coded0 = yes, with no difficulty= yes, but with difficulty2 = no alternate sources of supply.Subtractthis number from 200%, i.e.,high numbers mean the firm hashigh power over its suppliers.

    Supplier Reliance POS F BU What percentageof the sales of yourthree largestexternal vendors ismade to your business?Customer Concentration POB F BU Relative to leading competitors howconcentrated are your immediatecustomers?

    = more concentrated2 = about the same3 = less concentrated.Importance to Customer POB F I Indicate percentageof customers' totalannual purchasesaccounted for bytypes of product or services sold by

    your businessI = more than 25%2 = between 5% and 25%

  • 8/3/2019 Environment, Market Share, and Market Power

    9/19

    ENVIRONMENT, MARKET SHARE AND MARKET POWER 1167TABLE 2 (coni'd)

    Table 1 Time Variant Business UnitVariable Category or Invariantb or Industryc Definition3 = between 1%and 5%4 = between .25% and 1%5 = less than .25%.Few New Products POB V BU The percentage of the total sales forthis business accounted for byproducts not introduced.during the3 precedingyears, relative to

    competition.Orderof Entry POB F BU At time of entry your business wasviewed as1 = a later entrant2 = an early follower3 = one of the pioneers.Product Customization POB F I The products or services offeredbyyour business and your leadingcompetitors are0 = more or less standardized for allcustomers.I = custom designed for individualcustomers.Market Share Stability LCR V I The sum of the absolute values of theyearly change in dollar marketshare for the business unit and itsthree largest competitors.Real MarketGrowth LCR V I Percent change in real served marketgrowth.Diversity of Competitor LCR V I Variance in size among your top threeBalance competitors.

    Capacity Utilization LCR V BU The average percentageof standard(% Utilization) capacity utilized.No New Product LCR F I New product development has theDevelopment following pattern in your industry0 = annually, seasonally, orperiodically

    I = no regular pattern, with less than1 year lag between initial R&D andcommercial sale2 = no regular pattern, with a I to 2year lag between initial R&D andcommercial sale3 = no regular pattern, with a 2-5year lag between initial R&D andcommercial sale4 = no regular pattern,with a greaterthan 5 year lag between initialR&D and commercial sale5 = no new products.Product Life Cycle LCR F I The stage of development for yourproducts and services isI = decline stage2 = maturity stage3 = growth stage4 = introductory stage.Attractive Industry LCR F I This variable is the sum of thefollowing two measures:During the past five years have anycompetitors with at least 5% market

    share dropped out of your market.

  • 8/3/2019 Environment, Market Share, and Market Power

    10/19

    1168 WILLIAM BOULDING AND RICHARD STAELINTABLE 2 (cont'd)

    Table I Time Variant Business UnitVariable Category or Invariant' or Industryc Definition0 = yes= noDo you have a new competitor thathas gained at least a 5% marketshare in the last 5 years.0 = no1 = yes.Number of Competitors LCR F I For the last year of data being

    entered, approximately how manycompeting businesses were in theserved market?Ignore competitors with less than 1%of the served market.5 = 51 or more4 = 21-503 = 11-202 = 6-101= 1-5.Exit Barriers LCR V BU The net book value of plant andequipment divided by the grossbook value, subtractthis from themaximum of this value, i.e., largernumbers indicate older equipment.

    Capital Barriers LTE V I The ratio of natural log of the BU'sinvestment plus assets divided by 2,divided by the BU's market share.Patent Protection LTE F BU Does your business benefit frompatents on product or process?0 = no= on either product or process2 = on both product and process.

    SharedCosts LTE F BU This variableis the sum of the sharedfacilities, markets/distributionchannels, and marketing programs,where shared facilities are coded1 = less than 10%of P&E2 = between 10%and 80% of P&E3 = 80%or more of P&E,shared markets/distributionchannelsare coded1 = less than 25% of sales2 = 25% to 49%of sales3 = 50% to 74% of sales4 = 75% or more of sales,and shared marketing programsarecoded1 = less than 10%of marketing2 = between 10%and 80%ofmarketing3 = 80% or more of marketing.

    a Many of these variables have been recoded from their original values. Definitions come from the PIMScodebook (Strategic Planning Institute 1988) and data forms (StrategicPlanning Institute 1987).b Time variant = V. Time invariant = F (fixed).c Business unit = BU. Industry = I.

  • 8/3/2019 Environment, Market Share, and Market Power

    11/19

    ENVIRONMENT, MARKET SHARE AND MARKET POWER 1169each force so that the value of the firm with the lowest value equals zero (with all othersgreater than zero). Consequently, increasing values for each force imply increasinglygood competitive conditions.Also, note that we characterize an environment as one of high competitive rivalrywhen it hasftjv competitors, no recent entry, and recent exit. We believe that, holdingmarketsharefixed, a firm with few competitors more significantlyinfluences the demandof any one competitor, and thus worries about competitive reactions (e.g., Pepsi vs.Coke). Also, recent exit indicates difficult competition, while no recent entry indicatesunattractive conditions.6Finally, we acknowledge that some of our one-time measures may not accuratelyreflect conditions over the full period of analysis. Specifically, some of the "fixed"PIMSvariables may have a time-variant component.

    4. Method of AnalysisAs mentioned above, our goal is to estimate the influences on P, and C,. In order todo this we start with the basic accounting identities in ( la) and ( Ib). First, we convertall dollar amounts to constant dollars by multiplying each dollar amount by the GNPdeflator. We then rewrite the unobservable measured IS, by relating IS, to IS,-, with theuse of a growth factor (GS,). Specifically, we note that IS, IS_I *GS,, i.e., unobservedindustry sales in year t equal last years unobserved industry sales times the industrygrowth rate, which we do observe. Also, we note that the market share variable providedin the PIMS data base measures dollar market share ($MS,). Thus MS, $MS,/RP,,where RP, represents the firm's price relative to the industry average price. Rewriting(la) for year 1, and writing it for year I - 1 as well, we get

    TR, DTR,/Ki, IS,-,*GS,*P,*$MS,/RP, (5)and

    TR,I- DTRI /K- IS,, *P,_1*$MS,1/RP1I1. (6)Taking logs and then subtracting (6) from (5 ) yields

    In DTR, - In DTR,_1 I=n GS, + ln P, -ln P,_1+ ln $MS, - ln $MS,1-,ln RP, + In RP,-1, (7)

    or,ln P, - ln P,_1 y,, (8)

    where y, is an observable variable, i.e.,y,-ln DTR, - ln DTR,_1 - ln GS, -ln $MS, + ln $MS,-, + ln RP, - ln RP,-1.Following the same process of taking logs and then differencingfor (3) produces

    ln P, - In P,1 = O(M', - X[ ,I ) + ,I ,,-l . (9)Substituting (8) into (9) yieldsy = OM(X,, - Xi, ) + f1,j - .I- ( 10)

    Two things become clear from ( 10). First,even though we do not have a direct measurefor P, we can estimate the vector of weights : found in (3) using OLS (or some other6 We empirically verify that these three factors operate as conjectured by replicating the analysis withoutsumming the forces, but instead including all the factors as individual measures. The signs for the coefficientsfor each of these three factors agree with the sign for the competitive rivalry coefficient. However, we recognizethat one could make opposite conjectures with regard to these variables.

  • 8/3/2019 Environment, Market Share, and Market Power

    12/19

    1170 WILLIAM BOULDING AND RICHARD STAELINsimilar estimation procedure),assuming the error erm (E - ,, ) is multivariatenormal.Second, we obtain estimates of d even tlhoughwe do not observe firmspecific fixed factorspostulated to affect PI. Consequently, an attractive feature of this estimating equation isthat it allows us to estimate the effects of time varying factors on a firm's average price,wlhilecontrolling for the unobserved firm specific fixed factors, such as managerial skills.We obtain an estimable equation for C, in an analogous fashion. Following the sameprocedures as shown above yieldsln DTC, - ln DTC,_, ln GS, + ln C, - ln C,1

    + ln $MS,-ln $MS,_ -1n RP, + ln RP,_,. (11)Taking logs and firstdifferences on (4), and then substituting into (11) produces

    WI = (X2 X2,1 ) + 162.1(21- ( 12)where

    W,--ln DTC, - ln DTC,_ - ln GS, - ln $MS, + ln $MS,-, + ln RP, - ln RP,_,.Again, note that we can estimate ( 12) witlh he firm-specific component removed fromthe analysis. Therefore,using our estimating eqiations ( 10) and ( 12), we obtain estimatesof the parameters associated with the time varying factors in our structural equations(3) and (4).An overview of the procedures discussed in this section appears in Exhibit 1. We nextdiscuss estimation of our derived equations.

    5. EstimationThe first differencing estimation procedure discussed above controls for unobservedfixed factors, such as managerial expertise, that correlate with market share, price, andcost.7 However, to ensure that no time varying unobservables (such as luck) correlatewith market share, price, and cost we utilize a 2SLS estimation procedure in which weinstrument $MS,, thereby breakingthe potential dependency between the unobservabledeterminants of market share and price and cost. Following Jacobson ( 1989), we uselagged variables as instruments in order to control for possible contemporaneous cor-

    relation with the unobservables in equations ( 10) and ( 12).Finally, even though we use the GNP deflatot to transform all dollar figures intoconstant dollars, we include yearlytime dummies to account for any systematic misspec-ification of the deflator. Since these time dummies have no strategicimplications, we donot discuss them again.6. Results

    Table 3 presentsthe resultsof our price equation estimates. These resultsshow supportfor Pi, which suggeststhat firms obtain higher prices when they face better than averageindustryconditions or shield themselves from competitive forces in a betterthan averagefashion. In particular, he coefficientson the variables hought to measuredifferentaspectsof the degree of competitive rivalryand threat of entry are positive and significant at the0.01 level. Although the coefficient on the variable thought to measure differentaspectsof power over buyers is insignificant, these results provide general support for Porter'sconceptualization of how industry structure influences prices (and in turn, profits).Our estimates do not support P2, which implies a positive coefficient on the laggedmarketposition variableand a negativecoefficienton the quadratic aggedmnarketosition7 Because PIMS observations consist of large, well-establishedbusiness units, we consider managerialexpertisereasonably fixed from yearto year. This might not be the case for a small, young firm undergoing rapidchange.

  • 8/3/2019 Environment, Market Share, and Market Power

    13/19

    ENVIRONMENT, MARKET SHARE AND MARKET POWER 1171EXHIBIT 1

    Analysis OverviewStep Purpose

    (1) Deflate current dollar measures into constant dollars Controls for inflation(2) Arrangeterms according to accounting identities Allows creation of variables(e,g., price)(3) Expressunobserved industry sales from the identity in Enables elimination of unobserved industryterms of its year t - I value and an observed growth salesfactor for year t; lag the identity and express industrysales by its current (year t - 1) value(4) Take logs of identity for years t and t - 1 Causes unobserved values to appear in anadditive fashion(5) Difference the identity expression for years t and t - 1 Removes unobserved fixed effects, disguisefactor, and unobserved values expressedwith a growth factor(6) Difference the independent variables Allows estimation of equation of theoreticalinterest (e.g., price) using within groupestimation(7) Estimate using a two stage least squares procedure Controls for problems of contemporaneouscorrelation (use laggedinstruments) orerrorsin measurement

    variable. We initially obtained estimates for both of these parametersvery close to zero.An F-test could not reject the hypothesis that these parametersequal zero (either maineffects alone, with or without interactions in the model, or considered jointly with thelagged market position interactions). Therefore, we eliminate these variables from themodel, with no discernible effects on the remaining parameters.Thus, we do not findsupport for the idea that market share itself provides monopoly power in setting prices.We find support for P3, which suggests that the coefficients on dollar market sharechanges describe a negative relationship between market share change and price. Ourestimates suggesta strictlynegative association between market share changes and price.Both the linearand quadraticterms on the change in market share variables are negative,with the linear term significant at the 0.01 level.

    We find strong supportfor P4, in that we observe positive coefficientson dollar marketshare changes interacted with industry forces. These results, combined with the maineffect findings, indicate that firms in good environments, as measured by the industryforces, face more inelastic demand curves. In fact, the demand functions for firms invery good industry environments appearso inelastic that small increasesin market shareassociate with higher prices. This result can occur because market share is measured interms of dollar volume. Consequently,higherpricesin conjunction with inelasticdemandcan result in increased firmrevenues, thereby giving the firm a highershare of the marketdollars.In C, we speculate that a business unit's market position will differentially influenceprice depending on industryconditions as measuredby the force variables.The directionof this effect becomes moot as a model's test shows that marketposition contributes noexplanatorypowerto the model, eitheralone, or interactedwith environmental variables.Therefore, we remove these variablesfrom the model.Table 4 presents the results of our averagecost equation estimates. These results showno supportfor P5, which implies a negativecoefficienton the linearlaggedmarketpositionvariable and a positive coefficient on the quadratic term. We could not reject the nullhypothesis that the main effects of lagged market position equal zero (either with orwithout interactionsin the model). Therefore,we remove these variables from the model,with no discernible effiectson the remaining parameters.This result implies that market

  • 8/3/2019 Environment, Market Share, and Market Power

    14/19

    1172 WILLIAM BOULDING AND RICHARD STAELINTABLE 3

    A verage Price EqluationEstimatesa,LMP ZA$MSVariable Mean SD Main Effect Interaction Interaction

    Market PositionLMP 0.779 1.144 OcLMp2 1.916 10.006 OcA$MS 0.323 0.633 -0.2762***(0.0389)zA$MS2 0.506 0.771 -0.0036(0.0081)Firm PositionPQ 27.450 27.530 6.4E -4***

    (2.2E - 4)SF 3.032 0.985 6.5E - 4(5.1E - 3)AD 2.777 .1.064 -2.6E - 4(4.9E -3)PROMO 2.928 0.982 0.0026(0.0050)For-cesPOB 10.664 2.263 -0.0030 Oc 0.0136***(0.0027) (0.0028)LCR 9.183 2.214 0.0044*** 0c 0.0043***(0.0013) (0.0016)LTE 2.598 1.518 0.1109*** 0c 0.0399***(0.0066) (0.0065)

    a Equation R2 = 0.257.bStandard errorsin parentheses.cDropped from model after unable to reject the null hypothesis that all of the LMP coefficients equal zero.*** Significantat the 0.01 level.

    share position, in itself, does not provide market power over suppliers and distributors,and thus, lower averagecosts.Examining the coefficients on the linear and quadratic market share change variablesreveals strong support for P6, which predicts a negative relationship between marketshare changes and average costs. The coefficient on the linear term is negative and sig-nificant.The quadratic erm is also negative,but insignificant.Thus, as we expect, increasesin market share associate with lower average costs, perhaps due to spreading fixed costsover a higher volume.Turning to results related to our conjectures, C2 suggests that firms shielded fromadverse competitive conditions experience higher average costs relative to those firmsleft unprotected from adverse conditions. Estimates on the three environmental maineffects tested do not show strong effects.8Although all estimates have the conjectured

    sign, we only obtain a significantcoefficient for the lack of threat of entry variable. Thisfinding suggests hatwhen firmsface lesspressure rom competition; they become relativelyless cost-efficient. Therefore, we find limited empirical support for our conjecture C2.In C3 we speculate that high initial market shareposition leads to reduced cost benefitswhen firms also have power over their suppliers and buyers. Empirically, we find theopposite result. We find negative coefficients on both power over buyers (significant)and power over suppliers(insignificant). This finding, in conjunction with the null main8 No main effect estimate for power over supplierswas obtained since none of the variablesmeasuringdifferent

    aspects of this force vary over time in the PIMS database.

  • 8/3/2019 Environment, Market Share, and Market Power

    15/19

    ENVIRONMENT, MARKET SHARE AND MARKET POWER 1173TABLE 4

    Average Cost EquiationEstirnatesabLMP ZA$MSVariable Mean SD Main Effect Interaction Interaction

    Market Position 1.144LMP 0.779 10.006 OcLMP2 1.916 0.633 OczA$MS 0.323 -0.2144***(0.0740)zA$MS2 0.506 0.771 -0.0024(0.0104)Firm PositionPQ 27.450 27.530 3.9E - 4*

    (2.9E - 4)SF 3.032 0.985 6.1E - 4(6.6E - 3)AD 2.777 1.064 0.0027(0.0063)PROMO 2.928 0.982 -0.0069(0.0064)%Utilizationd 74.881 17.777 -8.OE - 4***(2.3E - 4)ForcesPOB 10.664 2.263 0.0030 -0.0052*** 0.0032(0.0043) (0.0017) (0.0041)POS 4.176 1.301 -0.0047 0.0076

    (0.0063) (0.0127)LCR 8.570 2.479 0.0020 5.9E - 4(0.0020) (2.4E - 3)LTE 2.598 1.518 0.1220*** 0.0534***(0.0086) (0.0085)

    a Equation R2 = 0.223.bStandard errorsin parentheses.cDropped from model after unable to reject the null hypothesis that the LMP main effects coefficients equalzero.d We remove the capacity utilization variablefrom the competitive rivalry measure. In this equation capacity

    utilization appearsto pick up the effect of spreadingfixed costs over a largervolume. Empirically,the capacityutilization variablereceives a differentsign from the other variablesincluded in the competitive rivalrymeasurewhen regressed against average cost.* Significant at the 0.10 level.*** Significant at the 0.01 level.

    effects finding, suggeststhat firms only benefit (in the form of statistically significant costreductions) from initial market share position when in a favorable position relative tobuyers. Only in this instance does market share translate into market power benefits.C4 states that market share changes come with different associated costs in varyingcompetitive environments. Our resultssuggestthat market share increases associate withless cost reduction in easy environments. All of the coefficients on the industry forces-market share change interactions are positive, although only the coefficient on the lackof threat of entry interaction term is significant. Thus, again we find limited support forthe conjecture of less cost-consciousness in easy environments.

    7. DiscussionThe overarching question we addressin this paper is the extent to which market shareaffiectsfirm profits. Toward this end we develop and empirically test a model of how

  • 8/3/2019 Environment, Market Share, and Market Power

    16/19

    1174 WILLIAM BOULDING AND RICHARD STAELIN

    market share influences the two components of firm profit, total revenue and total cost,through its influence on average price and average cost. Our modeling approach providesboth methodological advances as well as new insights into the managerial implicationsof market share. We discuss these below.Methodology

    We believe our methodological approach representsan important contribution, sinceresearchers now have a means of obtaining dollar value price and cost measures fromthe PIMS databasein addition to the ratio measures typicallyutilized (e.g., relativeprice).We use this approachto explicitly estimatethe effects of both monopoly and monoposonymarket power.Our empirical model utilizes a fixed effect estimation technique that makes prices andcosts comparable across the different business units in our sample, thereby allowing usto analyze firm specific data from different industries. We control for different industrystructuresby characterizingan industry in terms of the environmental forces proposedby Porter (1980). Such an approachmakes it possible to disentangle firm-specificmarketshare effects from industry structure effects on profits.Ourtheoreticalmodel also acknowledgesthe legitimateconcerns of previousresearchersthat unobserved variablesaffect both market share and profitability Jacobson and Aaker,1985, Rumelt and Wensley 1980). Since we control for both fixed and time-variantunobservables correlated with the endogenous variables in our process model, we cantest whether economic rents accrue to market share itself, rather than these unobservablefactors (e.g., firm-specificassets or luck). In this way our results provide insight into themarket power-efficiency debate (e.g., see Demsetz 1973 and 1974).Managerial Implications

    Our modeling approach also allows us to formally address the question of "is moremarket share better?"By substituting equation (3) into equation (1 a), and (4) into (l b),we obtain expressions for total revenues and total costs. This allows us to examine notjust the price-cost margin, where we found no direct effects of market position, but alsofirm profits, which reflect the volume effects of market share. We solve for the marginalrevenue product and marginal factor cost of market share.9Equating marginal revenuesand marginal cost provides the optimal firm market share solution. If the premise thatmarket share is good is correct, then more market share should be even better, implyingan optimal market share of 100%for each firm.The optimal firm market share depends on current firm and environmental charac-teristics. Holding fixed these values, we solve for the optimal market share for everybusiness unit in our sample. Figure 1 shows that over half of the business units have anoptimal market shareof less than 100%.However, the largenumber of "100%" olutionsindicates that some firms can improve profits via increases in market share.The firms estimated to make more money with an increase in market share operatein good environments (i.e., have power over buyers, power over suppliers, lack of com-petitive rivalry, and lack of threat of entry). It appears that these firms can increasevolume without an offsetting reduction in the price-cost margin. Thus, even though wefind market share per se does not provide market power in terms of higher prices orlower costs (see Tables 3 and 4), our results indicate that firms operating in good envi-ronments obtain benefits from market share increasesvia volume increases without con-comitant margin reductions.These findings must be stronglytempered by two related caveats that suggesta strongupward bias in our optimal market share results. First, if a firm in a good environment

    9Copiesof these solutions are available from the authors upon request.

  • 8/3/2019 Environment, Market Share, and Market Power

    17/19

    ENVIRONMENT, MARKET SHARE AND MARKET POWER 1175504540

    U 35C 30C)4 25

    20; 15

    1050

    1 10 20 30 40 50 60 70 80 900 to to to to to to to to to 'to 100 Other"*9 19 29 39 49 59 69 79 89 99Optimal Market Share

    *For the sample of 340 business units.** Multiple solutions obtained.

    FIGURE 1. Distribution of Optimal Market Shares.

    noticeably increases market share, the collusive solution breaks down and the economicrents associated with the industry structure may dissipate. Somewhat related, our opti-mality analysis assumes environmental factors remain fixed. However, as a firm movesto 100%market share our environmental measures reflecta less attractive environment.For example, exits must occur as the firm's market shareapproaches 100%.This in turnmeans that the average price after the increase in market share would be less since theenvironment is less friendly. Thus, a firm will benefit less than our analysis indicates,unless the firm can simultaneously increase market share and maintain market powervia some mechanism such as product differentiation.With respect to long run market share position, we find that, in general, market shareper se does not lead to monopoly rents. Since considerable evidence exists that marketshare is positively associated with profitability (e.g., Buzzell et al. 1975), this previouslyobserved relationship is probably due to managerial efficiency, returns to innovation orrisk, and other unobservable (to the researcher)factors that correlatewith both marketshare and firm profits (e.g., see Jacobson and Aaker 1985). However, we do find thatindustry structureand short term changes in market share strongly influence prices andcosts. In particular,we observe that firms in adverse competitive environments exhibitmore cost efficiency but also must lower price more to obtain share increases. We takethese results to imply that in terms of government intervention, antitrust enforcementshould not focus on higher share firms per se (for a specific example of this conclusionsee Fisher et al. 1983). Instead, if intervention is deemed socially desirable, this inter-vention should occur at the industry level.Future Research Implications and Implementation

    In setting strategy for the firm, the key issue is not market share, per se, but how toincrease market share to get volume benefits while maintaining economic rents. A firm

  • 8/3/2019 Environment, Market Share, and Market Power

    18/19

    1176 WILLIAM BOULDING AND RICHARD STAELINcould do this in one of two ways. First, a firm might be able to increase market sharewithout worsening the competitive environment, thus maintaining the rents accruingtoindustry structure. Such an approach suggests future work aimed at assessingthe natureof competitive reaction/rivalry. For example, one might define conditions under whichcompetitive response is minimized, or identify techniques available to firms (e.g., sig-naling) that minimize competitive response.Second, if industry rents dissipate when a firm chooses to increase market share, thefirm could substitute firm-specific rents for lost industry rents. This suggests future workassessing firm-specific specialized assets (i.e., inimitable assets) that enable profitablemarket share increases. We control for these unobservable factors in this paper, but donot address them directly. Future research should attempt to study questions such aswhat particular management skills enhance a firm's ability to gain market share, and,how long can a firm maintain its specialized assets?

    Our results emphasize the strategic importance of industry structure. Firms shouldcarefully choose and monitor their industry environment. In this regard, we find thePorter frameworkvery useful. We identify variablesfrom the PIMS database that measuredifferent aspects of Porter's forces. However, because of the strategic importance of in-dustry structure, and a lack of significance for the coefficients on some of our environ-mental measures (e.g., the insignificant effect of power over buyersin the price equation),we recommend future empirical work that focuses on obtaining relevant measures ofindustry structure and assessing their effects on profits.Finally, our methodological approach should enhance future research opportunitiesusing PIMS data, although implementation issues exist. For example, the process weemploy for controlling for unobserved fixed effects (e.g., managerial effectiveness) pre-cludes us from directly obtaining parameterestimates for observed fixed variables. How-ever, as shown by Hausman and Taylor (1981) and others (Amemiya and MaCurdy1986, Breusch et al. 1989), techniques exist for obtaining coefficient estimates on measuredtime invariant variables while controlling for unobserved fixed effects. Also, fixed effectestimation can increase the impact of measurement error. This errorsin variables(EIV)problem is one reason we instrument market share, our variable of theoretical interest.More generally, Griliches and Hausman ( 1986) present numerous solutions to the EIVproblem in the context of fixed effect estimation.

    Even if our basic approach sometimes requires more complicated estimation proce-dures, we believe the effort s well worth the cost. The advantagecontainedin ourapproach,which yields dollar metric measures, versus the more readily available ratio variablessuch as prices and costs relative to competition, is the ability to study the effects ofstrategicchoices in an equilibrium framework consistent with economic theory.'0'1 The authors gratefully acknowledge the Business Associates Fund at the Fuqua School of Business, TheStrategic Planning Institute, and The Wharton PIMS Research Center for their research support. We thankseminar participantsat Columbia University, Duke University, University of Arizona, Washington University,Stanford University, University of Chicago, and an anonymous reviewer for their helpful comments. Specialthanks are due to Mike Moore for his insights.

    ReferencesAMEMIYA, T. AND T. E. MACURDY, "Instrumental Variable Estimation of Error-Components Model," Econ-oinetlica, 54 (1986), 869-881.ANDERSON, CARL R. AND FRANT T. PAINE, "PIMS: A Reexamination," Acadcleinvf AanagenewntRe'., 3(1978), 602-612.ANTERASIAN, CATHY AND LYNN W. PHILLIPS, "Discontinuities, Value Delivery and the Share-Returns As-

    sociation: A Re-examination of the 'Share-Causes-Profits' Controversy," Marketing Science InstituteReport No. 88-109, 1988.BREUSCH, TREVOR S., GRAYHAM E. MIZON AND PETER SCHMIDT, "Efficient Estimation Using Panel Data,"Economnetrica, 7 (1989), 695-700.

  • 8/3/2019 Environment, Market Share, and Market Power

    19/19

    ENVIRONMENT, MARKET SHARE AND MARKET POWER 1177BUZZELL, ROBERT D. AND BRADLEY T. GALE, The PIAIS Principles, The Free Press, New York, 1987.

    5 AND RALPH G. M. SULTAN, "Market Share-A Key to Profitability,"Halvard Business Rev.,53 (1975), 97-106.CARPENTER, GREGORY S., LEE G. COOPER, DOMINIQUE M. HANSSENS AND DAVID F. MIDGLEY, "ModelingAsymmetric Competition," Marketing Sci., 7 (1988), 393-412.DAY, GEORGE AND DAVID B. MONTGOMERY, "Diagnosing the Experience Curve," J. Marketing, 47 (1983),44-58.DEMSETZ, HAROLD, "Industry Structure, Market Rivalry, and Public Policy," J. Lawl}nd Economics, 16

    (1973),1-9."Two Systems of Belief About Monopoly,' in H. J. Goldschmid, H. M. Mann, and J. F. Weston(Eds.), IndlustriialConcentrattioni: he NewvLearnoing,Little, Brown and Company, Boston, 1974.

    FISHER, FRANKLIN M., JOHN J. MCGOWAN AND JOEL E. GREENWOOD, Folded, Spincdled,ndMu1tilated:Economic AnaNlsisand U.S. vs. IBM, MIT Press, Cambridge, MA, 1983.FRUHAN, WILLIAM E. JR., "Pyrrhic Victories in Fights for MarketShare," Harvard Biusince.ssev., 50 (1974),412-423.GENERAL ELECTRICCOMPANY, General Electric Monogram, September-October, 1981.GRILICHES, ZVI AND JERRY A. HAUSMAN, "Errors in Variables in Panel Data," J. Econiomnetrics, 31 (1986),93-118.HAGERTY, MICHAELR., JAMES M. CARMAN AND GARY J. RUSSELL,"Estimating Elasticities with PIMS Data:Methodological Issues and Substantive Implications," J. MalrlketingRes., 25 (1988), 1-9.HANSSENS, OMINIQUE ., "MarketResponse, Competitive Behavior and Time SeriesAnalysis,"J.Marlketit1gRes. 17 (1980), 470-485.HAUSMAN, JERRYA. AND WILLIAME. TAYLOR,"Panel Data and Unobservable Individual Effects," Econometric,49 (1981), 1377-1398.JACOBSON, OBERT, Unobservable Effects and Business Performance," Marletin1gSci.,forthcoming.

    , AND DAVID A. AAKER, "Is Market Share Really All that It's Cracked Up to Be?," J. Aqrll.keting,9(1985),11-22.MARTIN, STEPHEN, "Market Power and/or Efficiency," Rev'.qf Economics and Statistics, 70 (1988), 331-335.MONTGOMERY,CYNTHIA A. AND BIRGER WERNERFELT,"Sources of Superior Performance: Market ShareVersus Industry Effects in the U.S. Brewing Industry," Northwestern Univ. Working Paper, 1989.MOORE, MICHAEL J. AND WILLIAM BOULDING, 'Economic and Econometric Analysis of Disguised Data,Fuqua School of Business, Duke Univ. WorkinigPaper, 1987.PORTER, MICHAEL, Competitive Strategy, The Free Press, New York, 1980.RUMELT, RICHARDAND ROBIN WENSLEY, "In Search of the Market Share Effect, Working Paper MGL-61,Univ. of California, Los Angeles, 1980.SCHROETER, OHNR., "Estimating he Degreeof MarketPowerin the Beef PackingIndustry," Rev. o EconomicsanidStaltistics, 70 (1988),158-162.STATEN, MICHAEL,JOHN UMBECK AND WILLIAMDUNKELBERG, "Market Share/Market Power Revisited,"

    J. Healtli Economics, 7 (1988), 73-83.STRATEGIC PLANNING INSTITUTE, The PIMS ProgramData Forin, Thie Strategic Planniiing Institute, Cambridge,MA, 1987.The PIMS Competitive Strategj' Rcessearchata Base, Thie Strategic PlanninigInstitute, Cambridge,MA, 1988.

    WERNERFELT, BIRGER, "The Relation Between MarketSlhareand Profitability, J. Butsinecsstrlateg,, (1985),67-74.