Strategic Management Journal, Vol. 19, 405–411 (1998)

COMMENTARY ON ‘ENTRY INTO NEW MARKETSEGMENTS IN MATURE INDUSTRIES:ENDOGENOUS AND EXOGENOUS SEGMENTATIONIN THE U.S. BREWING INDUSTRY’ BYA. SWAMINATHAN

WILL MITCHELL*University of Michigan Business School, Ann Arbor, Michigan, U.S.A.

Anand Swaminathan’s paper, ‘Entry into newmarket segments in mature industries: Endogen-ous and exogenous segmentation in the U.S.brewing industry’, offers three contributions.First, the focal purpose of the paper is to testcompeting ecological predictions about how newindustry segments arise. Second, the implicationsof the conceptual argument and empirical analysishelp us understand how the interplay betweenexogenous environmental change, endogenousindustrial change, and firm capabilities shapesindustry evolution. Third, the study provides anexample of how to reach generalizable con-clusions about the effects of idiosyncratic firmcapabilities. The first contribution provides a use-ful bridge between organizational ecology theoryand strategy research, while the second and thirdcontributions are fundamentally important inshaping our understanding of business strategyand our approaches to strategy research.

The primary focus of the paper is to compareniche formation and resource partitioning, whichare two ecological theories about industry seg-mentation, as alternative predictors of entry ratesto new industry segments in mature industries.Niche formation theory emphasizes exogenouschanges such as new consumer preferences,regulatory regimes, and technology as the driversof segmentation. In the niche formation view,the size of a new exogenously created segment

Key words: exogenous and endogenous industrychange; disequilibrium

* Correspondence to: Will Mitchell, University of MichiganBusiness School, 701 Tappan St., Ann Arbor, MI 48109-1237, U.S.A.

CCC 0143–2095/98/040405–07 $17.50 1998 John Wiley & Sons, Ltd.

determines the rate of new entry into the segment.Larger new segments will tend to have more newentrants, where sales revenue or unit sales offerpossible measures of segment size.

By contrast with the niche formation view,resource partitioning models argue that newindustry segments often arise as the result ofendogenous changes within the context of existingcompetition in an industry. In this view, incum-bent firms tend to grow by undertaking the mostgeneral and highest volume activities in an indus-try. By doing so, the incumbent firms leavespecialty segments for new entrants to the indus-try. With an argument that closely parallels thedevelopment of current ideas in the strategy litera-ture concerning firm-specific business capabilities,resource partitioning theory argues that the speci-alized niches arise because the commercial capa-bilities of the generalists do not suit many special-ized activities. A core underlying assumption ofthis view is that opportunities for new entrantsin a mature industry occur because the generalcapabilities that suit the mature market do not fitthe narrow needs of more specialized marketsegments, rather than because of managerialmyopia of the incumbent firms. Indeed, the proc-ess of creating general capabilities may activelyconflict with attempts to create specialized capa-bilities. In this view, it is the endogenous processof resource partitioning within a mature industrythat creates the incentives for firms to enter newlyemerging segments. The greater the degree ofpartitioning, which ecologists often measure interms of industry concentration, the greater therate of new entry into new segments.

Thus, the niche formation and resource par-

406 W. Mitchell

titioning ecological views provide different per-spectives on entry to mature industries and, inturn, on industry evolution. If new market seg-ments arise as the result of environmental changesoutside the context of existing competition in anindustry, then segment sales potential will bea strong predictor of entry into new segments.Alternatively, if new market segments arise asthe result of changes in the capabilities of firmswithin the context of existing competition, thenindustry concentration will be a strong predictorof entry rates into new segments. The nicheformation view is akin to a punctuated equilib-rium view of industry evolution, in which externalchanges can disturb a stable industry. Theresource partitioning view, by contrast, posits anongoing disequilibrium, in which an industrynever stabilizes.

The paper tests the two views in the contextof the U.S. brewing industry, by examiningentries into the microbrewery and brew pub seg-ments of the industry. The data include 335microbrewery foundings from 1939 to 1995, plus588 brew pub foundings from 1982 to 1995.This data set includes almost the full history ofmicrobrewery and brew pub foundings in theUnited States including all 50 states plus theDistrict of Columbia. The focal dependent vari-able is the number of foundings in a given stateduring a given year. The focal independent vari-ables are the volume of imports and beer industryconcentration. The paper defines both import vol-ume and industry concentration in terms of thetotal beer industry, including mass producers aswell as brew pubs and microbreweries. Importvolume provides a measure of exogenous changein consumer tastes for beer. If niche formation isthe primary explanation for industry segmen-tation, then entry into the microbrewery and brewpub segments will tend to rise with import vol-ume of imported beer. Industry concentration pro-vides a measure of endogenous change in thecompetitive conditions of the beer industry. Ifresource partitioning is the primary explanationfor industry segmentation, then entry into themicrobrewery and brew pub segments will tendto rise with industry concentration across the totalbeer industry. The analysis also controls for keyalternative and ancillary influences on entry rates,including potential market size, legitimizing andcompetitive influences of competitive density,institutional support in the political environment,

1998 John Wiley & Sons, Ltd. Strat. Mgmt. J.,Vol 19, 405–411 (1998)

and annual production output of firms in the newsegments. Several of the control variables haveimportant results, which I will address later inthis comment.

The empirical analysis in the paper, which Isummarize in Table 1 of this comment, supportsthe exogenous niche formation argument forindustry segmentation, while conflicting with theresource partitioning view. In both the micro-brewery segment and the brew pub segment, vol-ume of imports is a significant predictor of entry(row 10 of Table 1), while beer industry concen-tration has no significant effect on entry to eithernew segment (row 11 of Table 1). This compara-tive result contributes to the development of eco-logical research, by providing analysis that eco-logical researchers can test and compare withinan existing cumulative body of empirical research.

One strength of this paper is that it presentsecological reasoning in a language and style thatare relevant to strategy researchers. Table 1 inthis comment attempts to further clarify and inte-grate the use of ecological and strategic language.The strategic aspect of the contribution is partic-ularly important given the unfortunate tendencyof much ecological research to cloud its coreideas in theory-specific terminology and structure.The obscurity of ecological language sometimeshides the fact that many of the core ideas ofecological research directly apply to understand-ing industrial change and business strategy andperformance. Indeed, strategy researchers cangather substantial insights from behind the veilof ecological research. This paper helps pushaside some of that veil of terminology.

An immediate pay-off of the clarity is thatAnand identifies an intriguing possible interactionbetween industry-level processes, which are thefocus of ecological theory, and individual decisionmaking, which is often deemphasized in ecologi-cal research while being a key part of strategicanalysis. He finds that initial microbrewery den-sity has a lesser legitimizing influence in statesin which brew pubs are legal. Row 9 of Table 1in this comment summarizes the result. A partialexplanation for the result may simply be thatlegitimacy deriving from competitive presence isless necessary in states in which brewing legit-imacy partially derives from legal sanction. Anadditional explanation that Anand presents,though, is that potential entrants will quicklyswitch from considering microbrewery entry as

Commentary 407

Table 1. Summary of empirical results in Tables 3 and 4 of ‘Entry into new market segments in matureindustries’ (Anand Swaminathan)a

Variable in Table 3 (Table 4) Strategy terminology (1) (2)Microbrewery foundings Brewpub foundings

1939–95 1982–95Table 3, Model 6 Table 4, Model 11

1 State per capita annual personal Potential market size (+) .income

2 State per capita annual beer Potential market size . .consumption

3 Percentage of state population Potential market size . .living in dry areas

4 Brewpub legality Legal conditions +5 State microbrewery (brewpub) No. of direct competitors +,− +,−

density, density square6 Out-of-state microbrewery Regional trends + (+)

(brewpub) density7 State brewpub (microbrewery) No. of producers of . +

density complementary goods8 National mass producer density No. of producers of − .

substitute goods9 Brewpub legality× state Complement −

microbrewery density availability× directcompetition

10 Niche formation variable; Exogenously determined + +volume of imports segment sales potential

11 Resource partitioning variable: Endogenous creation of . .industry concentration specialized segments

aSigns indicate statistically significant results; signs in parentheses are borderline significant.

competition increases if they have the choice ofconsidering brew pub entry as an alternative.Thus, population- and firm-level processes mayintertwine in a generalizable fashion.

Well beyond the simple clarification of ecologi-cal and strategic language and the demonstrationthat industry- and firm-level processes can coexistwithin an ecological approach, the paper providessubstantial insights concerning business strategy.Strategy researchers will not be surprised to dis-cover that exogenous changes in industrialenvironments have major influences on industryentry. We have long documented that entrantsrespond to changes in environmental factors suchas new consumer tastes, technological change,and regulatory change. There is a risk, though,in leaving the implications at that first stoppingpoint, that is, in concluding that exogenouschange is the dominant explantion for indsutryevolution and that endogenous change plays atmost an ancillary role. While this conclusionmight attract those who prefer exogenously deter-mined equilibrium models of industry evolution,

1998 John Wiley & Sons, Ltd. Strat. Mgmt. J.,Vol 19, 405–411 (1998)

the simple reading of the empirical results ofthis paper obscures a deeper set of endogenouslyinfluenced disequilibirum processes.

Let us start by examining the effect of severalother variables in the analysis, which Table 1 alsosummarizes. I am particularly intrigued by theresults concerning the number of producers ofsubstitute goods (row 8), the number of directcompetitors (row 5), and the number of producersof complementary goods (row 7). Singly andjointly, these results suggest a process of exogen-ous and endogenous industry evolution.

Row 8 of Table 1 demonstrates a first inter-action between exogenous and endogenous indus-try evolution. Here, we find that microbreweryfoundings decline with number of producers ofsubstitute goods (column 1), that is, the numberof mass brewers, while brew pub foundings haveno relationship with mass producer density(column 2). It is reasonable to assume that beerproduced by mass producers is a closer substitutefor microbrewed beer than for the output of abrew pub, because a brew pub serves as a res-

408 W. Mitchell

taurant and entertainment site as well as a beerproducer. Consistent with this assumption, then,the presence of many producers of substitutegoods and, by further assumption, with a greatervariety of substitute goods, discouraged the entryof new firms into the new segment. Thus, theresults suggest an endogenous process in whichcompetitive conditions within the core industryinfluence entry rates into an exogenously creatednew segment of close substitutes. Even if a newindustry segment arises as a result of exogenouschanges in taste and technology, then, the stateof the existing industry at the time of the exogen-ous change will influence the rate at which newentrants respond to the change.

The second interaction arises in row 5 ofTable 1, which summarizes the influence of thevariables that deal with legitimization and compe-tition. These variables show a common ecologicalinfluence, in which increasing competition firstcreates legitimacy for a segment and encouragesentry and then inhibits further entry once thenumber of competitors reaches some critical mass.This, in itself, is an endogenous process in whichcompetition both encourages and inhibits entrywithin even an exogenously created new segment.

The third interaction arises in row 7 of Table 1,which shows that microbrewery foundings influ-ence brew pub foundings (column 2), but thatthe reverse relationship does not hold (column1). If we view microbreweries and brew pubs asproducers of complementary goods, rather thandirect competitors, this asymmetric result hasintriguing implications. The assumption of com-plementary is reasonable, as people who drink inbrew pubs are likely to be particularly commonpurchasers of microbrewery beer. The asymmetricinfluence of microbrewery density on brew pubfounding rates most likely stems from anendogenous temporal process, where the endogen-eity arises at the fringes of the mass industry andoccurs as one new industry segment leads to thegrowth of a second new segment. Recall thatmicrobreweries emerged before brew pubs. Asmicrobreweries diffused through the variousstates, the results then suggest that the diffusionled to the founding of an even more specializedform of beer consumption, the brew pub. Thelikely causality here is that beer drinkers whocame to like the microbrewery product gained anincreased taste for similar products sold in brewpubs. More generally, this result suggests a proc-

1998 John Wiley & Sons, Ltd. Strat. Mgmt. J.,Vol 19, 405–411 (1998)

ess in which the exogenous creation of one newmarket segment leads to the endogenous creationof a complementary new market segment. Thus,the results of influences that at first seem to besimply control variables help shape our under-standing of interactions between exogenous andendogenous industrial change.

Together, these results suggest a set of threeinterrelated evolutionary processes. First, the stateof competition in the core industry influencesentry into a new segment. Second, entry into thenew segmment first encourages and then discour-ages further entry into the segment, with thelegitimizing effect ending sooner when there arealternative segments in which to consider entry.Third, the growth of the new segment encouragesentry into a complementary new segment. Thus,an erroneous initial conclusion that exogenousinfluences dominate would obscure a more com-plex interaction of exogenous and endogenousindustry evolution. The existence of such an inter-related process should ring true to strategyresearchers who study how firms and industrieschange over time, and are well familiar with theway in which unanticipated events can set offchain reactions of subsequent events within anindustry. The pattern of the results helps gen-eralize our expectations of how these chain reac-tions will occur.

In addition to the insights concerning inter-actions between exogenous and endogenousindustry-level evolution, the results hint at issuesconcerning business-level capabilities. I am againintrigued by the complementary producer resultsin row 7 of Table 1, which raise a business-level question in addition to their implicationsfor understanding influences on segment-levelentry rates. The question is why the presence ofmicrobreweries encourages the entry of brewpubs, which typically are not owned by themicrobreweries, rather than the diversification ofthe microbreweries into the brew pub segment.Indeed, this question also arises at an earlier stageof industry evolution, in the sense that the massproducers for the most part left the new micro-breweries to develop the microbrewery segment,rather than diversifying themselves.

Two possible explanations for the prevalenceof new entrants in new market segments cometo mind, stemming from very different behavioralprocesses. One possible explanation for the pre-dominance of new entrants is that the mass pro-

Commentary 409

ducers suffered ongoing mass myopia concerningthe potential of the microbrewery market, and thatthe microbrewers then suffered similar myopiaconcerning the brew pub opportunity. This firstexplanation is not especially satisfying, especiallygiven the decades-long duration of the positedblindness. At the very least, the explanationwould suggest that we have been doing a partic-ularly bad job when we try to teach businessstudents, who then become business managers,about how to recognize and respond to changes intheir competitive environments. A second possibleexplanation is that the new segments requiredfirms to have very different business capabilitiesthan those needed in the earlier mass productioncore of the industry and, later, in the micro-brewery segment. In the second view, the estab-lished firms tended to avoid the new segmentsbecause they did not fit in them, rather thanbecause they did not see the changes in themarket. This second explanation seems morelikely than the first, particularly when we coupleit with the observation that each established seg-ment of the market continued to offer substantialsales and growth opportunities as each new seg-ment emerged. Thus, established firms had incen-tives to concentrate on adapting and strengtheningwhat they already knew how to do, rather thandiversify into market segments that required dif-ferent capabilities and knowledge.

Recent trends in the discount retail marketscorrespond to the industry and firm-level patternsthat this study of the beer industry suggest. Themass discount retail market became establishedduring the late 1960s and 1970s in the UnitedStates, with firms such as Kmart, Walmart, andTarget playing leading roles. More specializeddiscount retail market segments then arose duringthe late 1970s and 1980s, in product areas suchas hardware, sporting goods, books, and officesupplies. The first new specialized discount seg-ments arose at least in part due to exogenouschanges in consumer taste, mall growth, roadconstruction, and suburbanization. In turn, though,the success of the earlier specialized discountretail segments, such as hardware, provided mod-els and gave impetus to the birth and growth oflater complementary segments, such as books.Notably, few of the mass discount retailersexpanded into the specialty discount retail seg-ments, judging that the specialized segmentsrequired different skills than those they needed

1998 John Wiley & Sons, Ltd. Strat. Mgmt. J.,Vol 19, 405–411 (1998)

to compete in the mass discount core of theindustry. Equally notably, the one mass discountgeneralist that undertook a major expansion intospecialized discount retailing, Kmart, failed andhad exited specialty retail by 1997 after incurringsevere losses. The mass discount retailers thatfocused on adapting to changes within the coreelements of the mass discount market, such asWalmart’s development of multiline super stores,were much more successful. Although the evolu-tion of the beer and discount retail industriesdiffer on many details, there is an intriguing coresimilarity on these dimensions.

Before concluding this note, I would like tocomment briefly on a more general issue concern-ing this fit between this paper and trends instrategy research. One recent emphasis in strategyresearch is to consider strategy as an interactionof multiple causes at multiple levels of analysis.In part, this tendency mirrors trends in disciplinesthat closely relate to strategy, such as economicsand organizational theory. In part, though, re-searchers working within the tradition of strategicmanagement may be uniquely able to undertakesuch multilevel and multicausal research.

A historic strength of the strategic managementfield has been that strategy researchers have beenwilling to study subjects that cut across existingconceptual boundaries. The fundamental questionsof strategy research, concerning why businessesact and how those actions affect business per-formance, involve multiple causes. The causesare often idiosyncratic to particular businesses.They often involve many levels of analysis,including individual managers, intraorganizationalgroups, organizational routines, business andcorporate units, interorganizational businessalliances and social relationships, vertical andhorizontal industry relationships, market bound-aries, and societal forces. This disorderlinessgrates on many research norms and, perhaps asa result, researchers in many established fieldshave shied from addressing these kinds of issues,preferring instead to focus on small elements ofbusiness strategy that stem from more discretecauses. Strategy research, by contrast, has beenwilling to take on the complexity of business,sometimes within individual studies and moregenerally across the field of strategy research.

In some sense, one can liken the phenomenonof business strategy to an apple, which is delight-ful to look at and taste, and results from a

410 W. Mitchell

complex interweaving of sun, soil, color, scent,texture, and flavor. Many research traditions viewthe strategy apple as too complicated a subjectto study as a whole. Rather than study the apple,one can instead choose to study its seeds, or itsscent, or the tree on which the apple matures, orthe soil in which the tree grows, or the orchardistwho grows the tree, or the people who pick thefruit. Such reductionist approaches are necessary.The studies and fields often provide precisecharacterizations of their subjects and, in turn,we unquestionably learn much of value about theelements that define the context of the apple. Arisk of relying only on reductionist approaches,though, is that we may forget that we are studyingapples or, at least, that we may lose sight of thecomplex interactions that make up the apple asa whole. A deeper risk, which has a centralimpact on our conceptual understanding of theobjects of our research, is that we may mischarac-terize the whole by studying only the parts. Forinstance, an apple will not reproduce itself fromits own seeds. That is, if one plants the seeds ofan apple of which one is particularly fond, a treebearing the same type of apple will not emerge.If a tree grows at all, it may bear wizened fruitor no fruit, or very occasionally, a delightfulproduct that bears little or no resemblance to itsprogenitor. It one wants to reproduce a particulartype of apple, one must graft its branches toanother tree, rather than plant the seed that grewwithin the piece of fruit. Indeed, the most suc-cessful reproduction of apples often involvesusing root stock from different types of fruitaltogether, such as plum trees. Moreover, theappropriate rootstock varies drastically dependingon local temperature, light, and soil. Thus, if onestudies only the seeds of an apple or the tree onwhich a particular apple grows, one can reachlittle understanding of how that apple came intobeing and even less understanding of how toreproduce it. The strength of the strategy researchliterature is that it has been willing, as a field,to study the complex and disorderly phenomenonthat is the strategy apple.1

At the core, then, a grounding in strategicmanagement is a reflection of the idea that oneneeds something other than extensions of tra-ditional theories of market and social environ-

1 I appreciate Rebecca Henderson’s comments concerning anearlier version of this analogy.

1998 John Wiley & Sons, Ltd. Strat. Mgmt. J.,Vol 19, 405–411 (1998)

ments, social organizations, or individual be-haviors to understand the determinants of firmperformance. Instead, one needs perspectives thatfocus on how the intersections among individual,organizational, and environmental incentives pro-duce the entities that we recognize as businessorganizations and the activities and performancethat we recognize as business strategy.

Nonetheless, despite the real strength and valueof the strategy literature, its disorder is sometimesa weakness. Perhaps most centrally, it is some-times difficult to generalize beyond the boundsof a single research study or a small set ofstudies because of fundamental differences in thestructure of strategy research studies. As a result,to introduce another analogy from the physicalsciences, many well-crafted and insightful piecesof research become single beacons casting beamsof light through lenses with individual shapes,focal lengths, and filters. The resulting kaleido-scope of overlapping colors offers observers awonderful pastiche of images and may provideprecise characterizations of individual businesscircumstances. The kaleidoscope of images,though, sometimes provides too little of a generalrecognizable picture. The field of strategyresearch would offer clearer contributions if thefield provided more generalizable methods andresults that help us understand why businessesundertake actions and how the actions affect busi-ness performance outside the precise context ofindividual research studies.

The challenge for the field of strategy research,then, is to increase its generalizability whileretaining its willingness to address the disorderlymultilevel and multicausal phenomena of businessactivity and performance. This is a substantialchallenge, to put it mildly. Meeting the challengewill require many conceptual arguments andmany sets of research studies.

Anand Swaminathan’s study of segmentationin the brewing industry offers an example of oneapproach to this challenge. The study has anunderlying causal argument about firm capabilitiesthat closely aligns with many current argumentsin what is becoming known as the resource-basedview of the firm. In addition, though, the studysuggests some of the ways in which such idio-syncratic firm-specific capabilities can interactwith industry-level factors to influence businessactivity and industry evolution. In doing so, thestudy offers one example of how researchers

Commentary 411

can generalize the implications of firm-specificcapabilities for business strategy.

This study provides specific insights concern-ing the process of industry evolution and auseful example of how to carry out generaliz-able research about idiosyncratic firms. Theinitial results of the paper support the viewthat industry segmentation often results fromexogenouschanges in factors such as tastes, tech-

1998 John Wiley & Sons, Ltd. Strat. Mgmt. J.,Vol 19, 405–411 (1998)

nology, and regulation. Further analysis of theresults then reveals a more complex process inwhich exogenous environmental change, endogen-ous industrial change, and firm capabilities inter-act to shape industry evolution jointly. The resultsand the underlying arguments mesh nicely withcurrent strategy research, while providing concep-tual and empirical approaches to extending thatresearch.