entry into new market segments in mature industries: endogenous and exogenous segmentation in the...

Strategic Management Journal, Vol. 19, 389–404 (1998)

ENTRY INTO NEW MARKET SEGMENTS INMATURE INDUSTRIES: ENDOGENOUS ANDEXOGENOUS SEGMENTATION IN THE U.S.BREWING INDUSTRY

ANAND SWAMINATHAN*University of Michigan Business School, Ann Arbor, Michigan, U.S.A

I evaluate two processes, niche formation and resource-partitioning, that could independentlyaccount for the entry of firms into new market segments in mature industries. The nicheformation argument focuses on environmental changes that promote the entry of new firmswhereas the research-partitioning argument is based on the internal differentiation of a matureindustry into subgroups composed of specialist and generalists. In other words, the nicheformation and resource-partitioning accounts emphasize forces that are exogenous and endogen-ous to the industry, respectively. I attempt to resolve this theoretical tension by modeling theeffects of niche formation and resource-partitioning together on the founding of firms in themicrobrewery and brewpub segments of the U.S. brewing industry. I find that niche formationprovides a better explanation for both microbrewery and brewpub foundings. In addition, Ifind limited evidence that the process of resource-partitioning is being played out again withinthe microbrewery segment of the industry. Implications for the evolution of organizationalheterogeneity within industries are discussed. 1998 John Wiley & Sons, Ltd.

Strat. Mgmt. J., Vol. 19, 389–404, 1998

Historical records of several organizational popu-lations show that most organizations within themwere founded in brief periods (Stinchombe, 1965;Aldrich, 1979: 177). Stinchcombe (1965: 154)notes that ‘an examination of the history ofalmost any type of organization shows that thereare great spurts of foundation of organizations ofthe type, followed by relatively slower growth,perhaps to be followed bynew spurts, generallyof a fundamentally different kind of organizationin the same field.’ Sociologists explain this punc-tuated pattern in foundings in terms of factorsaffecting the distribution of resources in theenvironment. These include the changing role ofthe state, the development of a market-oriented

Key words: organizational founding, market entry,organizational ecology, entrepreneurship, specialiststrategies* Correspondence to: Anand Swaminathan University ofMichigan Business School, 701 Tappan St., Ann Arbor,MI 48109-1234, U.S.A.

CCC 0143–2095/98/040389–16 $17.50 1998 John Wiley & Sons, Ltd.

economy, increasing levels of urbanization,greater literacy because of better schooling, andpolitical revolution. Traditionally, this stream ofresearch has attempted to identify exogenous fac-tors that affect the emergence of entirely neworganizational populations.

This study departs from the traditionalapproach by examining processes that drive anew spurt in foundings within an existing popu-lation of business organizations, in other words amature industry. Industry maturity is often syn-onymous with a few dominant firms, high entrybarriers and a low rate of entry. However, matureindustries often show a dramatic increase in thenumber of firms. Typically this occurs as a resultof the founding of new kinds of organizationsthat are different from incumbent firms. In theparlance of Hannan and Freeman (1989), thesenew entrants represent new organizational forms—their formal structure, patterns of activity, andnormative order are different from those ofincumbent firms. It is important to develop an

390 A. Swaminathan

understanding of firm entry into new market seg-ments of mature industries due to the prominentrole of entrants in the renewal and growth ofsuch industries (Abernathy and Clark, 1985).

In this paper I examine two alternative expla-nations for firm entry into new market segmentsin mature industries: niche formation andresource-partitioning. The niche formation andresource-partitioning arguments differ in theextent to which they assume that the entry offirms into new market segments in a matureindustry is driven by forces that are exogenous orendogenous to the industry. The niche formationargument emphasizes forces that are largelyexogenous to the industry (Delacroix and Solt,1988). New market niches emerge as a result ofdiscontinuities in an industry’s environment.These discontinuities may reflect changes in tech-nology or consumer behavior (Tushman and And-erson, 1986; Delacroix and Solt, 1988). In eithercase, potential entrepreneurs recognize the oppor-tunities afforded by the formation of the newniche and enter the industry.

The alternative explanation for the entry offirms into new market segments in a matureindustry relies on a process of resource-partitioning (Carroll, 1985). According to theresource-partitioning model, as industries maturethey come to be dominated by a few generalistfirms. These generalist firms attempt to maximizetheir performance by drawing on the largest pos-sible resource space, the center of the market.This opens up pockets of resources on the periph-ery of the market. Entrepreneurs then foundspecialist firms to exploit these peripheralresources. This process generates outcomes thatare consistent with the observation of competitivefringes in many industries (Beesley and Hamilton,1984). Thus the resource-partitioning model laysgreater emphasis on changes endogenous to thegeneralist segment of industries. Movementtowards the center by generalists leads toresource-partitioning which in turn creates con-ditions that facilitate the entry of specialist firms.

The two explanations also imply varyingdegrees of managerial initiative in the foundingof new specialist firms. The niche formation argu-ment attaches a great deal of importance to theability of potential entrepreneurs to respond tothe emergence of a new niche. The resource-partitioning model, however, assumes that adap-tation constraints (Hannan and Freeman, 1984)

1998 John Wiley & Sons, Ltd. Strat. Mgmt. J.,Vol 19, 389–404 (1998)

prevent incumbent firms from moving into thespecialist niche. Instead, this space is likely to befilled by new firms that thrive on the periphery, inpart because they avoid competing with the domi-nant generalist firms. The niche formation andresource-partitioning hypotheses are tested usinglife-history data on the population of U.S. brew-eries in the period 1939–95.

The next section consists of a brief overview oftrends in the post-Prohibition American brewingindustry. I then discuss the niche formation andresource-partitioning models as they apply to theentry of firms into new market segments inmature industries. After this, I describe the dataand the methods to be used in testing the hypoth-eses. Finally, I present the findings and discusstheir implications for the evolution of organi-zational heterogeneity within industries.

THE U.S. BEER BREWINGINDUSTRY: 1933–95

The legality of the American industries producingalcoholic beverages has been subject to questionthroughout history. Several states have imposedProhibition at one time or another, the earliestbeing Maine in 1846. In 1919, 36 states ratifiedthe 18th Amendment to enact national Prohi-bition. This ban lasted until 1933, when it wasrepealed. Most of the breweries founded immedi-ately after Prohibition operated before Prohibition:617 out of the 842 breweries founded in 1933–34 were ‘restarts.’ New firm foundings are alsohigh in these 2 years (53 in 1933 and 172 in1934). Both restarts and new firm foundings fellto an insignificant number after 1950.

The number of breweries declined from amaximum of 933 in 1934 to a minimum of 43in 1981 and 1983. The decline in the number offirms reflects not only brewery failures, but alsothe greater incidence of mergers and acquisitions(Tremblay and Tremblay, 1988). Since the repealof Prohibition, and particularly in the post-WorldWar II period, the American brewing industryhas undergone rapid concentration. The industryfour-firm concentration ratio has risen from 11percent in 1935 to 78 percent in 1982 (U.S.Bureau of the Census, 1982) to 89.7 percent by1995 (Modern Brewery Age, 1996). Mostobservers consider this trend as evidence demon-strating the operation of economies of scale in

Entry into New Market Segments in Mature Industries 391

the industry (Scherer, 1980; Elzinga, 1986).While the number of firms fell from 710 in 1935to 43 in 1983, the total production of the industryincreased from 42 to 178 million barrels over thesame period. Thus the decline in the number offirms does not represent unfavorable conditionsfor the industry, but rather the increasing domi-nation of the industry by a few large generalistfirms (Keithahn, 1978).

The American brewing industry witnessed aspurt of new foundings beginning in the late1970s. This recent burst of foundings has beendriven by the emergence of organizational formsthat are new in the post-Prohibition era—themicrobrewery and the brewpub (Institute for[Fermentation and] Brewing Studies, variousyears; Carroll and Swaminathan, 1992). Bothmicrobreweries and brewpubs produce ale andbeer by traditional processes. Microbrewery prod-ucts are available through regular distributionchannels, whereas brewpub products are availableonly at the site of production. The firstmicrobrewery is recognized to be the New AlbionBrewery of Sonoma, California founded in 1977.The brewpub is even more recent, the first onebeing the Mendocino Brewing Company of Hop-land, California founded in 1983. At last count(January 1996), the number of brewpubs (516)and microbreweries (287) far exceeded the num-ber of generalist mass producers (24) in thebrewing industry.

Carroll and Swaminathan (1992) argue thatmass producers, microbreweries, and brewpubsconstitute separate organizational forms to theextent that they encounter very different environ-ments and respond differently to those distinctenvironments. An examination of common strat-egies followed by these organizational forms sug-gests strong mobility barriers across these threestrategic groups (Caves and Porter, 1977; Mascar-enhas and Aaker, 1989). The mass producer seg-ment of the industry is characterized by strongeconomies of scale in production and advertisingand extensive distribution networks. Themicrobrewers, by contrast, target their productsfor small upscale niches in the market and, atleast initially, are geographically localized. Devel-opment of such markets is sometimes achievedby promotional efforts but is often accomplishedthrough word of mouth. The cultivation of elitenetworks is thus crucial to the performance andsurvival of microbreweries. Gaining and main-


taining regular access to such consumers can beproblematic, especially given the dominance ofdistribution networks by the mass producers.Finally, it is not clear that brewpubs competewith the two other forms so much as they doagainst local drinking and dining establishments.Potential returns on investments can be quitehigh, however, since the packaging and distri-bution costs faced by other brewers are avoidedby this form.

In keeping with their localized specialist strat-egies, the microbreweries and brewpubs are muchsmaller in size. For the period 1939–95, theaverage annual production capacity of firmsranged from 1391 barrels for brewpubs to 9871barrels for microbreweries to 930,517 barrels formass producers. A specialist strategy is not theonly factor differentiating microbreweries andbrewpubs from mass producers. They differ alsoin their formal structures. In accordance withtheir smaller size they have fewer employees.More importantly, however, entrepreneurs whostart up microbreweries or brewpubs often requiretechnical and brewing skills in addition to theirbusiness acumen (Mares, 1991: 139). Mass pro-ducers may be family-run enterprises, but theyusually hire technical staff for daily operations.The normative order among microbreweries andbrewpubs is also considerably different. Co-operation and not competition seems to be thenorm in mutual interaction. For example, it iscommon for microbreweries and brewpubs in thesame area to negotiate shared bulk purchasingagreements with suppliers of raw materials, toshare distribution channels and to jointly fundpromotional activities. In the next section, Idevelop hypotheses that explain the entry ofmicrobreweries and brewpubs into the brewingindustry in terms of niche formation and resource-partitioning processes.

THEORY AND HYPOTHESES

Niche formation

Delacroix and Solt (1988: 54) argue that found-ings in the California wine industry are drivenby a niche formation process. They write:

A new niche may become available for a giventype of organization with the advent of newtechnologies to perform old tasks, with the open-ing of new environmental resources hitherto not

392 A. Swaminathan

accessible for tapping, or the emergence of newways to obtain resources from the environmenton the basis of unchanged technology.

According to Delacroix and Solt (1988), thelatter process of niche formation has been respon-sible for the recent surge in winery foundings. Intheir view, the new niche in the wine industryevolved out of changes in lifestyle and associatedconsumer preferences. A change in consumerpreferences is one among several exogenous fac-tors that may lead to the creation of a new niche.A new niche could also be formed as a result oftechnological discontinuities. For example, found-ings in the semiconductor manufacturing industryseem to be driven by technological innovation(Brittain and Freeman, 1980). Tushman and And-erson (1986) account for product substitutionthrough technological changes which destroy thecompetencies of incumbent firms. The emergenceof new product classes such as cement (in 1872),airlines (in 1924), and plain-paper copying (in1959) is attributed to basic technological inno-vations. Innovation in production technology doesnot seem to be a major force behind the entryof firms into new market segments in the brewingindustry. In fact, in brewing the technologyemployed by both microbreweries and brewpubsis primitive by industry standards—it closelyresembles the technology in use over a centuryago.

Abernathy and Clark (1985: 18) suggest thatthree specific kinds of environmental changesmight lead to the formation of a new niche.First, new technological options offer improvedperformance or new applications that cannot bemet by existing product designs. Second, changesin government policy, especially in regulatoryregimes, may favor revolutionary strategic devel-opment. Finally, changes in consumer preferencesmay impose requirements that can be met onlythrough new design approaches. Delacroix andSolt’s (1988) niche formation argument thus hasthe same flavor as the process of ‘niche creationinnovation’ (Abernathy and Clark, 1985: 10–11)where organizations build on existing technicalcompetence and apply it to emerging market seg-ments. Given an underlying change in consumerpreferences, the influx of firms into a new marketsegment or niche in a mature industry is likelyto be greater as the new market niche expandsin volume.


Proposition 1: The greater the volume of anew market niche in a mature industry, thehigher the founding rate of firms in that niche.

Delacroix and Solt (1988) use the level of wineimports as an indicator of the volume of a newniche comprising the table and sparkling winesegments. Increases in the level of wine importshave a strong positive effect on foundings in theCalifornia wine industry. Delacroix and Solt(1988) interpret this as evidence that niche forma-tion increases the founding rate. It is even moretrue in the case of brewing that imported productscome closest to resembling microbrewery andbrewpub products. Although Delacroix and Solt’s(1988) analysis is not able to identify the organi-zational form of entrants into new niches in thewine industry, in the brewing industry it is clearthat new niches are being occupied by organi-zations with new forms (Carroll and Swamina-than, 1992). The niche formation argument wouldsuggest the following hypothesis:

Hypothesis 1: The greater the volume of beerimports, the higher the founding rate ofmicrobreweries and brewpubs.

Resource-partitioning

Trends in organizational density—the number oforganizations in an industry—are often related totrends in industrial concentration. As the numberof organizations declines, the market share heldby the largest few firms typically increases. Car-roll (1985) developed a model of resource-partitioning to account for the mortality rates ofspecialist firms in environments characterized byvarying degrees of industrial concentration. Bydefinition, generalists depend on a wide rangeof environmental resources for survival, whereasspecialists survive only within a narrow range ofenvironmental resources.

The level of industrial concentration has impli-cations for the dispersion of resources within amarket. When the market is characterized by lowconcentration levels it approximates the econo-mists’ conception of a perfect, atomistic market.That is, the market is composed of a large numberof generalist firms, each of which cannot indi-vidually affect prevalent price levels. Firms tailortheir products or services to appeal to a slightlydifferent set of consumers. In terms of market


coverage, there is a certain degree of overlapnear the center, but the differentiated appealsdeveloped by individual firms ensures that eachgeneralist firm possesses a unique advantage incertain market segments. The existence of thesepartially intersecting strategies implies that alarger proportion of the total market is covered.The resource space available for specialists issmaller—there are fewer market segments leftto exploit.

As the level of concentration in the marketrises, the death rate of generalist firms increasesas they compete with each other to gain controlover the center of the market. The survivinggeneralist firms that come to dominate the marketare fewer in number and larger in size. Thedegree of overlap in generalist firm strategies ishigher since most of them try to exploit resourcesavailable at the market center. The total resourcespace covered by generalist firms is smaller thanin the case of a competitive, unconcentrated mar-ket where they offer differentiated products orservices. Therefore in the concentrated marketspecialist firms have access to greater resources.They can exploit peripheral market segmentswithout entering into direct competition with thelarger generalists. Greater resource availabilityshould improve the survival chances of suchspecialist firms as the market concentrates. Atthis point, the total market is partitioned intogeneralized and specialized segments. In a con-centrated market generalist and specialist firmsseem to operate in distinct resource spaceswhereas they relied on a common resource basein an unconcentrated market.

The resource-partitioning model can easily beextended to the founding process (Freeman andLomi, 1994; Lomi, 1995; Swaminathan, 1995).High concentration in the market implies thatspecialists can draw upon peripheral resourceswithout entering into direct competition with gen-eralists. Increasing levels of market concentrationfree more peripheral resources. Existing special-ists may grow and survive longer by exploitingsuch resources. The increased availability ofperipheral resources may also facilitate thefounding of specialist organizations. Therefore

Proposition 2: The greater the degree ofresource-partitioning within an industry, thehigher the founding rate of specialists.

In the brewing industry, generalists are represented


by mass producers. New organizational forms—microbreweries and brewpubs—are specialists.

Hypothesis 2: The greater the degree of mar-ket concentration in the brewing industry, thehigher the founding rate of microbreweriesand brewpubs.

The niche formation and resource-partitioninghypotheses need to be tested with reference to abaseline model of organizational founding amongmicrobreweries and brewpubs. In particular, thisbaseline model ought to address three importantinfluences on the founding rate of firms in newmarket segments within a mature industry: thecarrying capacity of the environment, density-dependent evolution, and institutional support.Below I sketch a baseline model of microbreweryand brewpub founding that takes these influencesinto account.

Carrying capacity

Organizational founding rates are likely to behigher when the excess carrying capacity of theenvironment is greater (Brittain and Freeman,1980; Pennings, 1982; Baum and Singh, 1994).The carrying capacity of the environment formicrobreweries and brewpubs is likely to varyacross states. I model these state-level differencesin terms of three variables: the per capita personalincome of a state’s residents, the per capita beerconsumption of a state’s residents, and the per-centage of a state’s population that lives in dryareas. Although there is very little cross-ownershipof firms in the wine and brewing industries, it isintriguing that new organizational forms in brewingemerged shortly after equivalent forms had emergedin the wine industry. It is appealing to argue thatdevelopments in both populations reflect a commonunderlying process—a change in lifestyle andconsumer demand for greater variety. Thesedevelopments are associated with the emergenceof small upscale niches comprising relativelyaffluent consumers. Therefore states with higherper capita incomes are likely to experience ahigher entry rate of microbreweries and brewpubs.On the other hand, states that have a large pro-portion of their populations living in dry areaslikely reflect persistent social norms that discour-age the sale and consumption of alcoholic bever-ages. In such states, microbrewery and brewpubfoundings are likely to be lower.

394 A. Swaminathan

Density-dependent evolution

According to a model of density-dependentorganizational evolution proposed by Hannan(1986), the founding rate within an organizationalpopulation increases when an organizational formreaches higher levels of legitimacy and decreaseswith increasing competition within the population(see also Hannan and Carroll, 1992). The numberof organizations (or density) corresponds to theprocesses of legitimation and competition. At lowlevels of density, each addition to the populationfacilitates the legitimation process and thereforeincreases founding rates. As density increases andapproaches the environmental carrying capacity,competitive processes set in and the effect ofdensity is reversed and it lowers the foundingrate. Thus, this model predicts that density depen-dence in rates of founding is nonmonotonic.

Although the competitive effects of density areintuitive, the positive effect of initial increases indensity on the organizational founding rate issubject to interpretations other than the sociallegitimation of the organizational form (Delacroixand Rao, 1994). In particular, the legitimatingeffects of density on the founding rate of organi-zations with new forms can also be interpretedas a case of learning at the population level(Miner and Haunschild, 1995). Small businessesare more likely to succeed if they are founded byentrepreneurs who have failed at earlier attempts(Mayer and Goldstein, 1961: 138–139). Stinch-combe (1965: 152) notes that ‘the level of organi-zational experience of a population is a maindeterminant of their capacity to form new organi-zations.’ I would argue that the acquisition oflearning and legitimacy through a proliferationin numbers of a new organizational form arecomplementary processes since they both occurat the population level. In fact, these processesmay act in ways that reinforce each other.

The density-dependent model of organizationalevolution has been validated in several organi-zational populations, including an earlier analysisof microbrewery and brewpub foundings (Carrolland Swaminathan, 1992). If microbreweries andbrewpubs are new organizational forms, then thedensity model should apply to them separately.Carroll and Swaminathan’s (1992) analysis ofmicrobrewery and brewpub founding rates over1975–89 supports this view. Their results showthat microbreweries and brewpubs are subject to


separate processes of legitimation and competitionbased on form-specific counts of density at thenational level. The treatment of density-dependenteffects in this study improves upon Carroll andSwaminathan’s (1992) analysis in significantways. First, in keeping with the localized strat-egies of microbreweries and brewpubs, I modelform-specific density dependence in foundings ofthese new organizational forms at the state level.Second, I model interdependence between massproducers, microbreweries, and brewpubs in adifferentiated manner. Microbreweries and brew-pubs in the same state are expected to exhibit amutualistic relationship. Although they producesimilar products, they do not compete directlysince brewpub products are consumed on theirpremises whereas microbrewery products haveto reach consumers through regular distributionchannels. Unlike microbreweries and brewpubs,the impact of generalist mass production brew-eries is likely to be felt nationwide. Mass pro-ducers are likely to exert a competitive effect onmicrobreweries since their products are designedfor at-home consumption and distributed throughthe same channels. The modeling choicesdescribed above are consistent with Hannan andCarroll’s (1992: 208–209) recommendation thatthe level of analysis chosen to model density-dependent effects ought to capture most of thecompetition occurring among similar organi-zations (see also Singh, 1993; Budros, 1994).

Institutional support

A supportive institutional environment is likelyto foster higher organizational founding rates.Institutional support is often provided in the formof government regulations that protect infant orendangered organizations (Carroll, Delacroix, andGoodstein, 1988), laws that structure industrycompetition in particular ways (Barnett and Car-roll, 1993), or financial incentives that encourageentrepreneurial activity (Tucker, Singh, and Mein-hard, 1990). Tuckeret al. (1990) found that theestablishment of the ‘Opportunities for Youth’program between 1971 and 1975 legitimated andprovided additional resources for voluntary socialservice organizations (VSSOs) in Canada. Conse-quently, specialist VSSOs were founded at amuch higher rate during this period. Budros(1992) found that the passage of New York’sInsurance Incorporation Act stimulated the found-


ing of insurance carriers in the state. Similarly,in the brewing industry the legalization of brew-pubs by individual states might signify a moresupportive institutional environment for neworganizational forms in the brewing industry. Bydefinition, brewpubs could not be founded beforethey became legal. A more intriguing possibilityis that microbreweries were founded at higherrates in states that legalized brewpubs.

DATA AND METHODS

Data

Bull, Friedrich, and Gottschalk’s (1984)AmericanBreweriesconstitutes the primary source of event-history data for the population of brewing firms.This data source includes information on allAmerican beer producers except those who pro-duce under contract to others. The data havebeen verified using annual lists of brewing firmspublished in theModern Brewery Age Bluebooks(Modern Brewery Age, various years) and theBrewers Almanac (U.S. Brewers Association,various years) which uses the Department ofAlcohol, Tobacco and Firearms as its primarysource. The historical coverage has been extendedto 1995 by including information gathered fromTremblay and Tremblay (1988) and theMicrobrewery Resource Handbooks(Institute for[Fermentation and] Brewing Studies, variousyears). From the event-history data, I calculatedcounts of density, foundings, and deaths in anygiven year. Four-firm concentration ratios werecalculated as the combined market share of thefour largest firms. These estimates are based onsales data reported in theModern Brewery AgeBluebooksand theMicrobrewery Resource Hand-books. Data on beer imports have been obtainedfrom the Brewers Almanac. Data on the totalproduction of microbreweries and brewpubs havebeen derived by aggregating individual firm pro-duction figures which are available in theModernBrewery Age Bluebooksand the MicrobreweryResource Handbooks. Descriptive statistics for thevariables used in the analyses are given in Tables1 and 2.

Since microbreweries are at risk of foundingin all states during the entire observation period,the analysis of microbrewery founding ratescovers 51 states (including the District ofColumbia) over a 57-year period (1939–95). Data


on per capita income and state population are notavailable for Alaska and Hawaii for a total of 42state-years. Therefore the analysis includes 2865(51 states× 57 years −42) state-year spells.Brewpubs are at risk of founding only after theorganizational form is legalized in a given state.California and Washington were the first statesto legalize the form in 1982. Given that brewpubsare such a recent phenomenon, the analysis ofbrewpub founding rates includes only 305 state-year spells.

Methods

In modeling the organizational founding process,I follow convention in defining the population asthe unit of analysis and treat foundings as eventsin a point process (Cox and Isham, 1980; Ambur-gey and Carroll, 1984; Amburgey, 1986). Becausethe dates of founding often record only the yearof founding, I do not know the ordering of eventswithin years. Nor do I know the exact durationbetween foundings. In doing work of this kind,organizational ecologists have typically assumeda constant rate of founding with log-linear depen-dence on covariates (Hannan and Freeman, 1989).This approach assumes that, conditional on thevalues of the covariates, a time series of annualcounts of foundings is the realization of a Poissonprocess. This implies that the number of found-ings in year t, Yt, is determined by the prob-ability law:

Pr(Yt = yt) = exp(−lt) lyt t/yt! (1)

The relationship between the founding rate,lt,and the vector of covariates,xt, is specified as fol-lows:

ln lt = a + bxt (2)

Some implications of the Poisson model arethat the expected number of events in a givenyear t is equal to the mean founding rate,lt, andthat the variance of the number of events equalslt. Poisson regression methods typically estimatethe regression parameters by using the Poissonprobability law in Equation 1 to form likelihoodsfor the data and then employ methods formaximum likelihood estimation. This approachhas important advantages over methods based onconventional regression analysis of time series

396 A. Swaminathan

Table 1. Descriptive statistics for variables used in the analysis of microbrewery foundingsa

Variable Mean S.D. Minimum Maximum

Number of state-level microbrewery foundings 0.117 0.680 0 19State microbrewery density 0.319 1.479 0 29Out-of-state microbrewery density 15.969 36.227 0 192State brewpub density 0.476 3.268 0 76National mass producer density 182.631 173.236 23 641Industry concentration (4-firm ratio as a 46.945 25.274 13.5 90.0percentage)Volume of imports (millions of gallons) 2.630 3.372 0.031 10.489Brewpub legality (1= legal; 0 = illegal) 0.107 0.309 0 1Brewpub legality× state microbrewery density 0.259 1.436 0 29State per capita annual personal income 10.487 3.876 1.817 25.246(thousands of constant 1987 dollars)State per capita annual beer consumption 18.275 6.682 1.6 40.4(gallons)% of state population living in dry areas 5.99 12.33 0 63.11

aN = 2865 for all variables

Table 2. Descriptive statistics for variables used in the analysis of brewpub foundingsa

Variable Mean S.D. Minimum Maximum

Number of state-level brewpub foundings 1.928 3.312 0 26State brewpub density 4.416 9.116 0 76Out-of-state brewpub density 173.328 106.324 0 352State microbrewery density 2.436 3.756 0 29National mass producer density 25.007 2.243 23 33Industry concentration (4-firm ratio as a 87.310 2.450 75.9 90.0percentage)Volume of imports (millions of gallons) 8.954 0.884 5.754 10.489State per capita annual personal income 16.152 2.504 11.63 25.246(thousands of constant 1987 dollars)State per capita annual beer consumption 23.447 3.822 12.7 37.7(gallons)% of state population living in dry areas 4.055 9.787 0 46.35

aN = 305 for all variables

data on foundings. Most importantly, OLS orGLS regression methods used in early studiesof organizational foundings (see, for example,Delacroix and Carroll, 1983) do not take intoaccount the non-negativity of event counts or thediscontinuous nature of count data.

However, assuming that a series of counts offoundings is a realization of a Poisson processmeans assuming that the occurrence of an event


is independent of previous events. It also meansassuming that the expected number of foundingsin a year equals the variance of the number offoundings in that year. Both these assumptionsare problematic in the analysis of founding rates,where the variance of event counts often exceedsthe mean, a condition called ‘overdispersion’.Either unobserved heterogeneity in founding ratesor positive ‘contagion’ can generate overdisper-


sion. In this case, contagion means that the occur-rence of an event affects the rate of subsequentoccurrence. There is reason to believe that bothsources of overdispersion are evident in theorganizational founding process. Adopting thePoisson model in such a situation can lead tomisleadingly small standard errors for the esti-mated coefficients (Hausman, Hall, and Griliches,1984). Instead, I use the negative binomial modelwhich overcomes this limitation of the Poissonmodel through the inclusion of an overdispersionparameter (Cameron and Trivedi, 1986; Ranger-Moore, Banaszak-Holl, and Hannan, 1991; Bar-ron, 1992). That is, the relationship between thefounding rate,lt, and the vector of covariates,xt, is now specified as follows:

ln lt = a + bxt + e (3)

where e has a gamma distribution. This modelassumes that the coefficient of variation of theexpected count increases linearly with theexpected count. It has an additional overdisper-sion parameter,u, where

Var(Yt) = E(Yt) (1 + uE[Yt]) (4)

Settingu = 0 in Equation 4 simplifies the negativebinomial model to a Poisson model. Thus onecan form likelihood ratio tests of the Poissonprocess vs. the negative binomial. The likelihoodratio test statistic is defined as

m = max L0/max L1 (5)

where L0 and L1 denote the likelihoods of thenull model (subject to, say,n constraints) andthe alternative model that relaxes the constraintsrespectively. With large samples,−2ln m is dis-tributed as a chi-square withn degrees of free-dom. I report this statistic as the log-likelihoodchi-squared ratio for the founding models in mytables. The log-likelihood chi-squared ratio com-pares the log-likelihood of each model with anull model that assumes a constant founding rate.The difference of log-likelihood chi-squared ratiosof a pair of hierarchically nested models in mytables has approximately a chi-square distributionunder the null hypothesis.

I use maximum likelihood methods availablein the statistical package LIMDEP (Greene, 1995)to estimate the founding models. Maximum likeli-


hood estimation revealed that models of foundingrates for both microbreweries and brewpubs con-sistently show evidence of overdispersion. HenceI report negative binomial regression models ofmicrobrewery and brewpub founding counts.

FINDINGS

Table 3 presents estimates from negative binomialregression models of state-level microbreweryfoundings. Model 1 shows that the variables mea-suring the carrying capacity of the environmentaffect the microbrewery founding rate in predict-able ways. States with more affluent residentsand higher beer consumption experience highermicrobrewery founding rates. Indirect institutionalsupport also encourages microbrewery foundings.Microbrewery foundings are accelerated by thelegalization of brewpubs, an organizational formthat also draws on the local market for suste-nance. The overdispersion parameter,u, is sta-tistically significant in all models in Table 3.

Density dependence in microbrewery foundingrates is addressed in Model 2. Microbreweryfounding rates initially rise and then fall withincreases in local (state-level) microbrewery den-sity. This result is consistent with the predictionsof Hannan’s (1986) model of density dependence.Nonlocal (out-of-state) microbrewery density alsohas a positive effect on the state-level foundingrate. The mutualistic effect of nonlocal micro-breweries is somewhat surprising given theexpansion of some prominent microbrewers intoregional and in some cases national distributionof their products. One interpretation of this resultis that entry into new market segments is acontagious process that operates on a broad geo-graphic level. A large number of viable firmsexisting in nonlocal areas may indicate that thenew market niche has a large carrying capacity,thus attracting potential local entrepreneurs.Microbrewery founding rates are not affected bystate-level brewpub density, but are negativelyimpacted by mass producer density. This is to beexpected as microbreweries preceded brewpubsas an organizational form and they compete moredirectly with mass producers in output markets.

In Model 3, I find that the effect of brewpublegalization on microbrewery foundings is morecomplex than it would seem from Model 1. Themain effect of brewpub legalization is still posi-

398A

.S

wa

min

ath

an

Table 3. Negative binomial regression models of state-level microbrewery foundings

Independent variables Modelsa

1 2 3 4 5 6

Constant −8.371* −2.907* −3.119* −4.743* −5.474* −2.308(0.960) (0.941) (0.955) (1.178) (2.445) (2.790)

State per capita annual personal income 0.3021* 0.0594 0.0645* 0.0618 0.0638 0.0622(thousands of constant 1987 dollars) (0.0467) (0.0319) (0.0326) (0.0333) (0.0329) (0.0332)

State per capita annual beer consumption (gallons) 0.0547* 0.0244 0.0199 0.0184 0.0200 0.0179(0.0262) (0.0238) (0.0244) (0.0249) (0.0245) (0.0250)

% of state population living in dry areas −0.0115 −0.0067 −0.0060 −0.0065 −0.0060 −0.0067(0.0143) (0.0116) (0.0118) (0.0119) (0.0118) (0.0120)

Brewpub legality 1.9707* 0.2648 0.5785* 0.5496* 0.5419* 0.5910*(1 = legal; 0 = illegal) (2.057) (0.1841) (0.2253) (0.2278) (0.2290) (0.2321)

State microbrewery density 0.3681* 0.5184* 0.5283* 0.5219* 0.5282*(0.0373) (0.0499) (0.0525) (0.0520) (0.0526)

(State microbrewery density)2/100 −1.0027* −0.8871* −0.9010* −0.8863* −0.9099*(0.1806) (0.2257) (0.2339) (0.2287) (0.2336)

Out-of-state microbrewery density 0.0075* 0.0068* 0.0041 0.0059* 0.0044*(0.0014) (0.0014) (0.0022) (0.0020) (0.0022)

State brewpub density −0.0019 0.0015 0.0044 0.0016 0.0054(0.0128) (0.0132) (0.0137) (0.0133) (0.0142)

National mass producer density −0.0394* −0.0359* −0.0182* −0.0207 −0.0296*(0.0069) (0.0070) (0.0085) (0.0122) (0.0127)

Brewpub legality× state microbrewery density −0.1879* −0.2039* −0.1923* −0.2046*(0.0625) (0.0658) (0.0641) (0.0657)

Volume of imports 0.1754* .2042*(millions of gallons) (0.0831) (0.1184)

Industry concentration 0.0243 −0.0321(4-firm ratio as a percentage) (0.0249) (0.0357)

u 2.1795* 0.2620* 0.2876* 0.3020* 0.2986* 0.2970*(overdispersion parameter) (0.3136) (0.1221) (0.1338) (0.1333) (0.1378) (0.1306)

Log likelihood chi-squared ratio 1228.2 1452.6 1459.0 1464.5 1459.8 1465.3

Degrees of freedom 5 10 11 12 12 13

Number of cases 2865 2865 2865 2865 2865 2865

Number of events 335 335 335 335 335 335

*p , 0.05aStandard errors are in parentheses.

1998

JohnW

iley&

Sons,

Ltd.S

trat.

Mg

mt.

J.,Vol

19,389

–404

(1998)


tive. But the interaction of brewpub legalizationwith microbrewery density exerts a negativeeffect on the microbrewery founding rate. In otherwords, the effect of brewpub legalization onmicrobrewery founding rates varies over levelsof microbrewery density. The estimates in Model3 suggest that if four or more microbreweriesexist in a state, the overall effect of brewpublegalization on microbrewery founding rates isnegative. One interpretation of this finding is thatonce brewpubs are legalized, entrepreneurs canchoose to enter either the more establishedmicrobrewery or the fledgling brewpub segment.Higher levels of microbrewery density under suchconditions imply greater competition. Thereforepotential entrepreneurs may choose to enter thebrewpub segment instead. This interpretation isalso consistent with an unfolding of the resource-partitioning process within the microbrewery seg-ment of the industry.

Models 4 and 5 test the niche formation andresource-partitioning arguments separately. Model4 supports Hypothesis 1. Microbrewery foundingrates increase with the volume of imports (theindicator of niche formation). Model 5 does notsupport Hypothesis 2—the effect of industry con-centration (the indicator of resource-partitioning)on the microbrewery founding rate is negligible.

Finally, Model 6 pits the niche formation andresource-partitioning explanations for the entry offirms into new market segments in mature indus-tries against each other. The results suggest thatniche formation is primarily responsible for theentry of microbreweries into the brewing industry.The effect of factors endogenous to the industryimplied in the resource-partitioning hypothesis areinsignificant once the exogenous changes sug-gested by the niche formation scenario are takeninto account. The effects of microbrewery densityand brewpub legalization on microbrewery found-ing rates are consistent with earlier models. Totest the robustness of Model 6, I added a variablesignifying the population mass of microbreweries(annual production output). Barnett and Ambur-gey (1990: 82–84) argued that if larger organi-zations within a population generate strongercompetition, then population mass should lowerthe founding rate. In this supplementary model,estimates of which are not provided here, I foundthat population mass does not affect the foundingrate of the microbreweries significantly. Parameterestimates of variables included in Model 6, how-


ever, are robust to the addition of the populationmass variable. The lack of a population masseffect may signify a great degree of differentiationwithin the microbrewery population.

Table 4 presents estimates from negativebinomial regression models of state-level brewpubfoundings. Model 7 shows that the carryingcapacity variables affect the brewpub foundingrate in intuitive ways. The effect of the incomevariable is statistically significant in Model 7, butloses its significance in subsequent models. Theoverdispersion parameter,u, is statistically sig-nificant in all models in Table 4. Model 8accounts for density-dependent effects on thebrewpub founding rate. As predicted by the den-sity model (Hannan, 1986), state-level brewpubdensity exerts a nonmonotonic effect on state-level brewpub founding rates. Given the localnature of the brewpub segment, it is not surprisingthat national mass producer density does not exertan appreciable effect on the brewpub foundingrate. Out-of-state brewpub density has a positive,but inconsistent effect on the brewpub foundingrate, adding credence to the ‘entrepreneurship ascontagion’ explanation offered earlier for nonlocaleffects on microbrewery foundings. In addition,state-level microbrewery density has a positiveeffect on the state-level brewpub founding rate.This result suggests that the spread of themicrobrewery organizational form had a legi-timating effect on the founding process of brew-pubs.

Models 9 through 11 test the two alternativeHypotheses 1 and 2 derived from niche formationand resource-partitioning theory. The results sug-gest that the founding of firms in the brewpubsegment of the U.S. brewing industry is mainlydriven by the niche formation process. Thegreater the volume of beer imports, the indicatorof niche formation, the greater the brewpubfounding rate. This result supports Hawley’s(1988) suggestion that niche emergence may bepartly responsible for the generation of neworganizations. As in the case of microbreweries,I found that the parameter estimates in Model 11were robust with respect to the addition of avariable, total brewpub production, that is a proxyfor population mass. It is interesting that theentry of firms into the microbrewery and brewpubsegments of the U.S. brewing industry is betterexplained by niche formation than by resource-partitioning. I will explore the implications of

400A

.S

wa

min

ath

an

Table 4. Negative binomial regression models of state-level brewpub foundings

Independent variables Modelsa

7 8 9 10 11

Constant −1.491 −2.652 −4.487* −3.147 11.120(1.083) (1.486) (1.495) (8.902) (9.389)

State per capita annual personal income 0.1417* 0.0013 −0.0028 0.0012 −0.0026(thousands of constant 1987 dollars) (0.0461) (0.0340) (0.0331) (0.0341) (0.0328)State per capita annual beer consumption −0.0060 0.0282 0.0231 0.0282 0.0229(gallons) (0.0229) (0.0203) (0.0193) (0.0204) (0.0201)% of state population living in dry areas −0.0152 −0.0041 −0.0056 −0.0042 −0.0052

(0.0153) (0.0116) (0.0111) (0.0116) (0.0109)State brewpub density 0.0978* 0.0986* 0.0977* 0.1025*

(0.0216) (0.0211) (0.0223) (0.0216)(State microbrewery density)2/100 −1.1222* −0.1209* −0.1220* −0.1250*

(0.0309) (0.0344) (0.0317) (0.0335)Out-of-state brewpub density 0.0027* 0.0002 0.0026* 0.0018

(0.0006) (0.0007) (0.0013) (0.0013)State microbrewery density 0.0844* 0.0725* 0.0842* 0.0741*

(0.0264) (0.0260) (0.0265) (0.0262)National mass producer density 0.0531 0.0061 0.0559 −0.0883

(0.0446) (0.0452) (0.0613) (0.0663)Volume of imports 0.4037* 0.4897*(millions of gallons) (0.0966) (0.1047)Industry concentration 0.0050 −0.1639(4-firm ratio as a percentage) (0.0903) (0.0969)u 1.3736* 0.5225* 0.4571* 0.5222* 0.4502*(overdispersion parameter) (0.1629) (0.0975) (0.0920) (0.0975) (0.0914)Log likelihood chi-squared ratio 476.9 604.1 621.2 604.1 624.9Degrees of freedom 4 9 10 10 11Number of cases 305 305 305 305 305Number of events 588 588 588 588 588

*p , 0.05aStandard errors are in parentheses.

1998

JohnW

iley&

Sons,

Ltd.S

trat.

Mg

mt.

J.,Vol

19,389

–404

(1998)


these findings among others in the discussionthat follows.

DISCUSSION

This paper examined two alternative explanationsfor the entry of firms into new market segmentsin mature industries. First, niche formation theoryderives its explanatory power from factors thatare exogenous to the industry such as changesin consumer taste or basic technology. Theseexogenous changes create unmet demand for newproducts and services, and entrepreneurs recogniz-ing potential opportunities found organizations,often with new forms. Second, resource-partitioning theory relies on structural changeswithin an industry that lead to the fragmenting ofthe industry into generalist and specialist niches.Generalist organizations focus on producingstandardized products and services for the massmarket. Opportunities in the specialist niche areignored either because they are unattractive dueto a lack of scale economies or because of organi-zational inertia. The reluctance or the inability ofgeneralist organizations to expand into thespecialist niche allows organizations with newspecialist forms to be founded in mature indus-tries. Tests of these two theories on the entry offirms into the microbrewery and brewpub seg-ments of the American brewing industry providestrong support for the niche formation hypothesis.The lack of support for the resource-partitioningmodel may stem from at least two limitations ofthis study. First, we may have to measure thedegree of resource-partitioning more appropri-ately. Second, we may need to reconceptualizethe resource-partitioning process.

Market concentration has been used widely asa measure of resource-partitioning, but it is aweak proxy at best. At the heart of the resource-partitioning model is an argument that a declinein organizational diversity within an industry willlead to the entry of new specialist firms. Ratherthan employ market concentration as an inversemeasure of organizational diversity, it would bepreferable to measure organizational diversitywithin an industry directly in terms of diversityin technologies, products, markets, and internalstructures.

Resource-partitioning has been typically con-ceptualized as a one-time phenomenon. For


example, the design of this study assumed thatincreasing market concentration and the ac-companying decline in product diversity amongmass production breweries would simultaneouslycreate opportunities for specialist firms in themicrobrewing and brewpub segments of the brew-ing industry. In contrast, I would like to suggesta different approach, one that conceptualizesresource-partitioning as a continuous, cyclicalprocess that repeats itself as industries and, moregenerally, market niches within them mature.

By virtue of their producing and marketingpackaged products, microbreweries exhibit greaterinterdependence with mass production breweries.The brewpub organizational form occupies aniche that is far removed from the mass producersegment. This reasoning is partly supported bythe effects of mass producer density on the entryof firms into the microbrewing and brewpub seg-ments. This study considers the effect of resource-partitioning due to structural changes in the massproducer segment on the entry of firms into newmarket segments in a mature industry. Resource-partitioning within the mass producer segment ofan industry is more likely to affect the entry offirms into the new market segment that has agreater overlap with the mass producer segment,in this case the microbrewery segment.

The intriguing effects of brewpub legalizationon microbrewery foundings suggest that similarforces may be at work within the microbrewingsegment of the industry. One interpretation ofthese effects is that brewpubs are a variation ofmicrobreweries that arise due to structuralchanges within the microbrewing segment of theindustry. Microbrewery foundings may be lowerin states that have legalized brewpubs becauseentrepreneurs react to overcrowding in themicrobrewery segment and found brewpubsinstead. This interpretation is consistent withresults that show a positive effect of microbrew-ery density on the brewpub founding rate in agiven state. Brewpub products may find easieracceptance in states with a significant microbrew-ery presence—consumers can now buy similarproducts for consumption at and away fromhome. In states where brewpubs are legal,microbreweries often serve as organizational blue-prints for entrepreneurs planning to enter theindustry. But if the microbrewing segment iscrowded or dominated by a few firms then poten-tial entrepreneurs are likely to found a brewpub,

402 A. Swaminathan

an organizational form that is new to the craftbrewing segment of the brewing industry. Athorough examination of the cyclical nature ofthe resource-partitioning would require data onmarket concentration in the microbrewery seg-ment of the industry. Reliable data on micro-brewery sales are now available for the past 8years (Institute for Brewing Studies, 1996), buta longer time frame is probably required for anadequate test of the modified resource-partitioninghypothesis. I plan to continue collecting suchdata so that I can examine this proposition inthe future.

My attempt to conceptualize resource-parti-tioning as a continuous cycle within an industryis consistent with punctuated models of industryevolution. High levels of industry concentrationin mature industries often reflect the success ofdominant mass production firms at producinglow-cost standardized products. But in directopposition to this very trend toward standardiz-ation emerges another trend towards differen-tiation. There is some anecdotal and empiricalevidence in support of such a dialectical, cyclicalprocess. Petersen and Berger (1975) argue, forinstance, that specialist forms in the music indus-try attempt to expand by standardizing their prod-ucts, thus renewing the concentration process.Similarly, Anderson and Tushman (1990) proposea cyclical model of technological change wherethe emergence of a dominant design sets thestage for an era of incremental change abruptlyfollowed by the next technological discontinuity.In examining the sources of dominant designs infour industries—cement, container glass, flatglass, and minicomputers—they find that on theaverage new entrants are more likely to introducedesigns that destroy the technological competenceof incumbent firms. While Anderson and Tush-man (1990) focus on technological evolution, thisstudy addresses the evolution of market nicheswithin an industry and the organizational formsthat occupy these niches. In this case, the emerg-ence of a dominant organizational form sets thestage for a discontinuity in product markets, adiscontinuity that is often introduced through theentry of organizations with new forms (see alsoRomanelli, 1991: 93–96 for a recent discussionof organizational speciation).

This paper evaluates two alternative theories,niche formation and resource-partitioning, thatseek to explain the entry of firms into new market


segments within a mature industry. Entrants withnew organizational forms contribute to anincrease in organizational diversity within anindustry. An understanding of the sources ofdiversity is central to an evolutionary perspectiveon organizations and industries. At an abstractlevel, a system characterized by greater diversitycan respond better to changing environmentalconditions. In terms of industry structure, onethat is composed of firms manufacturing a diverseset of products is more likely to satisfy theneeds of a heterogeneous market. Although theconclusions drawn here are based on findingsfrom the brewing industry, they are broadly gen-eralizable to the entry of firms into new marketsegments in mature industries that are charac-terized by low levels of organizational densityaccompanied by high degrees of concentration.Similar structural conditions exist in industriessuch as newspapers, book publishing, music re-cording, retailing, life insurance agencies, adver-tising, and managerial consulting. In fact, sincethe resource-partitioning process is likely to haveoccurred at an earlier time in some of theseorganizational populations, they might provideadded insight, particularly with regard to theprevalence of a cyclical process of resource-partitioning within newly emerging segments ofan industry.

ACKNOWLEDGEMENTS

Support for this research was provided by theUniversity of Michigan Business School. I wouldlike to thank Bill Barnett, Glen Dowell, BradKillaly and Will Mitchell for comments on earlierversions of this paper.

REFERENCES

Abernathy, W. and K. B. Clark (1985). ‘Innovation:Mapping the winds of creative destruction’,Research Policy,14, pp. 3–22.

Aldrich, H. E. (1979).Organizations and Environments.Prentice-Hall, Englewood Cliffs, NJ.

Amburgey, T. L. (1986). ‘Multivariate point processesin social research’, Social Science Research,15, pp.190–207.

Amburgey, T. L. and G. R. Carroll (1984). ‘Timeseries models for event counts’, Social ScienceResearch,13, pp. 38–54.

Anderson, P. and M. L. Tushman (1990). ‘Technologi-


cal discontinuities and dominant designs: A cyclicalmodel of technological change,’AdministrativeScience Quarterly,35, pp. 604–633.

Barnett, W. P. and T. L. Amburgey (1990). ‘Do largerorganizations generate stronger competition?’ InJ. V. Singh (ed.),Organizational Evolution: NewDirections. Sage, Newbury Park, CA, pp. 78–102.

Barnett, W. P. and G. R. Carroll (1993). ‘How insti-tutional constraints affected the organization of earlyU.S. telephony’, Journal of Law, Economics andOrganization,9, pp. 98–126.

Barron, D. N. (1992). ‘Analysis of event counts: Over-dispersion and autocorrelation’. In P. V. Marsden(ed.), Sociological Methodology, Vol. 12. BasilBlackwell, Cambridge, MA, pp. 217–243.

Baum, J. A. C. and J. V. Singh (1994). ‘Organizationalniches and the dynamics of organizational founding’,Organization Science,5, pp. 483–501.

Beesley, M. E. and R. T. Hamilton (1984). ‘Smallfirms’ seedbed role and the concept of turbulence’,Journal of Industrial Economics,33, pp. 217–232.

Brittain, J. W. and J. H. Freeman (1980). ‘Organiza-tional proliferation and density dependent selection’.In J. R. Kimberly, R. H. Miles and Associates(eds.),The Organizational Life Cycle. Jossey-Bass,San Fransisco, CA, pp. 291–338.

Budros, A. (1992). ‘The making of an industry: Organi-zational births in New York’s Life Insurance Indus-try, 1842–1904’,Social Forces,70, pp. 1013–1033.

Budros, A. (1994). ‘Analyzing unexpected densitydependence effects on organizational births in NewYork’s life insurance industry, 1842–1904’,Organi-zation Science,5, pp. 541–553.

Bull, D., M. Friedrich and R. Gottschalk (1984).Amer-ican Breweries. Bullworks, Trumbull, CT.

Cameron, A. and P. Trivedi (1986). ‘Econometric mod-els based on count data: Comparisons and appli-cation of some estimators and tests’,Journal ofApplied Econometrics,1, pp. 29–53.

Carroll, G. R. (1985). ‘Concentration and specializa-tion: Dynamics of niche width in populations oforganizations’,American Journal of Sociology,90,pp. 1261–1283.

Carroll, G. R., J. Delacroix and J. Goodstein (1988).‘The political environments of organizations: Anecological view’. In B. M. Staw and L. L. Cum-mings (eds.),Research in Organizational Behavior,Vol. 10. JAI Press, Greenwich, CT, pp. 359–392.

Carroll, G. R. and A. Swaminathan (1992). ‘Theorganizational ecology of strategic groups in theAmerican brewing industry from 1975 to 1990’,Industrial and Corporate Change,1(1), pp. 65–97.

Caves, R. E. and M. E. Porter (1977). ‘From entrybarriers to mobility barriers’,Quarterly Journal ofEconomics,91, pp. 241–262.

Cox, D. R. and V. Isham (1980).Point Processes.Chapman & Hall, London.

Delacroix, J. and G. R. Carroll (1983). ‘Organizationalfoundings: An ecological study of the newspaperindustries of Argentina and Ireland’,AdministrativeScience Quarterly,28, pp. 274–291.

Delacroix, J. and M. E. Solt (1988). ‘Niche formationand foundings in the California wine industry 1941–


84’. In G. R. Carroll (ed.),Ecological Models ofOrganizations. Ballinger, Cambridge, MA, pp. 53–70.

Delacroix, J. and H. Rao (1994). ‘Externalities andecological theory: Unbundling density dependence’.In J. A. C. Baum and J. V. Singh (eds.),TheEvolutionary Dynamics of Organizations. OxfordUniversity Press, New York, pp. 255–268.

Elzinga, K. G. (1986). ‘The beer industry’. In W.Adams (ed.),The Structure of American Industry,Macmillan, New York, pp. 203–238.

Freeman, J. H. and A. Lomi (1994). ‘Resource-partitioning and foundings of banking cooperativesin Italy.’ In J. A. C. Baum and J. V. Singh (eds.),The Evolutionary Dynamics of Organizations.Oxford University Press, New York, pp. 269–293.

Greene, W. H. (1995).LIMDEP Version 7.0 User’sManual. Economic Software Inc., New York.

Hannan, M. T. (1986). ‘Competitive and institutionalprocesses in organizational ecology’, TechnicalReport 86–13, Department of Sociology, CornellUniversity.

Hannan, M. T. and G. R. Carroll (1992).Dynamics ofOrganizational Populations: Density, Legitimationand Competition. Oxford University Press, NewYork.

Hannan, M. T. and J. H. Freeman (1984). ‘Structuralinertia and organizational change’,American Socio-logical Review,49, pp. 149–164.

Hannan, M. T. and J. H. Freeman (1989).Organiza-tional Ecology. Harvard University Press, Cam-bridge, MA.

Hausman, J., B. Hall and Z. Griliches (1984). ‘Econo-metric models for count data with an application tothe patents–R and D relationship’,Econometrica,52, pp. 909–938.

Hawley, A. H. (1988). ‘Foreword’. In G. R. Carroll(ed.), Ecological Models of Organizations.Ball-inger, Cambridge, MA, pp. xiii–xvi.

Institute for [Fermentation and] Brewing Studies(1986–87).Microbrewers Resource Handbook andDirectory. Institute for Fermentation and BrewingStudies, Boulder, CO.

Institute for [Fermentation and] Brewing Studies(1988). North American Microbrewers ResourceHandbook and Directory. Institute for Fermentationand Brewing Studies, Boulder, CO.

Institute for Brewing Studies (1989).North AmericanMicrobrewers Resource Directory. Institute forBrewing Studies, Boulder, CO.

Institute for Brewing Studies (1990–96).North Amer-ican Brewers Resource Directory. Institute for Brew-ing Studies, Boulder, CO.

Keithahn, C. F. (1978).The Brewing Industry. Bureauof Economic Research, Washington, DC.

Lomi, A. (1995). ‘The population ecology of organi-zational founding: Location dependence and unob-served heterogeneity’,Administrative Science Quar-terly, 40, pp. 111–144.

Mascarenhas, B. and D. Aaker (1989). ‘Mobility bar-riers and strategic groups’,Strategic ManagementJournal, 10(5), pp. 475–485.

Mares, W. (1991).Making Beer. Knopf, New York.

404 A. Swaminathan

Mayer, K. B. and S. Goldstein (1961).The First TwoYears: Problems of Small Firm Growth and Sur-vival. Small Business Administration, Washington,DC.

Miner, A. S. and P. R. Haunschild (1995). ‘Populationlevel learning’. In B. M. Staw and L. L. Cummings(eds.), Research in Organizational Behavior, Vol.17. JAI Press, Greenwich, CT, pp. 115–166.

Modern Brewery Age (various years).Modern BreweryAge Bluebook. Modern Brewery Age, Norwalk, CT.

Pennings, J. (1982). ‘Organizational birth frequencies:An empirical investigation’,Administrative ScienceQuarterly, 27, pp. 120–144.

Petersen, R. A. and D. G. Berger (1975). ‘Cycles insymbol production: The case of popular music’,American Sociological Review,40, pp. 157–178.

Ranger-Moore, J., J. Banaszak-Holl and M. T. Hannan(1991). ‘Density dependence in regulated industries:Founding rates of banks and life insurance com-panies’, Administrative Science Quarterly,36, pp.36–65.

Romanelli, E. (1991). ‘The evolution of new organi-zational forms’, Annual Review of Sociology,17,pp. 79–103.

Scherer, F. M. (1980).Industrial Market Structure andEconomic Performance(2nd ed.). Rand McNally,Chicago, IL.

Singh, J. V. (1993). ‘Review essay: Density dependence


theory—current issues, future promise’,AmericanJournal of Sociology,99, pp. 464–473.

Stinchcombe, A. L. (1965). ‘Social structures andorganizations’. In J. G. March (ed.),Handbook ofOrganizations. Rand McNally, Chicago, IL, pp.142–193.

Swaminathan, A. (1995). ‘The proliferation of specialistorganizations in the American wine industry, 1941–1990’, Administrative Science Quarterly,40, pp.653–680.

Tremblay, V. J. and C. H. Tremblay (1988). ‘Thedevelopment of acquisition: Evidence from the U.S.brewing industry’,Journal of Industrial Economics,37, pp. 21–46.

Tucker, D. J., J. V. Singh and A. G. Meinhard (1990).‘Organizational form, population dynamics, andinstitutional change: A study of founding patternsof voluntary associations’,Academy of ManagementJournal, 33, pp. 151–178.

Tushman, M. and P. Anderson (1986). ‘Technologicaldiscontinuities and organizational environments’,Administrative Science Quarterly,31, pp. 439–465.

U.S. Brewers Association (various years).BrewersAlmanac. U.S. Brewers Association, Washington,DC.

U.S. Bureau of the Census (various years).Censusof Manufactures. U.S. Government Printing Office,Washington, DC.

entry into new market segments in mature industries: endogenous and exogenous segmentation in the...

Documents