Role of Structural and Planning Autonomy in the Performance of Internal Corporate Ventures_2012

The Role of Structural and PlanningAutonomy in the Performance of InternalCorporate Venturesjsbm_363 469..497

by Kevin L. Johnson

Historically, autonomy has been treated as a broad one-dimensional construct.This paper proposes that there are two major types of autonomy for internal corporateventures (ICVs)—planning autonomy and structural autonomy—and examines theirrespective impact on performance. Little prior knowledge exists regarding theautonomy–performance relationship for ICVs despite the billions of dollars of corpo-rate investment in ICVs. In this study, I collect primary data on 38 ICVs at differentstages of development from both venture-level and corporate-level managers from overa dozen companies in the U.S. Midwest. I find that a negative relationship exists forplanning autonomy regardless of venture stage. However, for venture structuralautonomy, a more complex relationship was discovered, which ranged from positiveto negative to curvilinear based on the venture’s stage of development.

How can an established companycreate a successful new business?Although many factors come to mind andhave become a point of academic debate,given the parent–venture relationship,the issue of autonomy should be amongthe top candidates. This study providesan empirical contribution to the corpo-rate entrepreneurship literature regard-ing the role of autonomy which to datehas been primarily conceptual in nature.

The success of new products isequated with technical performance,value to customer, and synergy with firmcompetencies (Henard and Szymanski2001; Zirger and Maidique 1990). Otherthan a loose reference regarding man-agement involvement, the examinationof autonomy has been limited. Addition-ally, prior research suffered from threeimportant maladies: (1) common methodbias, a long recognized problem that

Kevin L. Johnson is assistant professor of Management in the Opus College of Business atthe University of St. Thomas.

Address correspondence to: K. L. Johnson, Management, Opus College of Business, Uni-versity of St. Thomas, MCH 316, 2115 Summit Avenue, St. Paul, MN 55105-1096, USA. Phone:651-962-5431. E-mail: [email protected].

Journal of Small Business Management 2012 50(3), pp. 469–497

JOHNSON 469

causes spurious results due to dependentand independent variables obtainedtogether (Campbell and Fiske 1959); (2)extensive sample contamination fromrelated phenomena (discussed in detaillater); and (3) ambiguous definitions.Nevertheless, a popular assumptiondriven by anecdotal evidence and im-plicit theories regarding the success of aninternal corporate venture (ICV) is thatthe more autonomy, the better. Conse-quently, research into the autonomy–performance relationship has beeninconclusive. The basic research ques-tion examined in this study is: Do ICVsdesigned with more autonomy performbetter than those designed with lessautonomy?

The major contribution of this studycomes from the identification and devel-opment of two types of ventureautonomy—structural and planning—that are significant to ICV performance.Whereas venture structural autonomy(VSA) represents the extent to which theventure’s operations are linked to thoseof other businesses of the corporation,venture planning autonomy (VPA) repre-sents the extent to which the venture’smanagement team is responsible forestablishing goals, timetables, and strat-egy for the venture. Another majorcontribution of this study is the first dis-covery of the complex relationshipbetween ICV performance and VSA.

The motivation to examine ICVs isdue to their potential strategic competi-tiveness for both small and large compa-nies and the lack of understanding ofwhat truly drives ICV performance, asshown by the inconsistency of findingsand performance. ICVs are themselvessmall businesses. Learning from theefforts and mistakes of large establishedcorporations can be crucial to the sur-vival of the independent small business.Furthermore, the necessary capital avail-able to the large corporation (Baysingerand Hoskisson 1989; Gooding andWagner 1985) to purchase an external

venture can be a significant constraint forsmaller enterprises. Finally, corporationsinvest billions of dollars in resources forinternal venture development projectseven though after years of operationsand irrespective of the measures ofsuccess (Block 1989; Campbell and Park2004; Garvin 2002, 2004), 50–99 percentfail to achieve their performance expec-tations (Birkinshaw 2005; Chesbrough2000). Yet ICVs continue to be pursuedover external ventures because whensuccessful, they represent more innova-tive growth (Antoncic and Prodan 2008;McCrea and Betts 2008).

Literature Reviewand HypothesesThe Autonomy–Performance Link

Numerous potential contributors toICV success and failure have beenexplored (MacMillan, Block, and Subba-narasimha 1986; Sykes and Block 1989)with limited results. Autonomy was firstselected for the study variable herebecause it is one of the four dimensionsof ICVs provided by Thornhill and Amit(2001) that is also arguably a distinguish-ing factor between corporate andtraditional entrepreneurship (e.g., ICVmanagers may be directed by parent cor-porations, whereas independent entre-preneurs have autonomy over thedecision making for their ventures)—theother dimensions are degree of related-ness, extent of innovation, and natureof sponsorship. Autonomy was alsoselected because it is also a dimension ofentrepreneurial orientation (EO) andtherefore may be a key factor in ventureperformance outcomes. A firm with prac-tices and policies designed around entre-preneurship is described as having anentrepreneurial orientation (Lumpkinand Dess 1996). The EO constructincludes dimensions of innovativeness,proactiveness, risk taking, competitiveaggressiveness, and autonomy (Lyon,Lumpkin, and Dess 2000; Miller 1983;Miller and Friesen 1982), and has been

JOURNAL OF SMALL BUSINESS MANAGEMENT470

attributed to the overall success of somefirms (McGrath and MacMillan 2000).Although there have been mixed resultsregarding a specific relationship betweenEO and established firm performance(Wiklund and Shepherd 2005), the mostrecent meta-analytical study still indi-cates a moderate (r = 0.24) relationship(Rauch et al. 2009), which may be aresult of the role of autonomy.

This study does not test the EO of anorganization but examines the autonomyrelationship. Despite recent papers onautonomy versus dependence (Robins,Tallman, and Fladmoe-Lindquist 2002),autonomy in international joint venturedecisions (Glaister, Husan, and Buckley2003), and the issue of control versusautonomy (Paik and Choi 2007), theimportance of autonomy on ICV perfor-mance has been generally neglected;meanwhile, existing scales “were oftenlimited to measurement of structuralautonomy without regard to strategicautonomy” (Lumpkin, Cogliser, andSchneider 2009, p. 53). “[F]rom an EOperspective [autonomy] refers primarilyto strategic autonomy” (Lumpkin,Cogliser, and Schneider 2009, p. 50).

In total, virtually by definition, newICVs represent innovative and proactivebehavior. A new ICV is also risky as anunknown (Matthews and Human 2004)but not always autonomous. An ICV islinked to an existing parent with whomthe type and level of autonomy may sub-stantially differ; therefore, it is conceiv-able that autonomy might explain a lot ofvariance in ICV performance. However,in order to ascertain the autonomy rela-tionship for ICVs, we must clearly distin-guish the ICV from related phenomena,thereby addressing an earlier malady.

Related Phenomena—Similar ButNot Sufficient

ICVs exist under the umbrella of cor-porate entrepreneurship—a broad termthat encompasses several different busi-ness development activities (Sharma and

Chrisman 1999). A variety of criteria havebeen used since 1977 to depict the ICVs(see Table 1). In general, a company thatinternally generates, funds, and developsnew businesses utilizing internalresources is directly engaging in internalcorporate venturing or the developmentof ICVs (Burgelman 1983, 1984a, 1984b,1984c; Campbell et al. 2003; Miles andCovin 2002). In the exploration of internalventures, researchers have comparedsmall versus large firms (Day 1994; Siegel,Siegel, and Macmillan 1993), independentventures versus corporate ventures(Simon and Shrader 1997; Zahra 1996),focused on a single industry (McGrath1995; Zahra and George 1999), and evendifferent countries (Thornhill and Amit2001). Throughout, the overlap of phe-nomena is understandable; however,scholars now recognize that managingand developing an ICV is not the same asrunning an established business, pursu-ing an external start-up, or even necessar-ily engaging in traditional productdevelopment. This distinction is crucial tothe advancement of ICV research.

Product Development. One commonoverlap with ICVs has been productdevelopment and the plethora of studiesthat exist (Brown and Eisenhardt 1995;Cooper, Edgett, and Kleinschmidt 2001;Gomes et al. 2001; Henard and Szyman-ski 2001; Ledwith 2000; Simon andHoughton 2003; Storey and Easingwood1996; Story, Smith, and Saker 2001).Today, scholars acknowledge that a newproduct is not necessarily the same thingas a new corporate venture (Greene,Brush, and Hart 1999), and that acommon problem for corporations hasbeen in the failure to understand thedistinctive features of new markets, busi-nesses, products, and so on (Garvin2002).

Product development is a broadconcept that represents both new andexisting products. A new product canrepresent product repositioning, product

JOHNSON 471

line extension, product modifications, anew product category for the firm, or apreviously nonexistent product to theworld (Cooper, Edgett, and Kleinschmidt2001; Crawford 1994; Palmer and Cole1995). Most importantly, scholars havenoted that, “this activity [new productdevelopment] does not always requirethe creation of a new organization. . . a

new product launch/market extensionvis-à-vis an “innovation” is not the samething as a new corporate venture”(Greene, Brush, and Hart 1999).

Independent Ventures. It may initiallyseem logical to equate an ICV to an inde-pendent venture (or start-up) given that anew ICV is a start-up within the context

Table 1Internal Corporate Venture Criteria Review

PresentProposal

1. Originated internally2. Considered distinct from company’s existing products/

services3. Intended as or became a new business4. Pursued within past seven years

Thornhill andAmit (2001)

A business was considered new if it had developed any three ofthe following:1. New markets2. New methods of distribution3. New products/services4. New technology

Zahra andGeorge(1999)

1. The firm had to be in existence for eight years or less2. The firm had to be headquartered in the United States3. The venture had to be active in one or more of the major

areas that constituted the biotechnology industryMcGrath

(1995)1. Attempt to enter a new market2. Sell a new product/service3. Employ a significantly different process, utilizing internal

resourcesDay (1994) 1. Start-up that originated internally

2. New to firm on at least two dimensions (product, market,technology)

3. Requires significant investment of company resourcesbeyond expenditure year

Includes new products based on new technologies that woulddisplace a company’s existing product. Does not includeproduct extensions or brand introductions.

Miller,Gartner,and Wilson(1988)

A business marketing a service or product that:1. The parent company has not previously marketed2. That required the parent company to obtain new

equipment, people, or knowledgevon Hippel

(1977)1. Developing a new product2. Bringing it to market3. Carrying it through at least it initial phases of marketplace

activity


of the established parent firm. The ICVcan have a venture champion and theindependent venture has its entrepre-neur. However, ICV champions tend tohave access to valuable resources andsupport from existing parent firms. Incontrast, the independent entrepreneurlacks the “deep pockets” of a parentorganization, but enjoys more decision-making freedom without the bureau-cracy of a parent organization. It is notuntil investors demand it that most entre-preneurs even engage in formal strategicplanning, compose written businessplans, and/or address other hurdles (Hoy2006).

When Chrisman, Bauerschmidt, andHofer proposed their model for the per-formance of independent ventures, theystated that “the determinants of perfor-mance of a new [independent] ventureand an established business are nearlyidentical” (emphasis added) (Chrisman,Bauerschmidt, and Hofer 1998). In theirmodel of independent venture perfor-mance, their referenced studies includedcorporate ventures, as well as indepen-dent ventures. Similarly, in Sandberg andHofer’s (1987) review of other ventureperformance studies, we find that manyscholars had relied upon the PIMS (ProfitImpact of Marketing Strategy) database(that is, product development data),venture capital projects, or independentstart-ups (see Table 2, which illustratesthis overlap in earlier work). Indeed,many early proclaimed ICV studies com-bined related phenomena to increase thesample size (but not the representative-ness), which likely led to the inconsis-tency of findings.

External Ventures and EstablishedBusinesses. A common misperceptionis that if we understand how to manageand grow established businesses, thenwe can do the same with ICVs. However,scholars have now revealed that some ofthe skills, capabilities, policies, andstructures that are pertinent to the

success of an established business mayin fact contribute to the failure of a newbusiness (Block 1982; Campbell et al.2003; Tushman and Nadler 1986). AnICV can be completely autonomous likean independent start-up, or it can beintegrated (like an acquisition) into theparent firm and subjected to the samerules and requirements of the existingbusiness(es). This suggests performanceimplications for the venture based on thenature or type of venture autonomy. Inthe next sections, the nature of ventureautonomy for this study is discussed andhypotheses are presented.

The Nature of VentureAutonomy

Autonomy has historically beentreated as a one-dimensional constructwith inconclusive findings drawn fromthis broad view (Rauch et al. 2009). It isa unique characteristic of ICVs and proneto implicit theories based on anecdotalevidence and personal experiences. Forinstance, during interviews, it is notuncommon for respondents to expresstheir belief that more (or less) autonomyis appropriate based on their experi-ences. In practice, autonomy can reflectmany conditions, such as tight versusloose controls, uniform versus tailoredpolicies, and centralized versus decen-tralized processes, which early researchbelieved to impact the performance ofestablished businesses (Block 1982). As aresult, recent researchers now suggestthat autonomy may be best examined asa multidimensional construct (Covin andMiles 2007).

A recent study unintentionally reflectstwo major types of autonomy implicit inthe extant literature when the authorssuggest that venture units need (1) suffi-cient separation and (2) to make theirown investment decisions (Campbellet al. 2003). First, the authors have indi-cated what I call the structural type ofautonomy that addresses the physicallocation and/or separation between the

JOHNSON 473

Tab

le2

1987

Ven

ture

Per

form

ance

Lit

erat

ure

Ove

rvie

wIl

lust

rati

ng

the

Com

mon

Ove

rlap

of

Rel

ated

Phen

om

ena

inEar

lier

Ven

ture

Stu

die

s

Auth

or(

s)Typ

eof

Ven

ture

Dep

enden

tV

aria

ble

sIn

dep

enden

tV

aria

ble

sD

ata

Sourc

eD

ata

Anal

ysis

Big

gadik

e(1

976,

1979

)CV

adiv

isio

ns

RO

I,RO

S,ca

shflow

,m

arket

shar

e,an

dth

ree

oth

ers

Star

t-up

goal

s,en

try

stra

tegy

,m

arket

char

acte

rist

ics,

rela

tednes

sto

par

ent,

com

pet

itiv

ere

action

PIM

Scst

art-up

ques

tionnai

re(b

yB

igga

dik

e)

Corr

elat

ion

coef

fici

ent

anal

ysis

Hobso

nan

dM

orr

ison

(198

3)

CV

div

isio

ns

RO

I,m

arket

shar

eSt

art-up

goal

s,ve

ntu

resp

onso

rs,

mar

ket

char

acte

rist

ics,

entr

yst

rate

gy

PIM

Sst

art-up

dat

abas

eG

raphic

com

par

isons

Sandber

g(1

986)

IVsb

New

ventu

reper

form

ance

(five

cate

gory

scal

e)In

dust

ryst

ruct

ure

,ve

ntu

rest

rate

gy,

entr

epre

neu

rV

entu

reca

pital

pro

ject

pro

posa

lsM

ann–W

hitney

U-tes

tan

dSp

earm

anra

nk

corr

elat

ion

Mac

Milla

nan

dD

ay(1

987)

CV

div

isio

ns

RO

I,m

arket

shar

eM

arket

conditio

ns,

star

t-up

goal

s,ve

ntu

rest

rate

gies

PIM

SSt

art-up

dat

abas

eCorr

elat

ion

anal

ysis

Mac

Milla

net

al.

(198

7)IV

sSa

les,

mar

ket

shar

e,pro

fits

,RO

I,an

dfo

ur

cost

mea

sure

s

Entr

epre

neu

rial

team

,pro

duct

feat

ure

s,m

arket

,an

dfinan

cial

char

acte

rist

ics

Ven

ture

capital

pro

ject

pro

posa

lques

tionnai

res

Clu

ster

,re

gres

sion,

and

fact

or

anal

ysis

McD

ouga

ll(1

989)

CV

div

isio

ns

and

IVs

RO

I,RO

S,m

arket

shar

e,ch

ange

inm

arket

shar

eV

entu

rest

rate

gy,

bar

rier

sto

entr

y,ve

ntu

reori

gin

Ques

tionnai

resu

rvey

Fact

or

anal

ysis

,cl

ust

eran

alys

is,

F-te

sts

Stuar

tan

dA

bet

ti(1

987)

IVs

Initia

lquan

tified

succ

ess

initia

lsu

bje

ctiv

esu

cces

sM

arket

char

acte

rist

ics,

pro

duct

feat

ure

s,ve

ntu

rest

rate

gy,

ventu

reorg

aniz

atio

n,

ventu

rele

ader

ship

Ques

tionnai

resu

rvey

Step

wis

ere

gres

sion

anal

ysis

a CV

,co

rpora

teve

ntu

re.

bIV

,in

dep

enden

tve

ntu

re.

c PIM

S,Pro

fit

Impac

tof

Mar

ket

ing

Stra

tegy

dat

abas

e(i

.e.,

pro

duct

dev

elopm

ent

dat

a).


venture and existing business units. Simi-larly, others have asserted that divisionscharged with identifying, launching, andgrowing promising new businesses inter-nally should do so by “creating protectedentrepreneurial islands, with their owndistinctive modes of operation, in themidst of the traditional organization’sformal structures and processes” (Garvin2002, p. 9). In other words, “sufficientseparation” and “entrepreneurial islands”both suggest that the ventures shouldbe self-contained business units whoseoperations are independent of the otherbusinesses. Second, the authors indicatewhat I consider the planning type ofautonomy. Planning autonomy addressesthe basic ideas of planning and control,such as setting goals, strategy, and evalu-ating progress.

In this study, I propose these twotypes (or dimensions) of ventureautonomy—VSA and VPA. Together,they reflect the physical and operationallinks between the venture and theparent, as well as the freedom of themanagers to establish the plan for theventure. From an analogous personalperspective, autonomy could reflectwhether your office is located at theheadquarters or at home; and whetheryou had the freedom to set your ownhours and goals. In turn, your productiv-ity might be impacted by these types andlevels of autonomy afforded you in yourwork.

Venture Autonomy HypothesesOur own experience-based bias favors

more autonomy. Given the inconsistencyof earlier findings, in this section, we seeboth the support and nonsupport foundin the literature based upon the availablesample at the time beginning with thesupport for planning autonomy.

Planning Autonomy Support. A directassociation was found between innova-tion and the commitment of manage-ment, as well as between management

attitudes and the establishment of aninnovative environment (Karagozoglu1988; Waters 2000). A strong correlationwas also reported between ventureperformance and parental separation(Birkinshaw, Batenburg, and Murray2002). The idea was that managersneeded to protect the ICV by making itindependent of the parent as quickly aspossible. Based on observations from thePIMS database, scholars suggested thatcorporate management should take a“hands-off” approach to ventures (Millerand Camp 1985). Accordingly, researchhas supported “independent” decision-making for new ventures, and attributedthe difference between success andfailure to management, finding greatersuccess when corporate managementstayed out of the way (Birkinshaw,Batenburg, and Murray 2002; Block andMacMillan 1993; Ginsberg and Hay 1994;MacMillan, Block, and Subbanarasimha1986; Sykes and Block 1989).

Similarly, some scholars have assertedthat venture-level managers need to beable to define the rules of the game anddirect expectations for their ventures(Ginsberg and Hay 1994). Tidd andTaurins (1999) stated that there is often aconflict between the parent’s goals andthe venture’s activities. Many scholars,even those initially biased towardventure–parent integration, have foundsupport for separating the developingventure from the parent (Birkinshaw,Batenburg, and Murray 2002). In a sub-sequent study using the PIMS database,Miller, Spann, and Lerner (1991) foundthat a high level of direct corporateinvolvement actually reduced perfor-mance in terms of both lower productquality and higher costs. The overallimplication is more authority for venturemanagement and less intrusion from cor-porate management. Therefore, wehypothesize:

H1a: The degree of VPA will be positivelyassociated with venture performance.

JOHNSON 475

Planning Autonomy Nonsupport. Theprevailing argument has been to shieldthe emerging business from the corpo-rate bureaucracy. However, although webelieve that venture managers need to befree (from parental distractions) to adaptand modify venture operations as newmarket information becomes available,the overall objectives and expectationsstill have to be agreed upon by theparent organization. The venture canalso benefit tremendously from theknowledge, experience, and guidance ofthe parent in much the same waythe corporation might benefit from thewisdom of its board of directors. Thesuggestion that venture managers shouldhave autonomy to define the rules (Gins-berg and Hay 1994) may neglect theexperience of senior corporate-levelexecutives, as well as create hostilitybetween venture managers and the man-agers of existing business units. Manage-ment should provide the venture somefreedom but must also pre-empt the con-flicts that can arise (Tidd and Taurins1999). Overall, there are no solid reasonsto believe venture-level managers will bebetter equipped to establish a successfulventure in the absence of corporate-levelinput. It is more likely that the moreinformed the guidance senior executivesprovide to venture managers, the betterthe venture will perform.

Finally, given that the parent’s assess-ment ultimately will determine the fate ofthe venture, it is prudent to have moreparental input from the corporate-levelexecutives in the planning process. Inessence, the ICV is not linked to theparent (after all, it is a “new” venture),but the parent should remain highlyinvolved in its development. Likewise,the ICV managers should welcome andcapitalize on the parental involvement,especially in planning activities and deci-sions so that performance expectationsare agreed upon, realistic, and/or moreeasily modified, if necessary, as moredetails of the targeted market of the ICV

become available. Thus, the argument isto not give planning autonomy to theventure but to have the planning directlydependent upon the corporate evalua-tors. Therefore, we hypothesize:

H1b: The degree of VPA will be negativelyassociated with venture performance.

Structural Autonomy Support. Asnoted, a popular assumption is that newventures need to be free from the stan-dard bureaucratic requirements in orderto adapt to changing market demands.Scholars believed that the many corpo-rate policies, procedures, and structuresdesigned to facilitate the management ofexisting businesses actually impeded thedevelopment of new ventures (Luther1984). Thornberry (2003) pointed outthat some companies spin out their ven-tures so that they will not be constrainedby established company policies andprotocols. In a study of ventures in fourlarge firms, Dougherty (1995) found thatnew product ventures that continued tooperate within the structure and pro-cesses established for existing businessunits failed. Consequently, the argu-ment is made that ventures should bestructurally autonomous. Therefore, wehypothesize:

H2a: The degree of VSA will be positivelyassociated with venture performance.

Structural Autonomy Nonsupport. Instark contrast to the argument for struc-tural autonomy is the consideration thatfamiliarity, experience, and active spon-sors with intimate knowledge of thebusiness may drive venture success. Forinstance, research has shown that sepa-ration bred conflict within many organi-zations (Birkinshaw, Batenburg, andMurray 2002; Garvin 2002, 2004;Sobkowiak 2002). Similarly, there hasbeen considerable research suggestingthat separation does not support ventureperformance (Birkinshaw, Batenburg,


and Murray 2002; Campbell et al. 2003;Garvin 2002, 2004; Luther 1984). Giventhe potential for increased organizationconflict and the lack of support specifi-cally for increased venture performance,we hypothesize:

H2b: The degree of VSA will be negativelyassociated with venture performance.

Venture Stages ofDevelopment

There are arguments both for andagainst autonomy. Drawing from theknowledge of independent ventures, webelieve that these mixed findings are dueto the various stages of development thata venture undergoes over time. As anindependent venture develops, what ittakes to continue its development alsochanges (Boeker and Karichalil 2002).Research has supported that organiza-tions do experience stages or phases ofdevelopment with different emphasis orchallenges at different stages (Drazin andKazanjian 1990; Kazanjian and Drazin1989, 1990). Extrapolating from these,we expect potential changes in theautonomous needs of an ICV as it deve-lops as well.

For instance, the degree of internalsupport and other venture needs shouldchange as management’s understandingof the venture and its targeted marketsincrease. Recent research has analyzedthe Corporate Entrepreneurship Assess-ment Instrument (CEAI) used to evaluatethe entrepreneurial readiness of corpora-tions (Hornsby, Kuratko, and Zahra2002) and presented many factors as wellas challenges regarding new venture pur-suits. Some of the factors presentedincluded the need for internal support(George and MacMillan 1985), knowl-edge and understanding of competi-tive markets (Zahra, Neubaum, andEl-Hagrassey 2002), and market attrac-tiveness (Chandler and Hanks 1994).Quite possibly, performance expecta-tions and satisfaction would also vary

with the venture’s stage of developmentas a function of market feedback. Ingeneral, the venture will need individu-als with diverse skills (Ensley, Pearson,and Amason 2002), and market-experienced personnel (Garvin 2002).

Consistent with Thornhill and Amit(2001), we have three stages of venturedevelopment: early stage (i.e., year offirst financial investment in the venture),middle stage (i.e., year the venturebegan to generate revenue), and estab-lished stage (i.e., year the venturebecame profitable). Perhaps one of themost critical stages for an independentventure is the early stage (or start-up). Itis a stage during which an independentventure must survive on limited start-upcapital. During this time, the entrepre-neur must acquire customers andenough revenues to support its opera-tions before the start-up capital isdepleted. In terms of its structure, weexpect the venture to be separated fromthe existing business so that it may reactquickly to market developments withoutthe corporate bureaucracy impeding itsgrowth.

In the early stage of development,the decision to pursue the venture, thebudget for the venture, as well as themotivation, style of management, strate-gies, objectives, and expectations, mustbe established by the parent andaccepted by the venture manager. Manyunknowns will also begin to beanswered, and the parent needs tomonitor the emerging venture as a result.Given the negative association betweenmature ventures versus the need forparental resources (Thornhill and Amit2001), once the venture matures into themiddle stage, we believe more autonomymay be appropriate. Structurally, oncethe venture reaches maturity, theparent should remove autonomy andintegrate the venture into the existingportfolio of businesses. In summary, thestage of development for the venturewill moderate the relationship between

JOHNSON 477

performance and autonomy. Therefore,we hypothesize:

H3: The relationship between autonomyand performance will be moderatedby the stage of development for theventure.

Research DesignThis research recognizes the trade-offs

of early research designs and seeks toimprove upon them where possible.Since executives can define their busi-ness development activities in terms ofcorporate venture capital, acquisitions,alliances, product developments, exten-sions, new products, and so on, withouta concise definition, researchers riskobtaining samples that are not represen-tative of the phenomena. Indeed, as dis-cussed in the literature review from vonHippel in 1977; Miller, Gartner, andWilson in 1988; Day in 1994; McGrathin 1995; Zahra and George in 1999;Thornhill and Amit in 2001; to Miles andCovin in 2002, we see the criteria used todefine the ICV varied (Cooper, Edgett,and Kleinschmidt 2001), and that therewas considerable overlap of related phe-nomena (Table 2). Consequently, equivo-cal results emerge from the studies. Inthis study, data were collected in amanner to address this and otherresearch design limitations in earlierstudies.

Data CollectionFactiva was used to create a list of 78

corporations across multiple industries,listed on the NASDAQ or NYSE, with atleast 20,000 employees, and over $1billion in sales. The purpose was to iden-tify companies with a history of internalventuring and to obtain a variety of dif-ferent internal venturing experiences.There were no intentional boundary con-ditions (i.e., limitations placed on theassumed applicability of our hypoth-eses). This is not an industry-specifictheory, but a theory for ICV performance

across industries and businesses. Somecompanies required that neither thecompany nor the respondents be explic-itly identified. Nevertheless, the data setincludes a fairly diverse set of industriesand businesses—automotive, oil and gas,chemical, consumer products—and rep-resents companies such as 3M, Whirl-pool, and P&G. Geographically, the dataset includes several U.S. states: Illinois,Indiana, Ohio, Kentucky, Michigan, Wis-consin, Minnesota, California, Georgia,Virginia, and Missouri, for a total of 38ICVs. Data regarding ventures that didnot meet the strict ICV definition andcriteria were rejected, resulting in asmaller but more representative sample.

The corporations were then contactedand screened for a history of pursuingnew business investments given that cor-porations are typically consistent in theirinvestment types (Amit, Livnat, andZarowin 1989). Fourteen of the corpora-tions either directly stated that they didnot pursue new ICVs, or it was deter-mined that they did not, as defined inthis study. Of the 64 remaining, a total of16 participated (25 percent responserate), which exceeds the response ratefor some new venture studies (McDou-gall and Robinson 1990).

To focus clearly on ICVs, respondentsfrom the participating companies wereasked to identify ventures that originatedinternally; were considered distinct fromthe company’s existing products or ser-vices; and were intended as or becamenew businesses. This built upon the cri-teria used in previous studies (seeTable 1) to eliminate or minimize con-tamination from related phenomena.Without a clear and explicit definition,researchers risk obtaining large samplesthat are not representative of the phe-nomena, since executives can definetheir business development activities interms of corporate venture capital, acqui-sitions, joint ventures, and strategic alli-ances, as well as product developmentand extensions. Ventures were also


limited to an age of no more than sevenyears, which is within the range of agesaccepted for defining new ventures(Lussier 1995; McDougall 1989; Shrader,Oviatt, and McDougall 2000; Zahra 1996;Zahra, Ireland, and Hitt 2000).

A final important characteristic of theresearch design was that the data werecollected using a two-part survey supple-mented by interviews. The independentdata were collected from the venturemanagers, who had the detailed knowl-edge regarding the operations of theventure, whereas the performance datawere collected from the senior execu-tives responsible for overseeing the busi-ness development activities and betterable to report the parent’s assessment ofventure performance. Some of thevarious positions and titles of respon-dents are shown in Figure 1. Theirrespective questionnaires were com-pleted independently, clearly separatingthe collection of dependent variable datafrom the independent variable data.

Sample SizeWithout enough cases, one must

either reduce the model or use tech-niques to compensate for size. When toofew cases exist, results are difficult tofind and can become highly sample spe-cific (the extreme is called a “casestudy”). When too many cases exist, find-ings can actually be of little value as well.In fact, it is often forgotten that a samplesize can be too large as well as too small,leading to type 1 and type II errors (i.e.,rejecting the truth versus accepting a lie).In both situations, the issue is exacer-bated if the sample is not representativeof the phenomenon. The sample sizeneeds to be large enough to detectimportant differences with high prob-ability. In other words, we need tocollect enough evidence to avoid con-demning an innocent man by rejectingthe presumption of innocence (Vogt1999). However, if a sample sizebecomes too large, even unimportant dif-

ferences (the untruths) can become sta-tistically significant with high probability(this reflects the saying, if we hear a lieoften enough, we might begin to think ittrue). The general assumption is that weneed a ratio of 20 to 1 in terms of samplesize to variables examined. Although thisnumber is desirable, it is actually muchhigher than necessary for statisticalanalysis as noted by statisticians.

Figure 1Positions and/or Titles of

Respondents

Business development manager

Chief executive officer

Chief financial officer

Director

Director of new product integration

Executive vice president

Front-line leader

General manager

Group vice president

Innovation consultant

Marketing consultant

President

Product/project manager

Senior vice president of business development

Vice president

Vice president of business development

Vice president of marketing

Vice president of innovation

JOHNSON 479

The general rule is that the ratioshould never fall below 5 to 1,meaning that there should be fiveobservations for each independentvariable in the variate. As this ratiofalls below 5 to 1, the researcherencounters the risk of “overfitting”the variate to the sample, makingthe results too specific to thesample and thus lacking genera-lizability (Hair et al. 1998, p. 166).

In his discussion of regressionmodels, Long (1997, p. 54) also states, in“the literature on the covariance struc-ture model, the rule of at least five obser-vations per parameter is often given.”Unfortunately, it seems that we will oftensacrifice the quality of our sample inorder to achieve a greater quantity ofsample. Similarly, we ignore observablepatterns in search of p-value confirma-tions. In this study, efforts have beenmade to improve upon both of theseissues.

VariablesFace and content validity were

checked by a group of researchers and

executives. Confirmatory factor analysiswas also conducted and showed properloadings, with all loadings below the0.600 threshold suppressed. The varimaxrotation method was used, as is standard,with the rotation converging in 10 itera-tions. Overall, the research variables dis-played good psychometric properties,which included strong Cronbach alphasof 0.78, 0.89, and 0.94 for VSA, VPA, andICV performance, respectively. Table 3provides the descriptive statistics andcorrelations.

Internal Corporate Venture (ICV) Perfor-mance. ICV performance was derivedfrom a measure by Thornhill and Amit(2001), and measured based on agree-ment with the following statements ratedon a seven-point Likert scale (1 = stronglydisagree to 7 = strongly agree):

(1) This venture generally meets (ormet) the expectations of the parentcorporation.

(2) The parent corporation views (orviewed) this venture as being suc-cessful overall.

Table 3Descriptive Statistics and Correlations

Variable Mean S.D.a 1 2 3 4 5

1 Internal CorporateVenturePerformance

4.378 1.64 1

2 Firm Size (log) 4.807 0.31 -0.114 13 Venture Age 4.546 2.94 0.011 0.432** 14 Venture Structural

Autonomy3.575 1.91 -0.131 0.143 0.291* 1

5 Venture PlanningAutonomy

3.985 1.43 -0.398** 0.026 -0.076 -0.025 1

aS.D., standard deviation.*Significant at 0.10 level (two-tailed).**Significant at 0.05 level (two-tailed).


(3) The parent corporation believes (orbelieved) that this venture achievedall of its milestones (i.e., eventscrucial to the venture’s successfuldevelopment) on schedule for eachstage of its development.

(4) This venture is performing (or per-formed) well in terms of the criteria(e.g., market share, returns, satisfac-tion, learning, or positioning) thatthe parent corporation considers (orconsidered) most important to theventure’s success.

Venture Structural Autonomy. VSArepresented the extent to which the ven-ture’s operations were linked to those ofother businesses of the corporation andwas measured based on agreement withthe following statements rated on aseven-point Likert scale (1 = strongly dis-agree to 7 = strongly agree):

(1) The venture operates as a self-contained business unit, with fewor weak structural or process link-ages with other businesses of thecorporation.

(2) The venture’s operations are notsignificantly constrained or dictatedby formal structural or process link-ages with other businesses of thecorporation.

(3) The venture operates in an indepen-dent manner vis-à-vis other busi-nesses of the corporation.

Venture Planning Autonomy. VPA rep-resented the extent to which the ven-ture’s management team was responsiblefor establishing goals, timetables, andstrategy for the venture. The questionstem for VPA was: “Who is responsiblefor each of the following planningand/or control related activities?” Plan-ning autonomy for the venture was thenmeasured based on the response to thefive items below using a scale rangingfrom 1 to 7, where 1 = the sole respon-sibility of a higher level(s) (e.g., CEO and

Corporate) of authority within the corpo-ration; 4 = equal responsibility ofventure-level and a higher level(s) ofauthority within the corporation; and,7 = the sole responsibility of venture-level management:

(1) Setting the venture’s goals.(2) Establishment of a timetable (if

applicable) for the achievement ofthe venture’s goals.

(3) Choice of formal criteria used tomeasure the venture’s performance.

(4) Formation or formulation of theventure’s business strategy.

(5) Decision to change (if necessary)the venture’s business strategy.

Controls. Three stages of venture devel-opment were used (Thornhill and Amit2001): Early stage (defined as the year offirst financial investment in the venture),middle stage (defined as the year theventure began to generate revenue), andestablished stage (defined as the year theventure became profitable). Whereassome ventures remain at a particularstage for an extended period, other ven-tures become established rather quickly.Therefore, controls were provided forthe potential effects of venture age,defined as the length of time the venturehas been established, the time since ter-mination, or the time since first financialinvestment, accordingly. Parent size wasmeasured by using the number ofemployees, as has been done in recentstudies to avoid potential complicationswith revenues (Lee, Lee, and Pennings2001; McGrath 2001).

As a test of robustness, a new hybridcontrol variable was created using a tech-nique that increases the power of smallsamples. The new hybrid control wascreated based on the original controls(e.g., firm size, venture age, and venturestage), which were then regressed oneach independent variable, and the stan-dardized coefficients used to computethe hybrid control. Thus, if testing five

JOHNSON 481

variables, then new controls would haveto be computed (one for each variable).The hybrid control for variable numberone would be: IV1 = b1 ¥ CV1 + b2 ¥CV2 + b3 ¥ CV3 + . . . bn ¥ CVn containingall the respective variance, which thenallows one hybrid control to be enteredin place of five original controls. A newregression is then run using the hybridcontrol variable.

Analysis and ResultsAutonomy and Performance

Despite having a sufficient samplesize (see Research Design), to corrobo-rate any effects that might be questioned,the new hybrid control variable wascreated (see Variables: Controls). Variousanalyses were then rerun using thehybrid control variable. This approachcan reduce R2 values and result in fewerfitting models, but, in return, provides arobustness check.

The analyses began with a graphicalexamination of the hypothesized rela-tionships between autonomy and perfor-mance. Graphical analysis is among theoldest and most versatile of analyticaltechniques, and has long been usedand accepted in a variety of research(Bernstein and Cowden 1937; Chan,Makino, and Isobe 2006; Ruamsook,Russell, and Thomchick 2007). TheJournal of Computational and GraphicalStatistics published by the American Sta-tistical Association (an establishedauthority in statistical techniques) isdevoted to extending the use of thisoften-misunderstood technique in statis-tics. The initial graphical analyses dis-played a negative relationship betweenVPA and ICV performance (Figure 2) insupport of H1b, but a curvilinear (orinverted-U) relationship between VSAand ICV performance in support of H3(Figure 3).

Figure 2Graphical Analysis of Overall Relationship between

Venture Planning Autonomy and Performance

7.006.005.004.003.002.001.00

Venture structural autonomy

7.00

6.00

5.00

4.00

3.00

2.00

1.00

ICV

per

form

ance

Early stage

Middle stage

Established stage

Overall

R2 quadratic = 0.239





Using the regression curve estimationanalysis of IBM SPSS Statistics package19, VSA, as displayed in Figure 3, has a R2

quadratic of 0.239, VSA beta coefficient of1.998 (significance = 0.015), and a VSA-squared beta coefficient of -2.166 (signifi-cance = 0.009), indicating the strongcurvilinear relationship. In comparison,VPA, as displayed in Figure 2, does notdisplay a significant curvilinear relation-ship (VPA beta significance = 0.516, andVPA-squared beta significance = 0.907).However, for VPA, we do see a significantlinear relationship (R2 linear = 0.210, VPAbeta coefficient = -0.459, significance =0.015). For further analysis, the VSAvariable was squared within a standardregression model, showing that thesecond order effects indeed reachedsignificance (beta = -2.294, p < .05). Thisrelationship was also tested using the

hybrid control and held (beta = -2.122,p < .05) (Table 4: models 1 and 2). Thus,the results of graphical analytical tech-niques are consistent with standardmethods, with the added benefit ofinstant visual information, confirmation,and interpretation.

Main Effects and Interactions. Usinglinear regression in SPSS, controls werefirst entered, followed by VSA and VPA,and multiplicative interaction variables.Given the findings from the graphicalanalysis, the task was to examine eachventure stage independently. Therefore,venture stage was used as a selectionvariable, which is a direct method toanalyze stages that can provide a morefine-grained analysis similar to somemore common approaches, such as mod-erated regression.

Figure 3Graphical Analysis of Curvilinear Relationship between

Venture Structural Autonomy and Performance and Stage ofDevelopment (Early, Middle, and Established) Trend Effects

ICV

per

form

ance

JOHNSON 483

Tab

le4

Sta

ndar

dC

urv

ilin

ear

Tes

tan

dH

ybri

dR

obust

nes

sC

hec

kfo

rP

erfo

rman

cean

dV

entu

reStr

uct

ure

Auto

nom

yR

elat

ionsh

ipR

egre

ssio

nR

esult

s

Var

iable

Sta

ndar

dM

odel

Acr

oss

Ven

ture

Sta

ges

Hyb

rid

Model

(Robust

nes

sC

hec

k)

Contr

ol

Model

1H

ybri

dC

ontr

ol

Model

2

BB

eta

BB

eta

BB

eta

BB

eta

Const

ant

3.77

1-0

.598

4.67

3(0

.385

)3.

099

(0.9

21)

(5.2

05)

(5.1

55)

Firm

Size

0.21

30.

046

0.80

00.

172

-0.0

49a

(0.2

65)

-0.0

38a

-0.0

11(0

.267

)-0

.009

(1.1

42)

(1.0

78)

Ven

ture

Age

-0.0

34-0

.740

-0.0

67-0

.144

(0.1

13)

(0.1

10)

VSA

b1.

464

2.16

0**

1.33

8(0

.615

)1.

975*

*(0

.639

)V

SA^2

-0.2

06(0

.084

)-2

.294

**-0

.191

(0.0

81)

-2.1

22**

R2

0.00

40.

232

0.00

10.

209

Adju

sted

R2

-0.0

830.

085

-0.0

400.

102

F-St

atis

tic

0.04

63.

113*

0.03

52.

892*

Stan

dar

der

ror

inpar

enth

esis

.a H

ybri

dco

ntr

ol.

bV

SA,

ventu

rest

ruct

ura

lau

tonom

y.*S

ignifi

cant

at.1

0le

vel.

**Si

gnifi

cant

at.0

5le

vel.


The first regression analysis was con-ducted across all venture developmentstages as with earlier research. Consistentwith the graphical analytical technique,VPA was negative and significant provid-ing strong support for H1b (beta =-0.606, p < .01) (Table 5, model 3). VSA,venture age, and firm size were notsignificant. There was no interactioneffect.

Next examined were ventures that hadprogressed beyond the early stage. Thenature of expectations might be less strin-gent during the early development stageversus later stages, thereby influencingperformance perceptions. VPA remainednegative and displayed a slightly greatereffect size in later venture developmentstages (beta = -0.685, p < .01) (Table 5,model 4). VSA and size remained notsignificant, whereas venture age reachedsignificance (p < .10).

Given the consistency of the VPAresults, the effect of VSA was partialedout, and another look taken of VPA acrossdevelopment stages. VPA remained sig-nificant (beta = -0.615, p < .01). Addition-ally, during later stages the effect size ofVPA (beta = -0.685, p < .01) and age bothincreased. With the hybrid control, theprevious results continued to hold (VPA:beta = -0.549, p < .01; VSA: not signifi-cant). The significant model accountedfor 20.6 percent of the variance (Table 5,model 5), adding to the strong support forVPA, H1b. Likewise, the lack of a directeffect from VSA provides support for thebasic premise that there are differenttypes of autonomy, and that the nature ofthe relationship between autonomy andICV performance is a function of the typeof autonomy.

In summary, H1b (VPA is negativelyassociated with venture performance)was supported, and H3 (the stage ofdevelopment of a venture moderates therelationship between performance andautonomy) was supported. H2 regardingVSA was not supported. However, it wasdiscovered that there is a relationship

between VSA and ICV performance toconsider within specific venture develop-ment stages. For instance, a positiveassociation between VSA and ventureperformance (similar to H2a) occurs withventures in the middle stage of develop-ment, whereas a negative associationwith venture performance (similar toH2b) occurs with ventures when theyreach the established stage. The earlystage displays an inverted-U relationship.

DiscussionEvery year, numerous companies, like

Kraft, Disney, 3M, as well as smaller,lesser known companies, seek to developnew businesses known as ICVs withintheir established businesses. Among themyriad of decisions that have to be madefor the successful development of the ICVis the decision regarding autonomy.Among many scholars, it is believed thatthe lack of autonomy for the venture leadsto venture failure (Birkinshaw, Baten-burg, and Murray 2002; Block and Mac-Millan 1993; Ginsberg and Hay 1994;MacMillan, Block, and Subbanarasimha1986; Miller and Camp 1985; Sykes andBlock 1989). Indeed, many of these earlystudies represent groundbreaking workthat has been instrumental in our initialunderstanding of corporate venturing.Yet early empirical studies reflect equivo-cal results, and ICV performance contin-ues to be random at best.

This study seeks to advance theresearch by empirically examining somepopular assumptions using a more rep-resentative sample of the phenomenon.This paper focuses on different types ordimensions of autonomy that historicallyhas been assumed to be a broad one-dimensional construct that positivelyimpacts ICV performance. However, twotypes of autonomy are distilled from theextant literature. Measures were subse-quently developed, tested, and found toproperly load on the two proposed types(or dimensions) of autonomy: VPAand VSA. These two types of venture

JOHNSON 485

Tab

le5

Per

form

ance

and

Ven

ture

Auto

nom

y(S

truct

ura

lan

dP

lannin

g):

Lin

ear

Reg

ress

ion

Res

ult

s

Var

iable

Acr

oss

All

Ven

ture

Sta

ges

Aft

erEar

lyV

entu

reSta

ge

Per

form

ance

and

Ven

ture

Pla

nnin

gA

uto

nom

yw

ith

Hyb

rid

Contr

ol

Tes

t

Contr

ol

Model

3C

ontr

ol

Model

4C

ontr

ol

Model

5

BB

eta

BB

eta

BB

eta

BB

eta

BB

eta

BB

eta

Const

ant

4.02

96.

223

0.95

94.

251

4.85

57.

076

(5.2

11)

(4.2

62)

(6.2

22)

(5.4

13)

(0.2

83)

(0.8

51)

Firm

size

0.27

50.

066

0.50

40.

121

0.97

30.

203

0.88

30.

184

0.09

8a0.

023a

0.70

6a0.

167a

(1.1

55)

(.94

3)(1

.357

)(1

.140

)V

entu

reag

e-0

.097

-0.2

41-0

.115

-0.2

86-0

.106

-0.2

86-0

.189

-0.5

08*

(0.9

43)

(0.8

50)

(0.1

11)

(0.1

01)

(0.1

05)

(0.1

03)

VSA

b-0

.206

-0.3

18-0

.015

-0.0

23(0

.128

)(0

.146

)V

PA

c-0

.575

-0.6

06**

*-0

.565

-0.6

85**

*-0

.521

-0.5

49**

*(0

.175

)(0

.171

)(0

.191

)R

20.

044

0.47

00.

063

0.47

40.

001

0.28

2A

dju

sted

R2

-0.0

570.

345

-0.0

540.

323

-0.0

490.

206

F-St

atis

tic

0.43

63.

771*

*0.

537

3.15

2**

0.01

13.

722*

*

Stan

dar

der

ror

inpar

enth

esis

.a H

ybri

dco

ntr

ol

vari

able

crea

ted.

bV

SA,

ventu

rest

ruct

ura

lau

tonom

y.c V

PA

,ve

ntu

repla

nnin

gau

tonom

y.*S

ignifi

cant

at0.

10le

vel.

**Si

gnifi

cant

at0.

05le

vel.

***S

ignifi

cant

at0.

01le

vel.


autonomy and the impact of venturestages are then analyzed in terms of theirrelationship with ICV performance givena sample of ICVs without relatedphenomena. The results are a newunderstanding of the autonomy–ICVperformance relationship.

VPA FindingsThe differences between what the

parent wants and what the venture candeliver creates conflicts that are believedto adversely impact performance lead-ing to conclusions for a “hands off”approach or, in other words, more VPA.Thus, popular belief is that an ICV is likean independent start-up and succeedswhen separated from the parent organi-zation and the venture managers are ableto exercise independent decision-making(Birkinshaw, Batenburg, and Murray2002; MacMillan, Block, and Subba-narasimha 1986; Sykes and Block 1989).

However, contrary to the popularbelief to provide more VPA, this studyfinds that VPA is negatively associatedwith ICV performance. Rather than allow-ing venture managers to define the rules(Ginsberg and Hay 1994), this study findsthat venture managers should embracethe venture–parent relationship and capi-talize on the needs and benefits of thatrelationship. Ultimately, the ventureexists for the parent organization, andthus we need to align the goals the parenthas for the venture with the actual capa-bilities of the venture, resulting in morerealistic expectations. Proper and cleargoal alignment will also reduce conflicts.This alignment is critical given that ICVsare really inventions of established busi-nesses (i.e., a parent organization), andwhat the parent wants is what it mustdeliver. These internal ventures mightbetter be called “inventures” since, like“inventions,” if they do not show promise,they might be scrapped or placed upon ashelf and virtually forgotten.

In summary, although both argumentsfor and against VPA are plausible, our

analysis of a more representative sampleindicates that less VPA is associated withthe more successful ventures. The effectis strong, significant, and explains from20 to 34 percent of the performance vari-ance (Table 5). Therefore, when it comesto the strategies, operations, and plan-ning, the data show that the moreinvolved the parent the better the perfor-mance of the ICV in terms (such asmarket share, returns, satisfaction, learn-ing, or positioning), which are importantto the parent. Of course, we ask whythese findings are contrary to popularbelief. The anecdotal evidence of thebelief of more autonomy neglects thetype of autonomy, as well as the ven-ture’s stage of development.

VSA FindingsThe second form of venture

autonomy—VSA—considers the linksbetween the venture and existing paren-tal business units. For a new ICV in thepurest sense, there would be no linkages,and the venture would be virtually inde-pendent. As revealed through the sepa-rate interviews, this is an extremely raresituation, since in the pursuit of a newbusiness there always tends to be someconnection with an existing business.Nevertheless, contrary to expectations,there was no significant linear relation-ship between VSA and ICV performance,thereby pouring even more doubt on thepopular belief for more autonomy.

Venture Stage Moderator. VPA is notimpacted by the venture’s stage of devel-opment, but VSA depends upon it.Although the stages of development foran ICV are distinct, research supports theimportance of stages of development orlife cycles for organizations (Drazin andKazanjian 1990; Kazanjian and Drazin1989, 1990). Similarly, we find that thenature of the relationship between VSAand ICV performance changes as theventure develops. A positive association(similar to H2A) occurs with ventures in

JOHNSON 487

the middle stage of their development,whereas a negative association occurs(similar to H2B) with ventures that havereached the established stage. Interest-ingly, during the early stage—perhapsthe most critical for a start-up as well—the relationship displays a more complexinverted-U shape. This reflects the deli-cate balance between the venture andparent in the initial positioning of theventure, and, although unexpected, it isintuitively appealing given that VSA rep-resents the operational linkages for adynamic ICV. The changing nature of therelationship between VSA and perfor-mance may also account for some of thepopular beliefs and inconsistent resultswhen autonomy is treated as a broadconstruct.

For instance, when a new venture isinitially pursued, we are aware of theconflicts that can arise between ventureand other business units (Tidd andTaurins 1999), which may lead us to seekto separate them, but research has alsoshown that separation can breed conflict(Birkinshaw, Batenburg, and Murray2002; Garvin 2002, 2004; Sobkowiak2002). This tension may be in part due tothe venture needing to be both differentand familiar at the same time, andneeding to be both free of the policiesimposed upon existing units (Thornberry2003) while still conforming to the estab-lished protocols for all units by virtual ofbeing a part of the parent organization.Thus, without structural autonomy, theventure is more likely to fail, and withtoo much structural autonomy, theventure is still likely to fail. Optimizationbecomes the key.

During the middle stage when man-agers are perhaps excited about thenew venture’s growth, the venture’sperformance benefits from autonomy.However, as we can see in the graph, thebenefit of additional VSA begins levelingoff and completely reverses upon theventure reaching the established stage(see Figure 3). Some of the findings of

this study are supported by a recent casestudy that suggested that venturesshould be managed in stages with stage-specific reviews, and that new marketsrequire learning by the organization andchanges in direction (McGrath, Keil, andTukiainen 2006). Without an understand-ing of this dynamic, a venture mightsimply by chance quickly complete theearly stage (recall venture age reachedsignificance after the early venture devel-opment stage, indicating a potentialbenefit to younger ventures later) andbenefit from the increasing level ofautonomy being provided to it becausemanagers were observing performanceimprovements unaware of the changingrelationship. Unfortunately, the ICV willthen encounter an unanticipated obstaclefrom the high level of autonomy; and, ifnot corrected, fail to become established.It will be terminated or abandoned.

Autonomy Types and Stagesin Action

Given that the findings of this studyare contrary to popular belief, it leads toa question of how can we reconcile theseresults? The differences are likely theresult of premises of the study asreflected in three issues: (1) having anuncontaminated sample, (2) distinguish-ing the different types of autonomy, and(3) considering the stage of the venture.Similarly, “Is there evidence in practiceof what these findings show?” Of course,evidence opposing popular belief isoften not apparent—otherwise it wouldprobably not be “popular.” Nevertheless,the need for VPA to reside at thecorporate-level (i.e., low VPA), and thedependence of VSA (e.g., operationalprotocols) on the venture stage isreflected in some recent history andevents.

For example, given its domination ofthe animation industry, the Walt DisneyCompany became complacent in itssuccess. Unfortunately, rather than con-tinue to internally innovate into new


characters, Disney invested resourcesinto retaining the Mickey Mouse charac-ter copyright. Disney won the copyrightextension (known as the Digital Millen-nium Copyright Act), but lost its competi-tive advantage in creativity andanimation. This led Disney to partnerwith a largely unknown, but upcomingcompany known as Pixar, and, eventu-ally Disney even acquired Pixar Studiosonce the alliance contract ended.

Despite the Pixar acquisition, withoutany other strategy, Disney’s downwardspiral would definitely continue if thePixar competencies were lost in theacquisition. However, rather than engagein the typical postacquisition integration,Disney CEO Bob Iger prudently keptPixar intact and gave control of Disney’sanimation operations to the former Pixarexecutives, Catmull and Lasseter. Aco-director at Disney, Mr. Musker, stated,“It’s unusual for Hollywood, making thekey creative people sort of in charge ofthe actual decisions and less the studioexecutives, to some degree” (emphasisadded) (Sanders 2008, B1). Mr. Musker isalluding to VPA, and, per his statement,it was not the norm.

In addition to VPA not being the normat Disney, it still is not. Disney executivesset Pixar within its portfolio and pro-moted the Pixar executives into Disneyexecutives, and gave them control overDisney’s animation operations. Thus,contrary to the impression that theventure managers were now in charge ofdecisions, Disney executives were over-seeing the many animation ideas beingpursued (in other words, the ICVs!).Recall that the corporate-level managersare those responsible for overseeing themany business development activities.The equivalent of venture-level manageris the writer and/or director. Mr. Las-seter, the new “executive” in charge ofDisney’s animation operations, replacedthe writer and director of an originalscript (an ICV equivalent) with two newdirectors (venture-level managers) based

on creative differences (Sanders 2008,B1). These actions by Disney reflect anunderstanding of the nature of ventureautonomy and performance consistentwith this study and contrary to popularassumptions.

LimitationsSteps taken to improve upon the

trade-offs of previous studies have anunfortunate trade-off as well—theyreduce the sample size. Additionally,many corporations tend to be reluctantto disclose their ICV data partly due tothe competitive nature of the infor-mation; partly due to inconsistenciesregarding what constitutes an ICV, andpartly due to their abysmal success rates(executives tend to disassociate them-selves with failures). There are also nopublicly available ICV databases con-taining the needed information forresearch purposes. The overall effect ofthese conditions is a smaller samplethan desired. Although the sample sizeis still sufficient (Hair et al. 1998; Long1997; Vogt 1999) and the sample itselfimproves upon previous studies, thesize presents a serious threat to theresearch by reducing the power ofthe study to find significance. However,despite the sample size, significant find-ings were made. Of course, significantfindings based on a small sample sizeare very compelling and support theneed for further research utilizing largersamples.

There are also biases and conflicts ofinterest to consider. VPA measures theextent to which the venture’s manage-ment team was responsible for establish-ing the goals, timetables, and strategy forthe venture. Among some of the items aresetting the goals and performance mea-surement for the venture. However, aventure manager may not consider theperformance of a venture independent ofhis own performance. Therefore, it ispossible that when venture managersevaluate ICV performance that their self-

JOHNSON 489

assessment could be more biased,whereas the upper-level executives mightnot be as impressed with the achieve-ments of the venture and reflect no per-sonal bias or connection in theirassessment. This negative correlationcould potentially be an artifact of lowVPA. Based on the data, we believe itpoints out the importance of the higher-level management involvement in thecriteria for the performance evaluation.We also believe this is an importantconsideration since the evaluation of theventure at each stage of development andwhether it continues to receive support isultimately determined by the upper-levelexecutives. During this study, perfor-mance information was collected fromboth the executive in charge as well as theventure manager. However, the executivein charge was used in the analyses forpreviously stated reasons.

Finally, there is the issue of survival.Indeed, as discovered in the supplemen-tal interviews, an ICV can becomeorphaned. Other ICVs are metaphori-cally kept on life support because noone wants to “pull the plug.” Thus,despite the best efforts, there arenotable limitations to every study.Although steps were taken to move inthe direction of causality, the associa-tions discovered in this study are notintended to describe causal relationshipsbut to identify potentially key variablesassociated with the phenomenon. Lastly,this is not intended to be an exhaustivelist of limitations but to remind thereader that all research has trade-offsthat we should consider with any con-clusions or subsequent recommenda-tions to management.

Implications for ManagementExecutives responsible for maintain-

ing and growing existing businessesoften look for a “balance between sta-bility and innovation” (Klavans, Shanley,and Evan 1985, p. 26). Of the growthavenues available (Garvin 2002, 2004),

the development of a new business isperhaps among the most rewarding andchallenging. Popular belief is that if youcan run an established business, thenyou can start a new business. However,we know that the skills needed are notalways the same. Additionally, implicitbeliefs appear to collide with businessrealities. Prior studies suggest that man-agers should cocoon new ventures, ortreat them as isolated islands, or takeother steps to shield and protectthe fledgling venture (Garvin 2002;Ginsberg and Hay 1994). However, newdata suggest that the prevailing sugges-tion to give greater autonomy to venturemanagers would not only underminethe venture–parent relationship, butadversely impact ICV performance.Findings from this study show a strongnegative association between VPA andperformance. The findings reveal tomanagers that the goals, timetables,strategy, performance measures, andoverall planning for the venture shouldbe diligently controlled by the parent soas to not undermine future developmentnor create conflict (i.e., sibling rivalry).Furthermore, the degree of operationallinks and protocol constraints shouldtake into account the actual develop-mental stage of the venture to determinethe appropriate role of autonomy. Ofcourse, high corporate-level involve-ment with the venture does not neces-sarily mean that the venture should besubjected to the same policies and pro-tocol for other business units. Highcorporate-level involvement simply rec-ognizes that the ICV needs the thought-ful tutelage of the parent though anactive planning role. The actual struc-ture of the venture, however, is a sepa-rate issue driven by the stage ofdevelopment of the venture.

In the early stages of venture devel-opment, managers search for the balancebetween control and freedom. Datasuggest that managers should put morefocus initially on understanding the


business distinctions and communicatingwhy the venture is of strategic interest.This must be accomplished withoutneglecting to acknowledge its link to theexisting businesses or markets. This linkprovides familiarity and comfort regard-ing the venture as it develops, and, as theventure moves into the middle stage ofdevelopment, managers need to maxi-mize structural autonomy, which willhave to be reversed as the venturebecomes established.

In summary, success depends uponunderstanding the types of autonomy andthe dynamic nature of the autonomy–performance relationship. Autonomyshould be actively monitored and tuned.VPA should be minimized and controlledby corporate executives regardless ofventure stage. However, the developmentstage of the venture must be taken intoaccount when determining the appropri-ate VSA. In the early stage, a carefulbalance of freedom and control must beachieved. During the middle stage, man-agement should shift focus to the distinc-tive value of the ICV and allow it tooperate more independently. In the finalstage, the venture should become a rec-ognized addition to the existing group ofbusinesses operating under the estab-lished protocols. Lastly, as the develop-mental time becomes prolonged and aventure becomes older, many venturesexperience diminishing performance,and, if not shut down, can becomeorphaned. Therefore, corporate- andventure-level managers need to be proac-tive to keep the venture moving forwardand prohibit lingering.

Future ResearchGiven the findings of this study, the

acquisition of larger samples couldprovide a deeper examination of the dis-covered effects of autonomy and venturestage, and provide greater generalizabil-ity. Keeping in mind the importance of arepresentative sample, researchers mightexamine the popular belief that similari-

ties between a venture and parent, spe-cifically in terms of their products andtechnologies, supports ICV performance;or, from a resource-based perspective,researchers might examine whether thetype of resources also impact ICV per-formance when we take into consider-ation the role of autonomy in thoserelationships.

This study found two dimensions toventure autonomy, but we can considernot only their impact on ICV pursuits, butother business development activities.For scholars, the development and dis-covery of two new measures of autonomythat are significant in ICV performanceopens up new research possibilities intoour understanding of internal venturesand other organizations. Furthermore, acase study focused on the experiences ofa single company could provide insightsinto the relationship between autonomyin terms of implementation strategies orspecific types of ICVs.

Future research should examine theneed for organizations to differentiatestructures in the parent–venture relation-ship. A particularly interesting studywould take the curvilinear nature of VSAand examine how organizations adaptstructures that best conform to perfor-mance expectations. Indeed, given thatduring the initial stage of developmentthe venture is adversely impacted byboth too much and too little structuralautonomy, we want to understand theoptimal level of VSA and the drivingfactors. Similarly, future research shouldexamine when an ICV is more likely tobe abandoned versus terminated. Under-standing the drivers of these outcomescould contribute to our understanding ofhow to avoid them. Lastly, we mightexamine the knowledge transfer thattakes place within organizations engagedin corporate entrepreneurship activitieswith a specific focus on ICVs.

Finally, autonomy is not the only vari-able that has been assumed to influenceperformance outcomes. Additional

JOHNSON 491

studies are needed to find and examineother factors. Ultimately, being able toaddress the question of a successful ICVdesign strategy for companies is where Ienvision this line of research taking us.

ConclusionsIn my effort to address several research

design problems found in earlier studiesand to highlight differences betweenthem, and by implication, the contribu-tion of this work, the tendency is tobecome overly critical. However, the con-tribution of prior scholars, which in factinspired this study, is fully recognized andacknowledged. Any discoveries hereinare in part a continuation of those earlyefforts, and, hopefully, in some smallpart, help advance our understanding ofcorporate venturing.

Representative samples are alwayssought, but determining what is repre-sentative and obtaining such a sample isboth challenging and imperative. There-fore, when related phenomena havebeen used and the resulting findings con-tinuously reflect equivocal results, thenwe need to reexamine our researchdesigns. Consequently, the sample crite-ria was derived from a review of theliterature and designed to eliminate con-tamination from related phenomena.Quantity was desired, but quality waspreferred; therefore, data were discardedif determined to be of a related phenom-enon. This step was taken with theexpectation that if the sample is repre-sentative of the truth, then the overallrelationship is more likely to be revealedin the analysis.

By also discerning some of the nuancesof the phenomenon in the extant litera-ture, two forms of autonomy were theo-rized: planning and structural. By deve-loping and examining these two types ofventure autonomy, it was (1) shown thatthere are indeed different types, (2)shown that the type of venture autonomyis independently important to perfor-mance; and (3) discovered that the nature

of some autonomy–performance relation-ships is influenced by the venture’s stageof development. Further improvementswere achieved by collecting the indepen-dent variables from the venture managersand the dependent variable separatelyfrom senior executives.

Consequently, the findings of thisresearch are important to companiesseeking innovative growth and develop-ment. The basic research question exam-ined is: Do ICVs designed with moreautonomy perform better than thosedesigned with less autonomy? I believewe can now offer a preliminary answerto this question based on the type ofautonomy and the venture’s stage ofdevelopment. From the empirical analy-sis of this primary data, we see thatdespite the desire for greater autonomyto establish the goals, timetables, or strat-egy by those who directly manage ICVs,an ICV is better served with lessautonomy. In contrast, the complexity ofstructural autonomy requires manage-ment to be cognizant of the developmentstage of the venture when determiningthe appropriate level of autonomy. Thereis an inverted-U relationship in the earlystage; a positive relationship in themiddle stage; and a negative relationshipat the established stage. This has neverbefore been empirically known, shown,or expected. Overall, the ICV has provento be an intriguing phenomenon withincorporate entrepreneurship, and, givenits value to organizational growth andinnovation, it remains a viable andimportant area for research.

ReferencesAmit, R., J. Livnat, and P. Zarowin

(1989). “The Mode of CorporateDiversification: Internal Venturesversus Acquisitions,” Managerial andDecision Economics 10, 89–100.

Antoncic, B., and I. Prodan (2008).“Alliances, Corporate Technolo-gical Entrepreneurship and FirmPerformance: Testing a Model on


Manufacturing Firms,” Technovation28(5), 257–265.

Baysinger, B. D., and R. E. Hoskisson(1989). “Diversification Strategy andR&D Intensity in Large MultiproductFirms,” Academy of ManagementJournal 32(2), 310–332.

Bernstein, E. M., and D. J. Cowden(1937). “Graphic Presentation ofTrend Data,” Southern EconomicJournal 3(4), 700–709.

Biggadike, R. (1976). “Entry, Strategy,and Performance,” Boston, MA: Doc-toral Dissertation, Harvard BusinessSchool.

——— (1979). “The Risky Business ofDiversification,” Harvard BusinessReview 57(3), 103–111.

Birkinshaw, J. (2005). “The Secret Diaryof Corporate Venturing,” BusinessStrategy Review 16(2), 19–24.

Birkinshaw, J., R. Batenburg, and G.Murray (2002). “Venturing toSucceed,” Business Strategy Review13(4), 10–17.

Block, Z. (1982). “Can Corporate Ventur-ing Succeed?,” The Journal of BusinessStrategy 3(2), 21–33.

——— (1989). “Damage Control for NewCorporate Ventures,” The Journal ofBusiness Strategy 10(2), 22–28.

Block, Z., and I. C. MacMillan (1993).Corporate Venturing: Creating NewBusinesses within the Firm. Boston:Harvard Business School Press.

Boeker, W., and R. Karichalil (2002).“Entrepreneurial Transitions: FactorsInfluencing Founder Departure,” TheAcademy of Management Journal45(4), 818–826.

Brown, S. L., and K. M. Eisenhardt(1995). “Product Development: PastResearch, Present Findings, andFuture Directions,” Academy of Man-agement Review 20(2), 343–378.

Burgelman, R. A. (1983). “A ProcessModel of Internal Corporate Venturingin the Diversified Major Firm,” Admin-istrative Science Quarterly 28, 223–244.

——— (1984a). “Designs for CorporateEntrepreneurship in EstablishedFirms,” California ManagementReview 26(3), 154–166.

——— (1984b). “Managing the InternalCorporate Venturing Process,” SloanManagement Review 25(2), 33–48.

——— (1984c). On the Interplay ofProcess and Content in Internal Cor-porate Ventures: Action and Cogni-tion in Strategy-Making.

Campbell, A., and R. Park (2004). “StopKissing Frogs,” Harvard BusinessReview 82(7/8), 27–28.

Campbell, A., J. Birkinshaw, A. Morrison,and R. Van Basten Batenburg (2003).“The Future of Corporate Venturing,”Sloan Management Review 45(1),30–37.

Campbell, D. T., and D. W. Fiske (1959).“Convergent and Discriminant Valida-tion by the Mutitrait-MultimethodMatrix,” Psychological Bulletin 56,81–105.

Chan, C. M., S. Makino, and T. Isobe(2006). “Interdependent Behavior inForeign Direct Investment: The Multi-level Effects of Prior Entry and PriorExit on Foreign Market Entry,”Journal of International BusinessStudies 37(5), 642–665.

Chandler, G. N., and S. H. Hanks (1994).“Market Attractiveness, Resource-Based Capabilities, Venture Strategies,and Venture Performance,” Journal ofBusiness Venturing 9, 331–349.

Chesbrough, H. W. (2000). “DesigningCorporate Ventures in the Shadow ofPrivate Venture Capital,” CaliforniaManagement Review 42(3), 31–49.

Chrisman, J. J., A. Bauerschmidt, and C.W. Hofer (1998). “The Determinantsof New Venture Performance: AnExtended Model,” EntrepreneurshipTheory and Practice 22(1), 5–29.

Cooper, R., S. Edgett, and E. Kleinschmidt(2001). “Portfolio Management forNew Product Development: Results ofan Industry Practices Study,” R&DManagement 31(4), 361–380.

JOHNSON 493

Covin, J. G., and M. P. Miles (2007).“Strategic Use of Corporate Ventur-ing,” Entrepreneurship Theory andPractice 31(2), 183–207.

Crawford, C. M. (1994). New ProductsManagement. Burr Ridge, IL: RichardD. Irwin, Inc.

Day, D. L. (1994). “Raising Radicals:Different Processes for Champion-ing Innovative Corporate Ventures,”Organizational Science 5(2), 148–172.

Dougherty, D. (1995). “Managing YourCore Incompetencies for CorporateVenturing,” Entrepreneurship Theoryand Practice 19(3), 113–135.

Drazin, R., and R. K. Kazanjian (1990).“Research Notes and Communica-tions: A Reanalysis of Miller and Frie-sen’s Life Cycle Data,” StrategicManagement Journal 11, 319–325.

Ensley, M. D., A. W. Pearson, and A. C.Amason (2002). “Understanding theDynamics of New Venture Top Man-agement Teams: Cohesion, Conflict,and New Venture Performance,”Journal of Business Venturing 17,365–386.

Garvin, D. A. (2002). “A Note on Corpo-rate Venturing and New Business Cre-ation,” Harvard Business School,1–20.

——— (2004). “What Every CEO ShouldKnow about Creating New Busi-nesses,” Harvard Business Review82(7/8), 18–21.

George, R., and I. C. MacMillan (1985).“Corporate Venturing: Venture Man-agement Challenges,” The Journal ofBusiness Strategy 6(2), 85–91.

Ginsberg, A., and M. Hay (1994). “Con-fronting the Challenges of CorporateEntrepreneurship,” European Man-agement Journal 12(4), 382–389.

Glaister, K. W., R. Husan, and P. J.Buckley (2003). “Decision-MakingAutonomy in UK International EquityJoint Ventures,” British Journal ofManagement 14(4), 305–322.

Gomes, J., P. D. Weerd-Nederhof, A.Pearson, and O. Fisscher (2001).

“Senior Management Support in theNew Product Development Process,”Creativity and Innovation Manage-ment 10(4), 234–242.

Gooding, R. Z., and J. A. W. Wagner III(1985). “A Meta-Analytic Review ofthe Relationship between Size andPerformance: The Productivity andEfficiency of Organizations and TheirSubunits,” Administrative ScienceQuarterly 30, 462–481.

Greene, P. G., C. G. Brush, and M. M.Hart (1999). “The Corporate VentureChampion: A Resource-BasedApproach to Role and Process,” Entre-preneurship Theory and Practice23(3), 103–122.

Hair, J. F., Jr, R. E. Anderson, R. L.Tatham, and W. C. Black (1998). Mul-tivariate Data Analysis, 5th ed. UpperSaddle River, NJ: Prentice-Hall.

Henard, D. H., and D. M. Szymanski(2001). “Why Some New Products AreMore Successful than Others,” Journalof Marketing Research 38, 362–375.

Hobson, E. L., and R. M. Morrison (1983).“How Do Corporate Start-up VenturesFare?,” in Frontiers of Entrepreneur-ship Research. Ed. J. A. Honaday, J. A.Timmons, and K. H. Vesper. Welles-ley, MA: Babson College Center forEntrepreneurial Studies.

Hornsby, J. S., D. F. Kuratko, and S. A.Zahra (2002). “Middle Managers’ Per-ception of the Internal Environmentfor Corporate Entrepreneurship:Assessing a Measurement Scale,”Journal of Business Venturing 17(3),253–273.

Hoy, F. (2006). “The Complicating Factorof Life Cycles in Corporate Venturing,”Entrepreneurship Theory and Practice30(6), 831–836.

Karagozoglu, N. (1988). “InnovativeBehavior of Firms in a DevelopingCountry: An Empirical Study,” Engi-neering Management International5(2), 121–129.

Kazanjian, R. K., and R. Drazin (1989).“An Empirical Test of a Stage of


Growth Progression Model,” Manage-ment Science 35(12), 1489–1503.

——— (1990). “A Stage-ContingentModel of Design and Growth forTechnology Based New Ventures,”Journal of Business Venturing 5(3),137–150.

Klavans, R., M. Shanley, and W. M. Evan(1985). “The Management of InternalCorporate Ventures: Entrepreneurshipand Innovation,” Columbia Journal ofWorld Business 20(2), 21–27.

Ledwith, A. (2000). “Management of NewProduct Development in Small Elec-tronic Firms,” Journal of EuropeanIndustrial Training 24, Product Devel-opment and Innovation, 137–148.

Lee, C., K. Lee, and J. M. Pennings (2001).“Internal Capabilities, External Net-works, and Performance: A Study ofTechnology-Based Ventures,” Strate-gic Management Journal 22, 615–640.

Long, J. S. (1997). Regression Models forCategorical and Limited DependentVariables (Vol. 7). Thousand Oaks,CA: Sage Publications, Inc.

Lumpkin, G. T., and G. G. Dess (1996).“Clarifying the Entrepreneurial Orien-tation Construct and Linking It to Per-formance,” Academy of ManagementReview 21(1), 135–172.

Lumpkin, G. T., C. C. Cogliser, and D. R.Schneider (2009). “Understanding andMeasuring Autonomy: An Entrepre-neurial Orientation Perspective,”Entrepreneurship Theory and Practice33(1), 47–69.

Lussier, R. N. (1995). “A NonfinancialBusiness Success versus Failure Pre-diction Model for Young Firms,”Journal of Small Business Manage-ment 33(1), 8–20.

Luther, J. W. (1984). “How CEOs Can HelpImprove the Odds for Market Success,”Management Review 73(7), 18–24.

Lyon, D. W., G. T. Lumpkin, and G. G.Dess (2000). “Enhancing Entrepre-neurial Orientation Research: Opera-tionalizing and Measuring a KeyStrategic Decision Making Process,”

Journal of Management 26(5), 1055–1085.

MacMillan, I. C., and D. L. Day (1987).“Corporate Ventures into IndustrialMarkets: Dynamics of AggressiveEntry,” Journal of Business Venturing,Winter.

MacMillan, I. C., Z. Block, and P. N.Subbanarasimha (1986). “CorporateVenturing: Alternatives, ObstaclesEncountered and Experience Effects,”Journal of Business Venturing 1(2),177–191.

Matthews, C. H., and S. E. Human (2004).“The Economic and CommunityContext for Entrepreneurship: Per-ceived Environmental Uncertainty,” inHandbook of Entrepreneurial Dynam-ics: The Process of Business Creation.Eds. W. G. Gartner, K. G. Shaver, N. M.Carter and P. D. Reynolds. ThousandOaks, CA: Sage, 94–103.

McCrea, E., and S. C. Betts (2008).“Failing to Learn from Failure: AnExploratory Study of Corporate Entre-preneurship Outcomes,” Academy ofStrategic Management Journal 7, 111–132.

McDougall, P. P. (1989). “InternationalVersus Domestic Entrepreneurship:New Venture Strategic Behavior andIndustry Structure,” Journal of Busi-ness Venturing 4, 387–399.

McDougall, P. P., and R. Robinson(1990). “New Venture Strategies: AnEmpirical Identification of Eight‘Archetypes’ of Competitive Strategiesfor Entry,” Strategic ManagementJournal 11, 447–467.

McGrath, R. G. (1995). “Advantage fromAdversity: Learning from Disappoint-ment in Internal Corporate Ventures,”Journal of Business Venturing 10(3),121–142.

——— (2001). “Exploratory Learning,Innovative Capacity, and ManagerialOversight,” Academy of ManagementJournal 44(1), 118–131.

McGrath, R. G., and I. C. MacMillan(2000). The Entrepreneurial Mindset.

JOHNSON 495

Boston, MA: Harvard Business SchoolPress.

McGrath, R. G., T. Keil, and T. Tukiainen(2006). “Extracting Value from Corpo-rate Venturing,” MIT Sloan Manage-ment Review 48(1), 50–56.

Miles, M. P., and J. G. Covin (2002).“Exploring the Practice of CorporateVenturing: Some Common Forms andTheir Organizational Implications,”Entrepreneurship Theory and Practice26(3), 21–40.

Miller, A., and B. Camp (1985). “Explor-ing Determinants of Success in Corpo-rate Ventures,” Journal of BusinessVenturing 1, 87–105.

Miller, A., W. B. Gartner, and R. Wilson(1988). “Entry Order and Its Relation-ship to Market Share and Competi-tive Advantage: A Study of NewCorporate Ventures,” Academy ofManagement Best Papers Proceedings64–68.

Miller, A., M. S. Spann, and L. Lerner(1991). “Competitive Advantages inNew Corporate Ventures: the Impactof Resource Sharing and ReportingLevel,” Journal of Business Venturing6(5), 335–350.

Miller, D. (1983). “The Correlates ofEntrepreneurship in Three Types ofFirms,” Management Science 29, 770–791.

Miller, D., and P. Friesen (1982). “Inno-vation in Conservative and Entrepre-neurial Firms: Two Models of StrategicMomentum,” Strategic ManagementJournal 3(1), 1–25.

Paik, Y., and D. Y. Choi (2007). “Control,Autonomy and Collaboration in theFast Food Industry,” InternationalSmall Business Journal 25(5), 539–562.

Palmer, A., and C. Cole (1995). ServicesMarketing: Principles and Practice.Englewood Cliffs, NJ: Prentice-Hall,Inc.

Rauch, A., J. Wiklund, G. T. Lumpkin,and M. Frese (2009). “EntrepreneurialOrientation and Business Perfor-

mance: An Assessment of pastResearch and Suggestions for theFuture,” Entrepreneurship: Theoryand Practice 33(3), 761–787.

Robins, J. A., S. Tallman, and K.Fladmoe-Lindquist (2002). “Autonomyand Dependence of InternationalCooperative Ventures: An Explorationof the Strategic Performance of U.S.Ventures in Mexico,” Strategic Man-agement Journal 23(10), 881–900.

Ruamsook, K., D. Russell, and E. Thom-chick (2007). “U.S. Sourcing fromLow-Cost Countries: A ComparativeAnalysis of Supplier Performance,”Journal of Supply Chain Management43(4), 16–30.

Sandberg, W. R. (1986). New VenturePerformance: The Role of Strategy andIndustry Structure. Lexington, MA:Lexington Books.

Sandberg, W. R., and C. W. Hofer (1987).“Improving New Venture Perfor-mance: the Role of Strategy, IndustryStructure, and the Entrepreneur,”Journal of Business Venturing 2, 5–28.

Sanders, P. (2008). “Disney Learns fromPixar: Storied Animation Studio’s New3-D Computer-Generated Film Show-cases Melding of Pixar Talent,” WallStreet Journal Marketplace, B1.

Sharma, P., and J. J. Chrisman (1999).“Toward a Reconciliation of the Defi-nitional Issues in the Field of Corpo-rate Entrepreneurship,” Entrepreneur-ship Theory and Practice 23(3), 11–27.

Shrader, R. C., B. M. Oviatt, and P. P.McDougall (2000). “How New Ven-tures Exploit Trade-Offs among Inter-national Risk Factors: Lessons for theAccelerated Internationalization of the21st Century,” Academy of Manage-ment Journal 43(6), 1227–1247.

Siegel, R., E. Siegel, and I. C. Macmillan(1993). “Characteristics DistinguishingHigh-Growth Ventures,” Journal ofBusiness Venturing 8, 169–180.

Simon, M., and S. M. Houghton (2003).“The Relationship between Overconfi-dence and the Introduction of Risky


Products,” The Academy of Manage-ment Journal 46(2), 139–149.

Simon, M., and R. C. Shrader (1997).“Corporate versus Independent NewVentures: Resource, Strategy, andPerformance Differences,” Journal ofBusiness Venturing 12(1), 47–66.

Sobkowiak, R. T. (2002). “CorporationStart-up Dynamics: How the ParentCorporation and the Start-up Have toChange,” Human Resource Planning25(3), 18–28.

Storey, C. D., and C. J. Easingwood(1996). “Determinants of New Pro-duct Performance: A Study in theFinancial Services Sector,” Interna-tional Journal of Industry Manage-ment 7(1), 32–55.

Story, V., G. Smith, and J. Saker (2001).“Developing Appropriate Measures ofNew Product Development: AContingency-Based Approach,” Inter-national Journal of Innovation Man-agement 5(1), 21–47.

Stuart, R., and P. A. Abetti (1987).“Start-up Ventures: Toward the Pre-diction of Initial Success,” Journal ofBusiness Venturing, Summer.

Sykes, H. B., and Z. Block (1989). “Cor-porate Venturing Obstacles: Sourcesand Solutions,” Journal of BusinessVenturing 4(3), 159–167.

Thornberry, N. E. (2003). “CorporateEntrepreneurship: Teaching Managersto Be Entrepreneurs,” The Journal ofManagement Development 22(4), 329–344.

Thornhill, S., and R. Amit (2001). “ADynamic Perspective of Internal Fit inCorporate Venturing,” Journal ofBusiness Venturing 16(1), 25–50.

Tidd, J., and S. Taurins (1999). “Learn orLeverage? Strategic Diversification andOrganizational Learning Through Cor-porate Ventures,” Creativity and Inno-vation Management 8(2), 122–129.

Tushman, M., and D. Nadler (1986).“Organizing for Innovation,” Califor-nia Management Review 28(3), 74–92.

Vogt, W. P. (1999). Dictionary of Statis-tics & Methodology: A NontechnicalGuide for the Social Sciences, 2nd ed.Thousand Oaks, CA: Sage Publica-tions, Inc.

von Hippel, E. A. (1977). “Successful andFailing Internal Corporate Ventures:An Empirical Analysis,” IndustrialMarketing Management 6, 163–174.

Waters, J. A. (2000). “Achieving Innova-tion or the Holy Grail? ManagingKnowledge or Managing Commit-ment?” International Journal of Tech-nology Management 20, 819–838.

Wiklund, J., and D. Shepherd (2005).“Entrepreneurial Orientation andSmall Business Performance: A Con-figurational Approach,” Journal ofBusiness Venturing 20(1), 71–91.

Zahra, S. A. (1996). “Technology Strategyand New Venture Performance: AStudy of Corporate Sponsored andIndependent Biotechnology Ven-tures,” Journal of Business Venturing11(4), 289–321.

Zahra, S. A., and G. George (1999).“Manufacturing Strategy and NewVenture Performance: A Comparisonof Independent and Corporate Ven-tures in the Biotechnology Industry,”The Journal of High Technology Man-agement Research 10(2), 313–345.

Zahra, S. A., R. D. Ireland, and M. A. Hitt(2000). “International Expansion byNew Venture Firms: InternationalDiversity, Mode of Market Entry,Technological Learning, and Perfor-mance,” Academy of ManagementJournal 43(5), 925–950.

Zahra, S. A., D. O. Neubaum, and G. M.El-Hagrassey (2002). “CompetitiveAnalysis and New Venture Perfor-mance: Understanding the Impact ofStrategic Uncertainty and VentureOrigin,” Entrepreneurship Theory andPractice 27(1), 1–28.

Zirger, B. J., and M. A. Maidique (1990).“A Model of New Product Develop-ment: An Empirical Test,” Manage-ment Science 36(7), 867–883.

JOHNSON 497

Documents

Role of Structural and Planning Autonomy in the Performance of Internal Corporate Ventures_2012