Published online 22 February 2008 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/smj.672 Received 21 September 2004; Final revision received 15 November 2007
RESEARCH NOTES AND COMMENTARIES
: THE EFFECT OF GOVERNANCE MODES AND
RELATEDNESS OF EXTERNAL BUSINESS
DEVELOPMENT ACTIVITIES ON INNOVATIVE
PERFORMANCE
THOMAS KEIL,1* MARKKU MAULA,1 HENRI SCHILDT,2 and SHAKER A. ZAHRA3**
1 Institute of Strategy and International Business, Helsinki University of Technology,
Finland
2 Tanaka Business School, Imperial College London, London, U.K.
3 Carlson School of Management, Gary Holmes Center for Entrepreneurial Studies, University of Minnesota, Minneapolis, Minnesota, U.S.A.
This study examines how different governance modes for external business development activities and venture relatedness affect a firm 's innovative performance. Building on research suggesting that interorganizational relationships enhance the innovative performance of firms, we propose that governance modes and venture relatedness interact in their effect on innovative performance. Analyzing a panel of the largest firms in four information and communication technology sectors, we find that degree of relatedness for corporate venture capital investments, alliances, joint
ventures, and acquisitions influences their impact on innovative performance. Copyright ? 2008 John Wiley & Sons, Ltd.
INTRODUCTION
Prior research has shown that external business development activities can enhance and comple ment firm-level innovative performance by tap
ping into external knowledge sources. Studies have shown that corporate venture capital (CVC) invest ments (Dushnitsky and Lenox, 2005b), alliances (e.g., Ahuja, 2000), joint ventures (e.g., Inkpen and Crossan, 1995), and acquisitions (e.g., Ahuja and
Katila, 2001) are positively related to the inno vative performance, especially patenting, of estab
lished corporations.
Though past studies have greatly enriched our understanding of the positive effects of external business development activities, they may have drawn a somewhat simplistic picture of these com plex relationships. Commonly, prior studies ana
lyze one or two forms of external relationships, separating these modes from their strategic con text. External relationships are frequently part of larger strategic programs that involve complex choices among different governance modes. The use of these governance modes might coevolve with strategic programs (Koza and Lewin, 1998). Consequently, traditional fixed effect regression models do not pick up the time-varying effect
of multiple simultaneous ventures a firm pursues.
Keywords: corporate venturing; business development; acquisition; alliance; corporate venture capital; innova tion, patents
* Correspondence to: Thomas Keil, Helsinki University of Tech nology, Institute of Strategy and International Business, P.O. Box 5500, FI-02015 TKK, Finland.
E-mail: thomas.keil@tkk.fi
**The authors are listed in alphabetical order, not order of contribution.
Recent research (Ahuja and Katila, 2001) also
suggests that other important contingency variables have been omitted. Learning theory, for exam ple, suggests that relatedness with partner company
(Ahuja and Katila, 2001; Sapienza., Parhankangas, and Autio, 2004) influences how external corpo rate ventures affect innovative performance. Thus,
prior results might have been confounded by the presence of other simultaneous relationships or contingencies, possibly yielding misleading results about the effect of external business development activities on firms' innovative performance.
In this research note, we reexamine prior find ings of a positive relationship between different forms of external relationships and innovative per formance (operationalized by patenting) by simul taneously incorporating various external business development activities within a single empirical model. In our analysis, we employ a longitudinal data set covering the CVC investments, alliances, joint ventures, and acquisitions of the largest cor porations operating in four information and com munications technology (ICT) industries between
1993 and 2000. In addition, we analyze the effects of relatedness as an additional contingency for the effects of different types of external ventures on
innovative performance.
THEORY AND HYPOTHESES
Technological complexity and rapid technological change increasingly compel firms to access knowl edge from external sources (Hill and Rothaer mel, 2003; Nicholls-Nixon and Woo, 2003). To
do so, firms engage in external business develop ment activities, often referred to as external cor
porate venturing (Schildt, Maula, and Keil, 2005), with companies within and outside their industry, enabling firms to acquire the knowledge needed to exploit opportunities, or leverage existing knowl edge and resources with the help of partners. Prior research has documented a link between
these external relationships and firm innovative ness (Ahuja, 2000; Ahuja and Katila, 2001; Dush nitsky and Lenox, 2005a; Inkpen and Crossan,
1995).
By forming a relationship with another firm, a company can achieve cost savings via risk shar ing, resource focusing, and accessing other firms' knowledge. Relationships act as conduits through
which knowledge can be accessed and even inter nalized, leading to development of new capabili
ties. Often, external relationships help companies combine new complementary knowledge with its preexisting internal knowledge, leading to novel
inventions. While the underlying process is one of knowledge access and organizational learning from a partner, the specific mechanisms differ depending on the governance mode chosen for the
relationship between two firms; therefore alterna tive governance modes might exhibit differences in their effectiveness and when they can be used (Nicholls-Nixon and Woo, 2003). We will examine four distinct forms of external relationships: CVC
investments, alliances, joint ventures, and acquisi tions.
External business development modes and innovative performance
Corporate venture capital
Many large corporations have initiated CVC pro grams that provide funding and related services
to entrepreneurial firms in return for an equity stake (Dushnitsky & Lenox, 2005a). A key objec tive of these investments is accessing knowledge and learning from the start-ups (Dushnitsky and Lenox, 2005a). Start-ups are frequently important
sources of innovative ideas and forerunners of new technologies and often commercialize new technologies ahead of industry incumbents (Hill and Rothaermel, 2003). Investments in innova tive start-ups may provide corporations with an early window on emerging technologies or busi ness models, providing new knowledge, and allow ing it to redirect its internal research and develop ment efforts. The due diligence process related to
investments also provides corporations with unique opportunities to learn about inventions by start-ups
(Dushnitsky and Lenox, 2005b). Firms might also gain new knowledge from the start-up by partici
pating on the venture's board, interacting with their
personnel, or undertaking joint projects.
Alliances
Various studies have found a link between strate gic alliances and innovative performance (e.g., Ahuja, 2000; Stuart, 2000). Alliances help com
panies increase their innovativeness by facilitating
Copyright ? 2008 John Wiley & Sons, Ltd. Strut. Mgmt. J., 29: 895-907 (2008)
access to partners' knowledge and complemen tary resources as well as joint knowledge cre ation (Grant and Baden-Fuller, 2004). In a learn ing alliance, the partners actively transfer knowl edge through technology sharing, personnel trans
fer, transfer of codified knowledge, or through joint
projects that provide for intense interactions and
thus facilitate the transfer of tacit knowledge (Lane and Lubatkin, 1998). Knowledge transfer can as well be an unexpected by-product of an alliance. In
the course of intensive cooperation, the firm might
learn about new technologies or new applications of its existing technologies.
Joint ventures
Though much of the alliance literature does not
differentiate between joint ventures and non-equity alliances, joint ventures entail the creation of a jointly owned separate entity that may help overcome problems of opportunism and facilitate knowledge transfer (Osborn and Baughn, 1990). While the same mechanisms exist in joint ventures
as in non-equity alliances, in the former the knowl edge access and learning is mediated through a mutually owned and staffed unit instead of occur
ring directly between the partners. Thus, the rela tionship between the joint venture and its parent is central to its learning effectiveness (Inkpen and Crossan, 1995).
Acquisitions
Finally, acquisitions provide a conduit for a cor poration to absorb the knowledge base of another firm (Ahuja & Katila, 2001) by completely inte grating the acquired firm into its operations. The expanded knowledge base may facilitate novel combination and integration of knowledge and cre ate economies of scale in R&D (Ahuja and Katila, 2001). Acquisitions are likely suited to the trans fer or combination of tacit knowledge (Ranft and Lord, 2002) because they allow for the intense interaction among partners and reduce transaction costs associated with the transfer of tacit knowl edge. In particular, acquisitions avoid conflicts of interest and moral hazards involved in other
learning relationships (Khanna, Gulati and Nohria, 1998). Recently, empirical research has shown a positive relationship between acquisitions and
innovative performance (Ahuja and Katila, 2001).
Based on the prior empirical findings and theo retical arguments linking external relationships to innovativeness, we hypothesize:
Hypothesis la: The number of CVC investments
is positively related to the firm's innovative per formance.
Hypothesis lb: The number of alliances is pos itively related to the firm's innovative perfor mance.
Hypothesis lc: The number of joint ventures is
positively related to the firm's innovative perfor mance.
Hypothesis Id: The number of acquisitions is positively related to the firm's innovative per formance.
Relatedness of ventures
Organizational learning theory suggests that the relationship between different external business development activities and innovative performance
is moderated by the relatedness between the focal company and its external partner. Cohen and Levinthal (1990) argue that knowledge absorp
tion is facilitated when the partners' knowledge
is related to the knowledge base of the focal firm.
Common skills, shared languages, and similar cog nitive structures enable partners to communicate and learn from each other (Lane and Lubatkin, 1998), enhancing learning and thereby increasing the innovative performance (Mowery, Oxley, and Silverman, 1996). Yet, when the knowledge of the
venture and the corporation are too closely related, the venture might possess little new knowledge for the corporation to absorb and therefore little learn
ing might occur (Sapienza et al, 2004) resulting in fewer innovations (Ahuja and Katila, 2001). Thus, to maximize innovation, firms might engage in ventures with partners, whose knowledge is mod
erately related to that of the firm (Ahuja and Katila,
2001; Shenkar and Li, 1999). Therefore:
Hypothesis 2: The relatedness of external corpo rate ventures (corporate venture capital invest ments, alliances, joint ventures, and acquisi
tions) should exhibit an inverted U-shaped mod eration effect on the relationship between the
number of ventures and the innovative perfor
mance of the firm.
SAMPLE AND METHODS
Data
To test the hypotheses, we collected data from the 110 largest companies (publicly traded in U.S. stock exchanges with revenues above 200 million U.S. dollars in 1989) in four ICT industries cov ering three-digit standard industrial classification (SIC) codes 357 (computer and office equipment), 366 (communications equipment), 367 (electronic components and accessories), and 737 (computer programming, data processing, and other computer
related services) for the period of 1993-2000. We focused on larger firms that are more likely to engage in external business development activi
ties and are more likely to report them publicly. To avoid survivor bias, we applied the selection
criteria at the beginning of the sample period and
followed the companies until the end of the sample
period or until they were acquired or ceased oper ations. This created an unbalanced panel dataset.
We used multiple sources in data collection. Data on external business development activities was gathered from Thomson Financial's SDC Plat
inum database (Thomson-SDC, 2007) which also provided the SIC codes for the companies. For CVC investments, the SIC code was derived using a conformance table of the Venture Economics
industry classification. When multiple subsequent business development activities with the same part ner were conducted within three years, we con
sidered only the first business development activ ity. Altogether, we identified over 4,000 external relationships. In addition, we extracted all suc cessfully granted patents of the focal firms from the Derwent patent database in November 2002 (Thomson-Derwent, 2002). To reflect the date of
innovation, patents were recorded according to the
date of application. Finally, we gathered financial data for the companies from Compustat (2007).
Dependent variable
We operationalized the dependent variable, the yearly innovative performance, as the number of successful patent applications filed during each year (e.g., Ahuja and Katila, 2001; Hall and Ziedo nis, 2001; Stuart, 2000). While the strength of
patent protection in the ICT industries has been relatively low (Cohen, Nelson, and Walsh, 2000), patents have become increasingly important in
these industries (Hall, 2005). Patent counts also correlate highly with alternative measures of inno vative performance (Hagedoorn and Cloodt, 2003). Independent variables
Our hypotheses highlighted the effects of three dimensions in the portfolio of external relation
ships on innovative performance: volume, gover nance modes, and relatedness. We thus created a distinct count variable for each type of exter nal business development activity, distinguished by the governance mode (CVC, alliances, joint ventures, and acquisitions) and the relatedness with the focal corporation. The volume of external rela tionships was measured as annually depreciated stocks of past announcements of new partnerships, investments, or acquisitions, reflecting the durable
yet depreciating effect of ventures (see. e.g., Ahuja and Katila, 2001; Dushnitsky and Lenox, 2005b). Following related prior research, we used 30 per
cent as the annual depreciation rate (e.g., Blundell, Griffith, and Van Reenen, 1995; Dushnitsky and Lenox, 2005a). Robustness analyses with alterna
tive depreciation rates, available upon request, did not change the results materially.
To measure venture relatedness, we adopted a
categorical measure based on SIC code match
between the venture and the focal firm (e.g., Haleblian and Finkelstein, 1999; Villalonga and McGahan, 2005). Despite their weaknesses (e.g., Markides and Williamson, 1996), the information
for the SIC based relatedness measures is the only information that is consistently available for all ventures in our sample. We classified ventures as Intra-industry when at least the first three digits of the SIC codes were identical, and as Related when at least the first digit matched between the
venture and the focal firm. All other ventures were
classified as Unrelated.
To test the hypothesized inverted U-shaped mod
eration effects of relatedness on the effects of four
different external venture governance modes on innovative performance, we included the cumu lative stock of each form of relationship in each relatedness category in our regression analyses. We had to rule out the conventional approach of using interaction terms (cumulative stock * relatedness)
Copyright ? 2008 John Wiley & Sons, Ltd. Strut. Mgmt. J., 29: 895-907 (2008)
for three reasons: 1) given the nonlinear relat edness moderation hypotheses (inverted U-shape), we would have needed two moderation terms (lin ear and squared) for each of the four external venture governance modes, increasing the com plexity of the model and creating difficult multi
collinearity problems; 2) given that there are years when the companies examined do not have cer tain types of external ventures, the annual mean relatedness would have been difficult to construct and interpret; and 3) our moderator?an SIC code based venture relatedness measure?is a categor
ical measure that lends itself less to analysis by inclusion of an interaction term (Aguinis, 2003). Our approach represents what we believe to be
the best compromise considering the availability
of data for our sample firms and the modeling pur
pose.
Control variables
We included several control variables to ensure valid results. R&D expenditure has been shown
to be positively related to patenting (Ahuja and Katila, 2001; Blundell et al, 1995; Hall and Ziedo
nis, 2001; Hausman, Hall, and Griliches, 1984). To separate R&D effects from size effects (Hall and Ziedonis, 2001), we used R&D intensity (the firm's R&D expenditure divided by its annual sales)
instead of using the R&D expenditure directly. To control for a small number of unreported R&D values in the Compustat database, we fol lowed prior research and imputed the unreported values, creating a dummy variable with a value
of 1 when the values were missing (Hall and
Ziedonis, 2001). We also included company size, measured as the logarithm of its annual sales,
as a control variable. Following earlier research (e.g., Blundell et al., 1995; Blundell, Griffith, and Windmeijer, 2002; Dushnitsky and Lenox, 2005b;
Stuart, 2000), we ran models including mea
sures of earlier patents (e.g., pre-sample mean patents and annually 30% depreciated stock of prior patents) to control for firm-specific fac
tors.
Analytical method
We modeled annual innovative performance in a panel model with fixed firm-effects to control con
stant firm-specific heterogeneity (a Hausman test
ruled out the use of a random effects specifica tion) and year dummies to control for macroeco nomic effects that are constant across companies. As is common in most studies explaining patent
counts, we found evidence of overdispersion in the dependent variable and thus used a negative binomial model instead of a more conventional Poisson model (e.g., Stuart, 2000). Finally, as an additional robustness test to control for serial cor relation and overdispersion, we used fixed effects Poisson with robust bootstrapped panel clustered
standard errors, obtaining largely similar results1.
RESULTS
Table 1 provides descriptive statistics and corre lations. The low correlation between independent and control variables (other than between the vari ables based on subsets of the relatedness cate gories) and acceptable variance inflation factor (VIF) statistics suggest that multicollinearity of variables is not a problem in our analyses.
Table 2 provides the results for all models using panel Poisson and panel negative binomial regres
sion models. Model 1 presents the base model with fixed effects Poisson specification. Model 2 uses
fixed effects negative binomial specification that is
robust to the overdispersion in the dependent vari able. As fixed effects in negative binomial speci
fication fail to capture all firm-specific effects, we
control them using a pre-sample mean of patent ing in Model 4, incorporating a dummy variable to indicate the lack of any pre-sample patents (Blun dell et al., 1995). Even though a Hausman test sug
gests that fixed effects specification is more appro
priate than the random effects, we provide random effects models since they incorporate firms with no patenting. Overall, the random effects model (Model 3) provides largely similar results. We use Model 4 to interpret our results.
1 Since past patenting tends to influence future patenting and might therefore introduce endogeneity biases, we also estimated
dynamic models (Blundell et ai, 1995; Blundell et al., 2002). However, given our complex data structure we could not achieve convergence when estimating our model with a dynamic linear feedback model that uses pre-sample information to account for firm heterogeneity as implemented in the ExpEnd program for GAUSS (Windmeijer, 2002). However, when testing various ad hoc methods for including past patenting in panel regressions suggested that while past patenting is an important determi nant of future patenting, the hypothesized effects of external business development activities remained largely unchanged and
significant.
t o o G cd C/2 O Oh ^ ? l/-} ?
Il s?.
.2 <u .2 ?3 C T3 I 5? 1) <U S 3 > S C C c g er g s A< o ^2 i>| ^ s ? u
!2 "p ^ s ? S ?2 C O 8Copyright ? 2008 John Wiley & Sons, Ltd. Strut. Mgmt. J., 29: 895-907 (2008)
S
g
o o o o ? ~ o o o o o o o o I I tj- m es ?< ?< ? ? ? ? o o o O ? ? o o ? III I m Tt m es o ? ? -h ? -^ ? o o o o d o ? ? o o lili no m Tt tj ? q ? oooooooo voovom'Tj-fnO^o do o o o ?odd oooooooooo 8 S oooooooooo oooooooooo oooooooooo m o ?' ? tn ? iti rt oooooooo oooooooooo I I I es ? ? q ? -^miocNin oooooooooo oooooooooo es es m r~ es r- no oooooooooo o o o o o o o o o ? ooor^ ? rtr~0\(Nr-o ?????? ? O? ? o o o o o o o* o o o oooooooooo t ? -^ or~<NTt-r~oo^o ? onO m ? ? ? ?o??? ? ? ?modd oooooooooo III II
n oo oo m ? ?- ? ? ?>/-) ? ? tj lOOO OOOOOOOO?< ??
odd ??????????
I I I m m CN ? ? ?oom ? r<">oo?>On lOO? oooooooooo
odd dddddddddd I lili -h ? (N (N^'tCAinM'-^n^ u-j ^h ? oooooooooo odd dddddddddd I III I II on ? ? es '?i o (S -< es ? oo >n t? ^?^ oooooo ? o ? ? odd ?????????? ooooooooo ? dddddddddd III I ooooooo ? ? es ?????????? P s q q q es q q oooooooooo lili I ? es m es O O O O O O ?????????? O t?- m m oooooooooo rt Ti- m ? o o o o ? Ti" 00 IT) o o o o o ? o dddddddddd inTj-Tt-o-ies ? ONTi- ? r mmmmmmesc^Ti-Ti ?????????? mTt-mmesoovmeses o o o o' o o* o o d o ?2 S S S /y; Tf vn no c~ oo on ON ON ON On On On On On On On On On >H >H ><><><>, >* y y Tt- *o no r- oo on es es es es es es Strut. Mgmt. J., 29: 895-907 (2008)
t
cd p 00 JU 73 Cl, cd cT-O oo 2 +L oo8^
8 P
SI
g-3
t? O GO <U Ph T3 t? u cd O CU fcj 73 ? a> 73 Cu u Cu ?? cd T3 OO 00 wi5 3
00? S
O 73E
i) S t? X)?) ? S ?
73 ^ cd R O T3 W> ? 2 >< t? X>E
t? X) O? 00 73 ^ O O 73 ?f ? ^2 v? ? -t?E
CU <? O oo 73 X ?D .oo 5 ?L ? o fe?fi
NOoooo^in(SOMHfHt^'Hfsi^fooo\'<t^xoN(N'H OO^hOOnOc^OOOOsOcDNOOOOOOOOOO H?HoddodHoddodHOHOH o o o ^minf00ooooa\ir)fnoo^0(smoofnin^o\00 aNO^ooxTj-^rHffi^oo^NDfnmoi'HMM^ooin o^^oonoonoooonoonoooooooc^oo ^'?^O????^?????^h'?^?^h'??? oooor^(NOooasr^^o^vo^^oo^tcnr^Ocncncn H(N^\O^OOOHHO\00\(SfnH^HiD(S\)-^ O0r-HOO\OOOOOONO<3NOOOC>OOOC3>0 HOHoddHOHdoddoHdodHodd xOvDr)(N(NhoohooomNO(Sinfnirnnh(SHom OO^OOnOOsOOO-? o o o O O ^ O ^ ? O O *-h' ? ? ? ?' O ? ? ?OOnOOOOnOOOONO M(N'Hina\OOOrHrHO\oa(Nfn^(^rtTt(STtfn OO^OOsOOOOOOnOOsOOOOnOOOOnO t^'?^h'???^?^h'?????^???^h'??? Moor)MONoooor^ino^NO'!tTtOMO'^t^a\cnNC)fn -HC^?nNOooooO'H'H(^oo\(Sr)'Ha\rH^MTtfn OO^hOOnOOOOOOsOcTnOOOOnOOOOnO HOHoddHOHdoddoHdodHodd 'vtooinMO^ooTH>n^oO'-iO'<t"nfn(r)ff)r-imONO (3NO'^'-H00OasOc0O00OyOOmO'-iOCNO00O OO^OOnOOnOOOOOOnOOOOOOOoOO r-H'?^'o?d??^? 'ddoHOHOHodd T3 t??
t? > T31
w 00 >.s
u > T3 > u t? T3 t? T3 O t? cd 13< s
jd T31
13 > t? 'o t? t? o T3 oCopyright ? 2008 John Wiley & Sons, Ltd. Strut. Mgmt. /., 29: 895-907 (2008)
S
<3
CDI
cd Oh cd O Cu fc T3 ? ?D 73 CU H^3
cd T3 T3 CD XE
c o? 5
cd p CD 73 ! +1? ooil?
^ CD 1 S3 to 1 t? a ' o o10 $ s ^
S * t? XiE
002 *s 'S '?
T3 cd a O 73 W) g NsH cd cd .t? --h CD . CD _ T3 ?? ?
g
Pi CD "t?ai
CD S t? X7^
O 73?3
W> ? <D 5 t? x> ^ -73 S o CD CD O oo 73 X ,?D .2 .P r7! <E ti. <D ^ ? m on ?n 0\0(Nh d ? -h' d vtOVOhM^O car^^rVOtNin^?ZVOl^ NOcnooNONO^^^iD00 ? O r-H <N O -h On m ON O d d \oinmo^>o O (N *-h d cN en d T-H Os O O O d d i-H* d t-h' d d d d d 00 On in On O d d On O d d ^o in ON O d d ON O ON O d dONO^foor-rN^-^-r-cx?^^NO^
ONO^oin^Or-HTtm^HO^^in00 ddvo (N Voodoo c? oo On O ^t in 1-1 On VO CN in ^h in <N r- t^ !> ?i o o o o o on r- On <N in <N S? ^cn-^ ^h' d d d d d ' oo O <N ^ S t?t? ?s
t? er M t? cd Oh 1-h O CN O O O P? "O CD t? O Cu t? t? 04 Q t??*?
.s ?
^73 CU <U 00 P v? cd cd cd CD 73etD
Oh O CU -O ?d C t? cd on CD *->gj
S2
13 -S CU .S 73?
O CU CD CD CD^?
i CD X ~ <t? ^O _, 73 13 o O ?? .0 g
?j ?^
8 al e
lois
^h ? o 73 ^ 5 fi 3? ?" G g .S 25 8 Sgt
CD w O M* &
O 73 ,CD CD fc 00 o3 W) CD 73 C CD O 73&"?
CD O on on -m CD CD -? er* CD ~^ cd rt? *\'S 8 i
> Oh 13 "cd 73 Oh o CD ? 73S ? 1 ^
? Sa ? -g ? u o u o X >^ CU _D ? Q cd 00 CD S ?t? o X! +* -rt CDS S fi S
>>? ?S ?
CD "^ cd ?d , Sx) ^ & ' O CD *t? H -D on 73 CD 73 CD g C -2 ?*.? ^_ ^ CD ^ *1> 3 O ?go ? t? .g1 s
> m?s s
g lO CD -g2? c^
? y o t? on g O CU.on mO|\t? S
^ ^ o u fi O (D CD 73 V Tg on? ^X> g
&* Sa
Sou. r-H O Xi ^3 S . "t? CD < ? CD *- V?1^
o eu C xi CD t a xiHypotheses la-Id predicted positive relation ships between the four types of business develop
ment activities and innovative performance. To test these hypotheses, we tested the significance of the combined effects of the three coefficients that mea
sured three relatedness categories using both Wald and likelihood ratio tests (Aguinis, 2003; Long and Freese, 2006) based on Model 4 in Table 2.2 Failing to support Hypotheses la and lb, the com bined effects of intra-industry, related and unre
lated partners were insignificant for CVC invest ments and alliances. However, for joint ventures,
the combined, effects were positive and significant.
For acquisitions, the combined effects were signifi
cant but negative, failing to support Hypothesis Id. Hypothesis 2 predicted an inverted U-shaped moderation effect of relatedness on the relation
ships between different types of external busi ness development activities and the innovative out put. We tested the differences between the coeffi
cients for Intra-industry, Related, Unrelated stocks
of external relationships using Wald tests3. Our hypotheses predicted a significantly more posi
tive effect on innovative performance for related
2 To test the combined effects in Hypotheses la-Id, we used Wald tests for linear hypotheses (implemented using-test comm-and in Stata 9.1) as well as likelihood ratio tests (imple mented using-lrtest-command in Stata 9.1) (c.f. Long and Freese, 2006; StataCorp, 2005). Similar results were obtained with both
tests.
3 To test differences between the coefficients in Hypothesis 2, we
used Wald tests for linear hypotheses about the parameters of the most recently fitted model (implemented using -test- command in Stata 9.1). The equivalent results can also be obtained using the
lincom- function in Stata (cf. Long and Freese, 2006; StataCorp,
2005).
Corporate venture capital investments Alliances
25 % 20 % 15% 10% ?2 5 % c o
S. 0%
-5% -10% -15% -20 % Joint venturesintraindustry related unrelated
25% 20% 15% 10% 5 % 0% -5% -10% -15% -20 % 25% 20% 15% 10% + 5% 0% -5 % + -10% -15%
intraindustry related unrelated
Acquisitions
intraindustry related unrelated
The marginal effects of different types of ventures are calculated based on Model 4 in Table 2 with other independent variables set to equal to the sample means.The dotted lines represent 95% confidence intervals assuming one-tailed statistical tests.
Figure 1. Estimated marginal effects of ventures on patenting rate
Copyright ? 2008 John Wiley & Sons, Ltd. Strut. Mgmt. J., 29: 895-907 (2008)
partners in comparison to intra-industry and unre lated partners. Figure 1 depicts the results by showing marginal effects (the percentage changes in the annual patenting rate) for an incremental
deal in the stocks of different types of ventures as
estimated in Model 4.4 Wald tests on the signifi cance of the differences in coefficients in Model 4 broadly supported the hypothesis in all governance modes except for acquisitions.
Our results support Hypothesis 2 regarding CVC investments. According to the Wald test, the coef
ficient for the middle relatedness category was sig nificantly greater than the other coefficients. Model
4 showed a 12 percent increase in the number of patents for a recent incremental CVC investment
in a related target. For alliances and joint ventures,
the related linkages were also positive and signifi cant. For each new alliance with a related partner,
the predicted increase in patents was 2 percent. For an incremental related JV, the figure was 4 percent.
In line with our predictions, intra-industry and unrelated alliances and joint ventures were neg ative and insignificant. The Wald tests of the differences between the coefficients were also at
least weakly significant, except between related and unrelated alliances (insignificant). Overall, for CVC investments, alliances, and joint ventures, the
results were largely as predicted in our hypotheses.
The results for acquisitions, however, were con
tradictory to those hypothesized: the intra-industry
category had significantly higher effects than the related category (opposite to the hypothesis, as
can be seen in Figure 1), and the difference
between the related and unrelated external ventures
was insignificant. Intra-industry acquisitions were positive and significant (+3% effect). However, the effects of related and unrelated acquisitions on innovative output were significantly negative
(?5% in both cases).
To establish robustness and allow comparison with earlier research (e.g., Dushnitsky and Lenox,
2005b), we tested additional models. Model 5 incorporated a lagged dependent variable, a depre ciated stock of patents (Blundell et al, 1995) (oth erwise the same as Model 2). Models 6 and 7 used 200 bootstrapped observations with fixed effects Poisson specification to control for potential serial
correlation and overdispersion. Model 6 was the same as Model 1 except for standard errors that were more robust. Model 7 added the lagged patent
stock to Model 6. Overall, results remained largely
similar in robustness tests.
DISCUSSION AND CONCLUSIONS
The results improve our understanding of the rela tionship between external business development activities and the firm's innovative performance. Prior empirical research has been limited in its
typical focus on exploring the effects of a sin gle governance mode. Our study joins a small number of studies (e.g., Rothaermel and Hess, 2007; Schildt, Maula, and Keil, 2005) that has begun to investigate how multiple simultaneous
relationships affect firm learning and innovation. For instance, Rothaermel and Hess (2007) show a
simultaneous effect of R&D investments, alliances and acquisitions in the biotech industry on inno vative output. Schildt et al. (2005) examine differ ences in the effect of CVC investments, alliances, joint ventures and acquisitions on explorative and
exploitative learning. Our study adds to these prior
studies by providing a more fine-grained exami nation of the relationship between different gov ernance modes (CVC investments, alliances, joint ventures and acquisitions) and innovative perfor mance.
Our results further extend past studies by exam ining how the relationship between various exter nal business development activities and innovative performance is contingent upon venture related ness. Specifically, we find that alliances, joint ven
tures, and CVC investments in related industries (middle category of relatedness) have a signifi cantly positive correlation with increases in inno vative performance. These results indicate that acquisitions provide greatest benefits for innova
tive performance when the acquirer and the target are in the same industry (Haleblian and Finkel
stein, 1999). Acquisitions in related or unrelated industries seem to contribute less to innovative performance despite some widely held beliefs to
the contrary about their learning benefits. We can only speculate that in acquisitions outside the firm's major industry, integration difficulties out weigh potential learning opportunities. Overall,
our results suggest that each governance mode can be used to stimulate a company's innovative
4 In Figure 1, the effects are marginal effects based on Model 4 in Table 1 with other independent variables set to equal to
the sample means. In this model, the calculated marginal effects are almost identical to the percentage changes implied by the
incidence-rate ratios in Table 1, Model 4.
performance, yet the benefits depend on related ness of the partners and targets and only joint ventures show an aggregate positive relationship with innovative performance
Our results are subject to some limitations. Our empirical test was confined to four ICT indus tries during the years 1993-2000 and therefore may not be generalizable to other contexts. Future
research could also build on our study by utiliz ing more fine-grained measures for relatedness, as well as other relationship characteristics. Future
research can also examine different learning out comes such as the extent of technological change. There may also be a trade-off between learning
and other financial benefits in the choice of gover nance mode that depends on the extent of venture
relatedness. Given the research suggesting nonlin ear effects of alliances on innovative performance
(e.g., Rothaermel, 2001), future studies could seek to create a research setting that facilitates com parison of the functional forms of the effects across governance modes. Clearly, prior research suggests diminishing marginal returns to various business development activities such as alliances.
Altogether, our study prompts future researchers to
investigate the total portfolio of external business development relationships rather than examining certain governance modes in isolation. The inter actions between external governance modes with strategy, resources, and competitive environment
open a fruitful avenue for future exploration.
ACKNOWLEDGEMENTS:
We would like to acknowledge the financial sup port from The Research Programme for Advanced Technology Policy (ProACT) of the Ministry of Trade and Industry of Finland and Tekes, the Finnish Funding Agency for Technology and Inno vation. We are indebted to Herman Aguinis, Cyril Bouquet, Yuval Deutsch, Eileen Fischer, Mike Ornstein, Willow Sheremata, Peter Tryfos, Wim Vanhaverbeke, Bel?n Villalonga, and Frank Wind meijer for advice and comments on earlier ver
sions of this paper. An early version of this paper was published in the Frontiers of Entrepreneur ship Research 2003, the best paper proceedings of the 2003 Babson Kauffman Entrepreneurship Research Conference, in which the paper was
awarded the Stevens Institute of Technology Wes ley J. Howe Award for Excellence in Research on the Topic of Corporate Entrepreneurship.
REFERENCES
Aguinis H. 2003. Regression Analysis for Categorical Moderators. Guilford Publications: New York. Ahuja G. 2000. Collaboration networks, structural holes,
and innovation: a longitudinal study. Administrative
Science Quarterly 45(3): 425-455.
Ahuja G, Katila R. 2001. Technological acquisitions and the innovation performance of acquiring firms: a longitudinal study. Strategic Management Journal 22(3): 197-220.
BlundellR, Griffith R, Van Reenen J. 1995. Dynamic
count data models of technological innovation. Economic Journal 105(429): 333-344.
Blundell R, Griffith R, Windmeijer F. 2002. Individual effects and dynamics in count data models. Journal of Econometrics 108(1): 113-131.
Cohen WM, Levinthal DA. 1990. Absorptive capacity:
a new perspective on learning and innovation.
Administrative Science Quarterly 35(1): 128-152. Cohen WM, Nelson RR, Walsh JP. 2000. Protecting their
intellectual assets: appropriability conditions and why U.S. manufacturing firms patent (or not). NBER
working paper no. 7552 National Bureau of Economic Research: Cambridge, MA.
Compustat. 2007. Standard & Poor's Compustat Database. McGraw Hill : New York.
Dushnitsky G, Lenox MJ. 2005a. When do firms undertake R&D by investing in new ventures?
Strategic Management Journal 26(10): 947-965. Dushnitsky G, Lenox MJ. 2005b. When do incumbents
learn from entrepreneurial ventures? Corporate
venture capital and investing firm innovation rates.
Research Policy 34(5): 615-639.
Grant RM, Baden-Fuller C. 2004. A knowledge access ing theory of strategic alliances. Journal of Manage
ment Studies 41(1): 61-84.
Hagedoorn J, Cloodt M. 2003. Measuring innovative performance: is there an advantage in using multiple
indicators? Research Policy 32(8): 1365-1379.
Haleblian J, Finkelstein S. 1999. The influence of
organizational acquisition experience on acquisition
performance: a behavioral learning perspective.
Administrative Science Quarterly 44(1): 29-56. Hall BH. 2005. Exploring the patent explosion. Journal
of Technology Transfer 30(1/2): 35-48.
Hall BH, Ziedonis RH. 2001. The patent paradox revisited: an empirical study of patenting in the U.S. semiconductor industry, 1979-1995. Rand Journal of
Economics 32(1): 101-128.
Hausman J, Hall BH, Griliches Z. 1984. Econometric models for count data with an application to the patents-R&D relationship. Econometrica 52(4): 909-938.
Hill CWL, Rothaermel FT. 2003. The performance of incumbent firms in the face of radical technological Copyright ? 2008 John Wiley & Sons, Ltd. Strut. Mgmt. J., 29: 895-907 (2008)
innovation. Academy Of Management Review 28(2): 257-274.
Inkpen AC, Crossan MM. 1995. Believing is seeing:
joint ventures and organization learning. Journal of
Management Studies 32(5): 595-618.
Khanna T, Gulati R, Nohria N. 1998. The dynamics
of learning alliances: competition, cooperation, and relative scope. Strategic Management Journal 19(3):
193-210.
Koza MP, Lewin AY. 1998. The co-evolution of strategic alliances. Organization Science 9(3): 255-264. Lane PJ, Lubatkin M. 1998. Relative absorptive capac
ity and interorganizational learning. Strategic Manage ment Journal 19(5): 461-477.
Long JS, Freese J. 2006. Regression Models for Categor ical Dependent Variables Using Stata, (2nd edn). Stata Press: College Station, Texas.
Markides CC, Williamson PJ. 1996. Corporate diver
sification and organizational structure: a resource based view. Academy of Management Journal 39(2):
340-367.
Mowery DC, Oxley JE, Silverman BS. 1996. Strategic alliances and interfirm knowledge transfer. Strategic Management Journal, Winter Special Issue 17:
77-91.
Nicholls-Nixon CL, Woo CY. 2003. Technology sourcing and output of established firms in a regime of encompassing technological change. Strategic Management Journal 24(7): 651-666.
Osborn RN, Baughn CC. 1990. Forms of interorganiza tional governance for multinational alliances. Academy of Management Journal 33(3): 503-519.
Ranft AL, Lord MD. 2002. Acquiring new technolo gies and capabilities: a grounded model of acqui
sition implementation. Organization Science 13(4): 420-441.
Rothaermel FT. 2001. Incumbent's advantage through exploiting complementary assets via interfirm cooper ation. Strategic Management Journal, June-July Spe cial Issue 22: 687-699.
Rothaermel FT, Hess AM. 2007. Building dynamic capabilities: innovation driven by individual-, firm-, and network-level effects. Organization Science 18(6): 898-921.
Sapienza HJ, Parhankangas A, Autio E. 2004. Knowl edge relatedness and post-spin-off growth. Journal of Business Venturing 19(6): 809-829.
Schildt H, Maula M, Keil T. 2005. Explorative and
exploitative learning from external corporate ven tures. Entrepreneurship Theory and Practice 29(4): 493-515.
Shenkar O, Li J. 1999. Knowledge search in international cooperative ventures. Organization Science 10(2):
134-143.
StataCorp. 2005. Stata Base Reference Manual: Release 9/. StataCorp LP: College Station, TX.
Stuart TE. 2000. Interorganizational alliances and the
performance of firms: a study of growth and
innovation rates in a high-technology industry. Strategic Management Journal 21(8): 791-811.
Thomson-Derwent. 2002. Thomson Derwent World
Patents Index. Thomson Financial: New York. Thomson-SDC. 2007. Thomson-Securities Data Corpo
ration SDC Platinum Database. Thomson Financial:
New York.
Villalonga B, McGahan AM. 2005. The choice among
acquisitions, alliances, and divestitures. Strategic Management Journal 26(13): 1183-1208.
Windmeijer F. 2002. ExpEnd, A Gauss Programme for Non-Linear GMM Estimation of Exponential Models with Endogenous Re gres sors for Cross Section and Panel Data. CEMMAP working paper no. CWP 14/02. Institute for Fiscal Studies, Centre
for Microdata Methods and Practice: London, UK. Available at SSRN: http://ssrn.com/abstract=340323
(accessed 12 Janury 2007).
