A summary of the paper -- Mowery, David C., Joanne E. Oxley, and Brian S. Silverman. "Strategic alliances and interfirm knowledge transfer." (1996).

1. Mowery, David C., Joanne E. Oxley, and Brian S. Silverman. "Strategic
alliances and interfirm knowledge transfer." (1996).
Strategic Alliances
and
Interfirm Knowledge Transfer
2015/10/14 Wed.
Megan Sun (孫敏倫)

2. INTRODUCTION (1/3)
-- MOTIVATION
Firm-specific knowledge in competitive strategy
Resource-based
• Stressing the capture of rents through the protection and deployment
of these resources. (Penrose, 1959; Barney, 1986; Wernerfelt, 1984)
Dynamic capabilities
• Focusing in particular on the development, more than the exploitation,
of firm-specific. (Teece, Pisano, and Shuen, 1997)
• Dynamic capabilities >> Organizational learning >> Strategic Alliance
Knowledge-based
• Emphasize the capacity of the firm to integrate tacit knowledge (Grant,
1995)

3. INTRODUCTION (2/3)
-- MOTIVATION
Alliances may serve other purposes, however…
Knowledge in firm strategy suggests additional motives and effects of alliance formation
Firms use interfirm collaboration to gain access to other firm’s capabilities,
supporting more focused, intensive exploitation of existing capabilities
within each firm (Grant and Baden-Fuller, 1995; Nakamura, 1996)
Empirical assessment hampered by assertion, it’s because of…
The difficulty of measuring the technological and other capabilities of firms.
Using the citation patterns of partner firms’ patent portfolios, we measure
changes in the extent to which their tech resources “overlap” with their partners’
tech portfolios as a result of participation in an alliance

4. TRENDS AND MOTIVES IN
STRATEGIC ALLIANCES (1/2)
Motives for the formation of these recent alliances includes:
• Spread the costs and risks of innovations (Mowery, 1988)
• Collaboration between users and suppliers of new products as a means of
coordinating and formulating tech stands to increase market power
(Grindley, 1995)
• Acquisition of new technical skills or technological capabilities from partner
firms (Mariti and Smiley, 1983; Hamel, 1989; Shan, 1990)
Alliances have become widespread in tech-intensive industries in which they
were of little or no importance prior to 1975. The activities involve higher levels of
knowledge exchange and tech transfer among participants.

5. TRENDS AND MOTIVES IN
STRATEGIC ALLIANCES (2/2)
By combining some of the incentive structures of markets with the monitoring
capabilities and administrative controls associated with hierarch, alliances can
provide a superior means to gain access to tech and other complex capabilities.

6. HYPOTHESES (1/4)
Opportunities for interfirm transfer of capabilities afforded by different alliance
structures influences the choice among them (Kogut, 1988)
Interfirm transfer of tech capabilities will be enhanced in equity joint ventures;
participants will display higher increases in tech overlap after formation of equity
joint ventures than after formation of contract-based alliances.
Hypothesis 1a
An extension of this logic suggests that interfirm knowledge transfers should be
more limited in unilateral contract-based alliances, such as licensing
agreements. Bilateral contractual arrangements includes tech sharing or joint
development agreements.
Participants in unilateral contract-based alliances will display a lower increase in
tech overlap after the formation of the alliance than participants in bilateral
contract-based alliances.
Hypothesis 1b

7. HYPOTHESES (2/4)
Possession of relevant tech skills facilitates inward tech transfer(Rosenberg,
1991; Agmon, 1991) and firms tend to establish alliances with firms which have
overlapping tech capabilities (Mowery, 1997)
The extent of a firm’s absorption of tech capabilities from its alliance partners will
be positively related to its pre-alliance level of tech overlap with partner firms.
Hypothesis 2a
R&D investment is a necessary condition for the creation of absorptive
capacity. Scholars use R&D intensity as a proxy for absorptive capacity.
The extent of a firm’s absorption of technological capabilities from its alliance
partners will be positively related to its R&D intensity.
Hypothesis 2b
“Absorptive capacity” is a necessary condition for a firm’s successful
exploitation of tech capabilities or knowledge outside its boundaries.
Larger firms have more resources and more diverse tech portfolios. Therefore,
they are more likely to possess tech that is relevant to the alliance in question.
The extent of a firm’s absorption of tech capabilities from its alliance partners will
be positively related to its size.
Hypothesis 2c

8. HYPOTHESES (3/4)
Interfirm learning is closely related to its “intent to learn” (Hamel, 1991)
• Japanese company emerged from an alliance stronger than its partner
• A survey of U.S. and Japanese firms, suggests that Japanese firms are more
effective in commercializing innovations based on external sources of tech
than are U.S. firms. (Mansfield, 1988)
• National Traits: British firms share the alleged arrogance and lack of receptivity
of U.S. firms and French firms’ ability to build competencies and learn from
alliances approaches that of the Japanese.
Japanese companies will absorb more of the tech capabilities of alliance
partners than will firms from other countries.
Hypothesis 3

9. HYPOTHESES (4/4)
Participation in alliances need not always increase tech overlap
• An alliance may enable one firm to gain access to key knowledge based
capabilities of another without internalizing or acquiring that capability. E.g., one
firm designs and the other manufactures an advanced semiconductor device
• An alliance enables its members to specialize in different but complementary areas
of tech. Partner firms may experience divergence in tech capabilities over the
course of the alliance (reflected in lower levels of tech overlap)
• Convergent development through interfirm knowledge transfer are likely to
produce inconclusive results. (Mowery et al, 1997)
• One test for alliance effects in the presence of both convergent and divergent
development examines the absolute value of changes in tech overlap.
The presence of divergence and convergence within the alliance population
will prevent the observation of a consistent post-alliance increase in the tech
overlap of alliance partner firms.
Hypothesis 4a
The absolute value of the pre- vs. post- collaboration changes in tech overlap
will be greter for alliance partners than for a similar sample of non-allying firm
pairs over the same time period.
Hypothesis 4b

10. MEASURE (1/2)
-- TECH CAPABILITIES & INTERFIRM KNOWLEDGE TRANSFER
Previous research on interfirm transfer of tech capabilities in strategic alliances
• Case studies or small-scale surveys (Lyles, 1988; Sobrero and Roberts, 1996)
• R&D intensity used as a proxy for tech resources(Montgomery and Hariharan,
1991; Nakamura et al., 1996)
• Patent data as firms’ tech resources (Silverman, 1996; Mowery et al. 1997)
Advantages of Patent Data
• Better measures of the output of R&D activities and capabilities development
• More disaggregated measure
Limitation of Patent Data
• Systematic differences among industries
• Codified knowledge that may not capture flows of the tacit knowledge. We
assume that the codified knowledge represented by patents and tacit
knowledge are complements, rather than substitutes. That codified
knowledge flows and the tacit knowledge flows of interest are closely linked.

11. MEASURE (2/2)
-- TECH CAPABILITIES & INTERFIRM KNOWLEDGE TRANSFER
• Citation of prior patents thus serve as an indicator of the tech lineage of
new patents
• As Firmi acquires tech knowledge from its partner in an alliance, Firmj. We
should see a higher rate of citation of Firmj’s patents in new patents applied
for by Firmi
• An increase in this measure is an indication of the degree to which Firmi is
acquiring technology-based capabilities from Firmj, i.e., of the extent of
interfirm knowledge transfer in the alliance and of the tech overlap
between the two companies.

12. EMPIRICAL ANALYSIS (1/3)
-- SAMPLE AND METHODS
• Cooperative Agreements and Technology Indicators(CATI) database
• This database includes 9000 alliances involving some 5000 firms
• Collect 792 alliances data
• Involves at least one U.S. partner
• Alliance formation between 1985 to 1986
• Control sample (for Hypo.4)
• 858 frim pairs of non-allied firm by randomly generating in the CATI sample
132
226
434
Equity joint
ventures (16%)
Unilateral
contract-based
alliances (29%)
Bilateral contract-
based alliances
(55%)
280
102
410
Both are U.S. firms
(35%)
A Japanese+ A
U.S. firm (13%)
Elsewhere + A U.S.
firm (52%)

13. EMPIRICAL ANALYSIS (2/3)
-- SAMPLE AND METHODS
Micropatent data base
• Patent granted in the United States since 1975
• Constructed patent portfolios for each firm >> Multi-national or multidivisional
firms poses a challenge
• Solution: Firm-level patent portfolio
• 1985 edition of Who Owns Whom (North American Edition)
• Match firms in the sample with their relevant parent company, subsidiaries and/or
“sister ” subsidiaries
• Collect totally 275,000 patents in 838 firms from 1975-1994
Compustat
• Other info. for U.S. firms, such as SIC code, based on sales and R&D spending
Compact Disclosure’s “Worldscope Global” data base
• Other info. for non- U.S. firms, such as SIC code, based on sales and R&D
spending

14. EMPIRICAL ANALYSIS (2/3)
-- SAMPLE AND METHODS
聯盟前雙向
的KT值
聯盟前後雙向KT
的改變量
聯盟前後單
向的KT值(吸
收能力)
用Sales替代
Size
聯盟前對該
聯盟技術的
應用程度(單
向)

15. RESULTS(1/6)
--
• n : Alliance and convergence develop
• We restrict our analysis to alliances in which such
transfers take place, i.e., those exhibiting tech
“convergence” (DPCTCRSS >0)
• H1a: The positive and significant coefficient for
EQUITY implies that equity-based joint ventures
support higher levels of transfer than contract-based
alliances.
• Extra: Convergent alliances involving firms in the
same product lines or markets experience lower
levels of interfirm knowledge transfers than those
spanning
Interfirm transfer of tech capabilities will be enhanced in equity joint ventures;
participants will display higher increases in tech overlap after formation of equity
joint ventures than after formation of contract-based alliances.
Hypothesis 1a (Correct)

16. RESULTS(2/6)
--
• n : non-equity alliance and convergence develop
• Using the sample of 147 non-equity alliances exhibiting
convergent development
• H1b : Unilateral support lower levels of interfirm
knowledge transfer
• 因為DPCTCRSS都是正的，且UNILAT係數為負，則當策略聯
盟為單向時（UNILAT=1），DPCTCRSS就會「正的少一
點」，即lower increase的概念。
Participants in unilateral contract-based alliances will display a lower increase in
tech overlap after the formation of the alliance than participants in bilateral
contract-based alliances.
Hypothesis 1b (Correct)

17. RESULTS(3/6)
--
• n : Alliance and both development (convergence
and divergence)
• FIRMLRN : absorb capabilities
• H2a : Use only PRECROSS as a measure of absorptive
capacity
The extent of a firm’s absorption of tech capabilities from its alliance partners will
be positively related to its pre-alliance level of tech overlap with partner firms.
Hypothesis 2a (Correct)

18. RESULTS(4/6)
--
• n : Subset of 838 alliance data with R&D and sales
data
• FIRMLRN : absorb capabilities
• H2b and H2c :
• Large firms with in this sample thus appear to absorb
fewer capabilities from their alliance partners.
• Relatively R&D intensive firms do not exhibit superior
capabilities absorption in alliances.
• Due in the reduction in sample size, the coefficients
for both EQUITY and US-nonUS are not significant.
The extent of a firm’s absorption of technological capabilities from its alliance
partners will be positively related to its R&D intensity.
Hypothesis 2b (No)
The extent of a firm’s absorption of tech capabilities from its alliance partners will
be positively related to its size.
Hypothesis 2c (No)

19. RESULTS(5/6)
• Foreign firms (FORGNCO) don’t
display significantly lower levels of
capabilities acquisition than U.S.
firms.
• H3: No support for the supposedly
superior learning abilities of
Japanese companies. If such a
difference exists at all for Japan, it’s
precisely the opposite of that
articulated in the “received wisdom”
Japanese companies will absorb more of the tech capabilities of alliance
partners than will firms from other countries.
Hypothesis 3 (No)

20. RESULTS(6/6)
• n : all the alliance, includes non-allies
• “Domestic” alliances, in which all
member firms share a common home
country, are likely to produce different
patterns of interfirm tech transfer and
learning than that found in international
alliances.
• H4a : Consistent with H4a, the coefficient
on ALLIES in Equation 1 is very small and
stastistically insignificant (有聯盟的人>>
ALLIES為1，才會影響DPCTCRSS，Eq 1因為
考慮了di / co，所以會互相抵銷；Eq 2是用
絕對值，可見聯盟前後的KT差其實非常明顯)
• Megan: 丟ALLIES是為了測量「聯盟」與否的
影響性。若沒有ALLIES變數，則無法知道是否
滿足4b。
The presence of divergence and convergence within the alliance population
will prevent the observation of a consistent post-alliance increase in the tech
overlap of alliance partner firms.
Hypothesis 4a (Correct)
The absolute value of the pre- vs. post- collaboration changes in tech overlap
will be greater for alliance partners than for a similar sample of non-allying firm
pairs over the same time period.
Hypothesis 4b (Correct)

21. CONCLUSION(1/2)
--
• Uses a novel technique for measuring change in firms’ tech capabilities that
allows us to track the effects of alliance activity on interfirm knowledge
transfers and the transfer of tech-based capabilities from on partner to
another.
• H1a >> Equity joint ventures appear to be more effective conduits for the
transfer of complex capabilities than are contract-based alliances such as
licensing agreements. (Kogut, 1988)
• H1b >> Lower levels of transfer occur in unilateral contracts than in bilateral
non-equity arrangements (Oxley, 1996)
• H2a >> Experience in related tech areas is an important determinant of
absorptive capacity (depends on the pre-alliance relationship between the
two firms’ patent portfolios) (Cohen and Levinthal, 1990)
• Further research on this issue requires better measures of the structure and
activities of individual alliances.

22. CONCLUSION(2/2)
--
• Extra >> U.S. firms’ alliances with non-U.S. firms seem to result in lower levels
of interfirm knowledge transfer than those involving only U.S. companies.
(Gulati, 1996)
• Convergent/divergent development: Our results suggest the need for a
richer conceptual framework in considering the effects of alliance activity
on firm-specific knowledge and capabilities.
• Measure of changing firm-specific tech capabilities have considerable
promise for broader application to the analysis of firm strategy and tech
innovation.

23. FUTURE WORK AFTER THIS THEORY
• Cited 3249 times.
Measure Knowledge Diffusion
• SINGH, JASJTT. "MULTINATIONAL FIRMS AND KNOWLEDGE DIFFUSION:
EVIDENCE USING PATENT CITATION DATA." Academy of Management
Proceedings. Vol. 2004. No. 1. Academy of Management, 2004.
• Appleyard, Melissa M., and Gretchen A. Kalsow. "Knowledge diffusion in the
semiconductor industry." Journal of Knowledge Management 3.4 (1999):
288-295.
Tech capability (absorb capability)
• Kim, Changsu, and Jaeyong Song. "Creating new technology through
alliances: An empirical investigation of joint patents." Technovation 27.8
(2007): 461-470.
引用其管理上的結論
• e.g. 「有股份的KT比較好」的結論

24. THANK YOU FOR LISTENING 