Content uploaded by Uwe Cantner
Author content
All content in this area was uploaded by Uwe Cantner on Nov 24, 2016
Content may be subject to copyright.
2
1995-1997
1999-2001
cooperation, scientist mobility
1. Knowledge exchange and innovation
2. Network structures and innovation
3. Network dynamics
3
1. Knowledge exchange and innovation
2. Network structures and innovation
3. Network dynamics
4
Knowledge exchange and innovation
•Innovation network: actors exchanging information and knowledge
in order to innovate collectively
•Innovation: (re)combination of different pieces of information and
knowledge
•Conditions:
–Incentives: Knowledge bases of the potentially cooperating actors required to
be different
–Mutual understanding: the cognitive or technological proximity between them
should neither be too large nor too narrow
–Controllability: relies on trust, itself being determined by the institutional and
the social proximity among the cooperating partners
5
Network formation
•Network formation
–mutual benefits for individual actors to cooperate in generating new ideas
–the resulting interaction structures meeting the requirements of mutual
understanding and reciprocal incentives
•Modestly specialized groups of actors (in regions: Cantner and Graf 2004)
–related but not too similar technological capabilities
•Knowledge bases of the cooperating actors show a certain degree
of overlap (Cantner and Meder 2007)
–reciprocal absorptive capacities: overlapping part
enhances understanding and eases the aspired
knowledge transfer between I and II
–reciprocal incentives: non-overlapping parts of
their knowledge bases are valuable for the partners
6
A B C D
C D E F
I
II
1. Knowledge exchange and innovation
2. Network structures and innovation
3. Network dynamics
7
Actors’ performance and relational embeddedness I
•Number of ties (Cantner, Conti and Meder 2010)
–inverted-u shaped relationship between the number of innovation
collaborations and the innovative performance of actors in the Jena innovation
system
•Centrality (Graf and Krüger 2009)
–strong positive influence of degree centrality on innovative performance
–betweenness centrality tends to have a negative effect on patenting,
–being positioned at the interface between the local network and external
actors does not seem to be unequivocally positive as the authors find a u-
shaped relationship between the intensity of being a gatekeeper and
subsequent patenting activity
–being a member of the main component is not significantly related to
innovative performance
8
Actors’ performance and relational embeddedness II
•Type of ties
–arm's-length and embedded ties: a mix as the optimal network structure,
because each type of relation serves different functions: “Embedded ties
enrich the network, while arm's-length ties prevent the complete insulation of
the network from market demands and new possibilities” (Uzzi 1997)
–weak versus strong ties: weak ties induce improved performance whereas
strong ties seem to have a negative effect (e.g. Rowley, Behrens and
Krackhardt 2000; Ruef 2002)
9
Actors’ performance and relational embeddedness III
•Type of partners as a determinant of the innovative capacity of
firms in a network of innovators:
–cooperating with other firms leads to a higher innovative capacity up to a
certain number of ties (Cantner, Conti and Meder 2010)
–cooperating with public research institutes show a significantly higher
innovative capacity (Cantner, Conti and Meder 2010)
–intermediaries (Cantner, Meder and Wolf 2011) induce cooperation success in
terms of innovations generated
10
11
1995-1997
1999-2001
cooperation, scientist mobility
1. Knowledge exchange and innovation
2. Network structures and innovation
3. Network dynamics
12
Innovation network internal dynamics
•With respect to the dynamic relationships between exchanging
knowledge, technological overlap and collaboration three different
internal dynamics can be distinguished:
1. By the very nature of frequently exchanging knowledge there is a tendency of
the technological proximity between two actors I and II to narrow down.
Consequently, the innovative potentials become exploited.
(Cowan, Jonard and Zimmermann 2006).
13
A B C D
C D E F
I
II
A B C D
A B C D
E F
E F
I
II
Cooperation
G
G
Innovation network internal dynamics
2. Another dynamics is related to the potential of any one partner I and/or II to
be able to create new knowledge by collaborating with other partners. New
knowledge leads to heterogeneity among actors and by this to cooperation
potentials. This may counteract dynamics 1 by creating potentials with respect
other partners. (Cantner and Graf 2010)
3. By continuously exchanging knowledge with the same partner a higher degree
of mutual trust is built up, alleviating and furthering collaboration in the
future (Cantner, Meder and Wolf 2011)
trade-off (between 1. and 3.): less creative potential vs. trust
•Homophily or proximity in terms of
–cognitive dimension
–social dimension (weak ties versus strong ties)
–geographical dimension (face-to-face versus distant)
may reduce the creative potential over time / by repeated cooperation
and affect the individuals’ and system’s performance negatively
14
15
32 persistent innovators
1995-1997
1999-2001
cooperation, scientist mobility
Realized exchanges
Innovation network internal dynamics: Connectedness and
Overlap
•Network core
–R1: exchanging knowhow increases the degree of overlap between two
partners (as in Mowery, Oxley and Silverman 1998).
–R2: Network position and innovation performance constitute a virtuous cycle,
implying that the rate of acquisition of information and knowledge from the
outside is closely linked to the internally generated expertise and vice versa.
16
Permanent actors (32): development of mean degree (within)
Overlap
Cooperation
Scientist mobility
1995–1997
Mean degree
S.D.
15.125
(14.914)
2.563
(5.346)
0.938
(1.900)
1999–2001
Mean degree
S.D.
19.563
(17.692)
3.938
(6.710)
2.500
(2.700)
Significance of difference between mean degrees
Wilcoxon rank sum test
W
130
49
20
p-value
0.011
0.1
0.000
Note: One-sided tests are performed with H0 as no difference between samples and H1 in the direction of the
observed differences.
Innovation network internal dynamics: Continuity II
•Trust, creative potential and
flexibility (I):
•How do network incumbents interact
and build linkages?
Can we identify a time persistent
pattern of linkages?
•Results
–R1: linkages do not seem to be
persistent but rather short term
–R2: linkages in 99-01 are best
explained by mobility of
researchers in the same period
–R3: technological overlap is a
necessary condition for building a
linkage
17
Model
M1
Method
network regression
dep. Variable
cooperation99-01
Pr(≥|t|)
Pr(≥b)
cooperation95-97
-0,082***
0,154
1,000
scientist mobility95-97
-0,136**
0,43
0,989
scientist mobility99-01
0,410***
0
0,004
tech. overlap95-97
0,075*
0,361
0,072
(tech. overlap95-97)2
0,038**
0,014
0,014
public linkages
0,277*
0,051
0,077
private linkages
-0,109
0,178
0,842
intercept
0,051
0,431
0,894
mult. R2 (adj.)
0,153
(0,141)
# of obs. (nodes)
496
(32)
significance-levels according to QAP: ***≤0.01, **≤0.05,
*≤0.1; significance is the minimum of Pr(>b) (which is
documented) and Pr(<b); # of permutations: 1000
Innovation network internal dynamics: Continuity I
•Trust, creative potential and flexibility (II):
–This raises the question whether the relevant level of trust is related to the
dyad relation only or to the system in general
•In the latter case, this would imply that a frequent switching between
collaboration partners is just taking advantage of specific creative
potentials embedded in a general sphere of broad trust
•In combination to that, in the Jena case, the connectedness to incumbent
collaboration partners increased over time, just sustaining the
interpretation of a system wide level of trust
18
Innovation network internal dynamics: Triadic closure
•Triadic closure
–For network dynamics this implies that two actors having one cooperation
partner in common are more likely to form a new tie than two actors who are
not connected indirectly
–Consequently, former indirect ties among actors are turned into direct ones
leading to closed triads in a network; clique development (Skvortez 1991)
–An explicit test of the mechanisms guiding network evolution is performed for
the inventor network in German biotechnology between 1970 and 1995
(ter Wal 2009)
19
Innovation network external dynamics: periphery
•Entering and Exiting:
–Network core and periphery
–Disconnecting from actors - connecting to entirely new partners
–low benefits/trust –higher benefits
20
Mean degree permanent actors w.r.t. entering vs. exiting actors
Overlap
Cooperation
Scientist mobility
1995–1997
Exit
Exit
Exit
Mean degree
S.D.
22.531
(22.361)
2.375
(5.375)
1.719
(2.976)
1999–2001
Entry
Entry
Entry
Mean degree
S.D.
37.250
(38.756)
7.438
(17.005)
3.125
(4.030)
Significance of difference between mean degrees
Wilcoxon rank sum test
W
101.5
34.5
66.5
p-value
0.002
0.136
0.008
Note: One-sided tests are performed with H0 as no difference between samples and H1 in the
direction of the observed differences.
Innovation network external dynamics: Openness of the system I
•Entering and Exiting:
–R1: actors entering the system tie themselves rather close to the core of the
network made up by actors with a relatively high number of ties and
collaboration partners (Cantner and Graf 2006)
–R2: actors exiting the system are connected mainly to actors in the periphery
of the network.
21
Entering (157) vs. exiting (107) actors: mean degree w.r.t. permanent actors
Overlap
Cooperation
Scientist mobility
1995–1997
Exit
Exit
Exit
Mean degree
S.D.
6.738
(4.187)
0.710
(1.873)
0.514
(0.883)
1999–2001
Entry
Entry
Entry
Mean degree
S.D.
7.592
(4.825)
1.516
(2.623)
0.637
(1.415)
Significance of difference between mean degrees
Mann–Whitney
W
7906
7111
8613
p-value
0.208
0.003
0.668
Note: One-sided tests are performed with H0 as no difference between samples and H1 in the
direction of the observed differences.
Innovation network internal dynamics: Geographical reach I
•Region internals versus
externals
•How do the Jena innovators draw on
Jena external cooperation partners?
•Results
–R1: concerning all actors
(persistent and temporary
innovators) we find the share of
external linkages to decrease
over time
–R2: concerning the persistent
actors we find a drastic decrease
in the share of external linkages
22
variable
ratio of external to internal linkages
all innovators
1995-1997
1999-2001
cooperation
1,65
1,59
private actors
scientist mobility
2,09
1,74
cooperation
1,86
1,52
public actors
scientist mobility
1,77
1,27
only persistent innovators
cooperation
0,50
0,13
private actors
scientist mobility
2,25
0,69
cooperation
1,25
0,08
public actors
scientist mobility
1,25
0,33
Innovation network internal dynamics: Geographical reach II
Jena
Halle
23
homeworks
•Yet to perform:
–Longer time spans, panels
–More cases
–In-depth data
•This all is currently worked on in a project related to the German
„Spitzencluster“-Initiative
24
Mastertitelformat bearbeiten
THANK YOU FOR YOUR ATTENTION !
25
Jena: Cooperation and exchange structures
26
Dependent
variable
co99-01
I
II
III
Pr(t)
Pr(
)
Pr(t)
Pr(
)
Pr(t)
Pr(
)
Intercept
0.010**
0.873
0.969
0.010**
0.868
0.977
0.051
0.431
0.894
co95-97
-0.103***
0.078
1.000
-0.099***
0.085
1.000
-0.082***
0.154
1.000
sm95-97
-0.131***
0.451
0.996
-0.076
0.657
0.802
-0.136**
0.430
0.989
sm99-01
0.404***
0.000
0.006
0.410***
0.000
0.004
tech95-97
0.291***
0.000
0.001
0.242***
0.000
0.002
0.075*
0.361
0.072
tech95-97-squared
0.038**
0.014
0.014
Public linkages
0.337**
0.019
0.039
0.250*
0.079
0.087
0.277*
0.051
0.077
Private linkages
-0.093
0.261
0.782
-0.100
0.220
0.788
-0.109
0.178
0.842
Mult. R2 (adj.)
0.111
(0.102)
0.142
(0.132)
0.153
(0.141)
Obs. (nodes)
496
(32)
496
(32)
496
(32)
Significance-levels according to QAP: ***0.01, **0.05, *0.1; Significance is the minimum of Pr(>
) (which is documented) and
Pr(<
); No. Of Permutations: 1000
R1: technological overlap is a necessary condition for cooperation
R2: cooperations do not seem to be persistent but rather short term
R3 cooperations are best explained by mobility of researchers in the same period
Innovation network external dynamics: Openness of the system II
•Entering and Exiting:
27
Entering (157) vs. exiting (107) actors: mean degree within group
Overlap
Cooperation
Scientist mobility
1995–1997
Exit
Exit
Exit
Mean degree
S.D.
8.792
(5.984)
3.084
(5.207)
0.561
(1.361)
1999–2001
Entry
Entry
Entry
Mean degree
S.D.
18.191
(16.196)
2.242
(4.424)
0.497
(0.965)
Significance of difference between mean degrees
Mann–Whitney
W
5854
9191.5
8166.5
p-value
0.000
0.066
0.695
Note: One-sided tests are performed with H0 as no difference between samples and H1 in the direction
of the observed differences.