The impact of internationalisation of research on firm market value Paola Criscuolo Tanaka Business School Imperial College London

1
The impact of internationalisation of research
on firm market valuePaola CriscuoloTanaka
Business SchoolImperial College London
2
Trends in RD internationalisation
Source National Science Foundation Science and
Engineering Indicators
3
Motivations for RD internationalisation process

• Home-base exploiting
• To support local production
• Home-base augmenting
• To tap into local pockets of excellence (Florida,
  1997, Kuemmerle 1999, Patel and Vega 1999)

4
New RD organizational structures
Each RD unit assumes a leading role in the
creation of technological assets in a particular
area (Chiesa,1996, Gassmann von Zedtwitz 1999
Criscuolo Narula, 2007)

• New technologies are mostly created at centre

5
Global RD networks Novartis
6
Why MNEs are adopting this complex structure?

• The assumption is that MNE can
• efficiently fish for knowledge in a global
  pool, harness that knowledge for innovation, and
  then harvest its value for its stakeholders Doz
  et al. (2001) From Global to Metanational

7
But firms are struggling...

• Doz et al. (2006) found that
• coordination across research sites is weak the
  number of cross-border projects is relatively
  low foreign subsidiaries are not tapping into a
  broad range of external partners
• As one of Siemens CEO acknowledged
• we still need much more co-operation with
  outstanding universities around the globe our
  research co-operation still has a distinctly
  German flavour
• Cross-borders knowledge transfer and integration
  are hindered by organizational and technological
  distance and MNEs suffer from organizational
  inertia (Criscuolo and Narula, 2007)

8
Are MNEs benefiting from global RD strategies?

• Impact on innovative performance (mixed
  evidence)
• Japanese Pharmaceutical MNEs with foreign RD
  produce more patents (Penner-Hahn Shaver,
  2005)
• Pharmaceutical MNEs with a greater number of RD
  locations produce less patents (Furman et al.
  2006)
• Dispersing RD has negative impact on patents
  quality, however when innovative teams build on
  knowledge from different locations come up with
  patents with higher value (Singh 2007)

9
Are MNEs benefiting from global RD strategies?

• What is the impact on financial performance of
  dispersing RD?

10
Internationalisation of production and firm
performance

• Internalization theory (Buckley Casson, 1976)
  FDI occurs when firms can maximize profits by
  creating internal markets for intangible assets
• Mixed empirical evidence
• But evidence that RD intensity has a moderating
  impact
• Mock and Young (1991) state the value of
  multinationality stems from the possession of
  intangible assets and the value of these
  intangible assets increases with the degree of
  multinationality
• Benefits of internationalization are linked to
  RD internationalization

11
Organizational learning perspective

• Focus on learning opportunities generated by a
  geographically dispersed network of operation
  (Hitt et al. 1997, Barkema Vermeulen 1998,
  Zahra et al. 2000)
• Ability to absorb external knowledge depends on
  existing RD capabilities (Cohen Levintahl,
  1990)
• Thus need to have a geographically dispersed
  network of RD facilities

12
Market value RD

• Technological activities ? intangible capital ?
  future profits ? reflected in market valuation
• Markets value RD stock, innovation output, but
  also innovation quality (Hall et al. 2005) and
  coherence of firm knowledge base (Nesta
  Saviotti, 2006)
• Investors appear to value not only quantity but
  also nature and properties of firm knowledge
  capital

13
Hypotheses

• H1 Geographical dispersion of research
  activities has a positive impact on firm market
  value
• H2 The higher the level of firms local
  embeddedness in host regions scientific community
  the higher their market value
• H3 The more that firms are able to integrate
  knowledge from foreign locations the higher their
  market value

14
Embeddedness

• Concept initially defined at the level of the
  individual
• Granovetter (1985) most individual behaviours
  are closely embedded in networks of interpersonal
  relationships. The strength of these trust-based
  relationships in a social network is defined as
  embeddedness
• Uzzi (1996) and Gulati (1999) applied the concept
  of embeddedness to firms
• Key elements of an embeddeded relationship
  (Gordon and McCann 2000)
• firms are more willing to engage in risky
  co-operative projects
• firms may act as a group in order to support and
  achieve a common and mutually beneficial objective

15
Local embeddedness

• Foreign subsidiaries are outsiders in the host
  region (Solvell Zander, 1995)
• Crucial to be embedded in the local network to
  access information knowledge, especially if
  tacit
• Local embeddedness has positive impact on foreign
  subsidiary performance and innovation (Andersson,
  Forsgren Holm, 2002)

16
Hypotheses

• H1 Geographical dispersion of research
  activities has a positive impact on firm market
  value
• H2 The higher the level of firms local
  embeddedness in host-region scientific
  communities the higher their market value
• H3 The more that firms are able to integrate
  knowledge from foreign locations the higher their
  market value

17
Knowledge integration

• Knowledge integration is crucial to create value
• the critical source of competitive advantage is
  knowledge integration rather than knowledge
  itself (Grant, 1996)
• Synergies between RD sites and international
  cross-fertilization of knowledge might lead to
  technological upgrading (Bartlett Ghoshal 1990,
  Zander 1999)

18
Hypotheses

• H1 Geographical dispersion of research
  activities has a positive impact on firm market
  value
• H2 The higher the level of firms local
  embeddedness in host regions scientific community
  the higher their market value
• H3 The more that firms are able to integrate
  knowledge from foreign locations the higher their
  market value

19
Data Publications

• Scientific publications by the 1998 Fortune 500
  pharmaceutical and chemical MNEs (29 firms)
  between 1990 - 2005
• To capture the geography of their research
  activities
• In science-based sectors firms publish selected
  results to increase their visibility and to be
  plugged into the scientific community (Cockburn
  Henderson, 1998, Tijssen 2004)

20
Data Co-authorship

• Co-authorship behaviour by foreign subsidiaries
  scientists
• To capture both local embeddedness and knowledge
  integration from multiple locations
• Co-authorship captures knowledge exchanges since
  it often entails face-to-face interactions,
  extensive discussions, joint problem solving,
  exchanges of information and tacit knowledge
• CHI indicator to assign each publication to a
  particular sub-discipline (e.g. Microbiology,
  Pharmacology)

21
Publication co-authorship trends
22
The model

• v log of Tobins Q proxied by ratio of market
  capitalisation divided by total assets
• rda (RD intensity) log RD expenditure over
  total assets
• inta (Intangible assets) log investments in other
  intangible assets over total assets
• X control variables firm size (log sales),
  financial leverage (log ratio total liabilities
  over total assets), publication intensity (log
  ratio publications over RD expenditure), time
  dummies

23
Geographical dispersion
Herfindahl-based index of geographic distribution
of firm i publications in countries n at time t.
If equals to 0 ? publications concentrated in one
country
24
Local embeddedness

• Weighted average of subsidiaries co-publications
  with local counterparts

Weights share of co-authorships in region n, in
discipline k and year t over total number of
co-authorships in that region
25
What do I mean by local co-authorship?

• A publication that includes two scientists from
  GSK subsidiary in Research Triangle Park (North
  Caroline), two scientists from the University of
  Duke (North Caroline), and one scientist from the
  University of Wisconsin, would generate six
  co-authorships but only four will be considered
  as local

26
Knowledge integration indicator
Novartis intra-firm co-authorship network in 2005
density of intra-firm co-authorship network
27
Econometric issues

• Problems of heteroskedasticity and
  autocorrelation
• Kmenta (1986) autoregressive heteroskedastic
  model
• But suitable only when N, the number of firms, is
  lower than T, the number of periods
• Fixed effect AR(1) and adjust for panel
  heteroskedasticity using the Whites correction

28
Results from fixed effect Kmenta (1986)
autoregressive heteroskedastic model
H1
H2
H3
29
Findings

• Investors seem to perceive the creation of a
  global network of research facilities as an
  opportunity to increase future profits
• Higher levels of local embeddedness are
  associated to higher market values
• Investors do not to value positively efforts to
  achieve synergies across geographical boundaries

30
Implications

• Global RD strategy
• Broaden the search outside the national
  boundaries
• Being open and connected to the local scientific
  community
• Do not desperately seek synergy (Goold
  Cambell 1998, Kretschmer Puranam, 2007)
• Support for the organizational learning
  perspective on internationalisation
• Extend our understanding of the relationship
  between multinationality and firm performance

31
Limitations

• Econometric issues
• RD intensity is predetermined rather than
  exogenous
• persistence in market value shocks requires
  lagged values of Tobins Q and generalised method
  of moments (GMM) estimator
• Local embeddedness is a characteristic of the
  subsidiary not of the MNE
• Local embeddedness captured with a very crude
  indicator

32
Future research

• Increase the sample of firms
• Identify collaborative partners and develop a
  better measure of embeddedness
• Explore impact of local vs non-local scientific
  collaborations by foreign subsidiaries