Does collaboration improve innovation outputs

1
Does collaboration improve innovation outputs?

• Anthony Arundel Catalina Bordoy
• MERIT, University of Maastricht
• Forthcoming in Caloghirou, Y., Constantelou, A.
  and Vonortas, N.
• Knowledge Flows in European Industry Mechanisms
  and Policy Implications, Routledge

2
1. Collaboration innovation theory

• Cooperation plays a central role in innovation
  (Georghiou, 1998)
• Technological collaboration as the dominant
  form of producing knowledge (Antonelli, 1999)
• Collaboration improves the transfer of tacit
  knowledge (Senker, 1995)

3
2. Empirical evidence

• Patent citations, data on research partnerships
  (CATI), and external RD funding show that
  collaboration is widespread
• Problems is collaboration growing, as predicted?
• Is it only prevalent in specific sectors and in
  RD intensive firms?

4
2.1 Empirical evidence

• 1997 CIS Only 28 of innovative manufacturing
  firms developed one or more product innovations
  between 1994-96 through collaboration.
• Public-private co-publications have been
  declining between 1995-2001

5
(No Transcript)
6
(No Transcript)
7
3. Collaboration outcomes

• If collaboration is beneficial, then
• Should lead to more innovations or more
  economically valuable innovations.
• Higher profits
• Higher share of sales from innovative products
• But, firms in sectors with high competitive
  push may collaborate simply to stay in business.

8
3.1 Empirical evidence for outcomes

• Quality of the firms innovations (proxied by the
  innovativeness of the firm)
• All research finds a positive correlation between
  collaboration and more innovative firms
• BUT these studies cant directly link
  collaboration to the quality of the firms
  innovations

9
3.1 Empirical evidence for outcomes

• 2. Collaboration the innovation sales share
• Ambiguous results both positive and negative
• Best evidence in favour
• Mohnen Thierren (2003)
• Use 1999 Canadian Innovation Survey, find a
  positive relationship between any collaboration
  and the innovation sales share (control for firm
  size, sector, RD status, government subsidies,
  use of public science).

10
Summary of evidence so far

• Collaboration (and other external knowledge
  sourcing) positively correlated with
  innovativeness of firms.
• Majority of firms give greater importance to
  their in-house innovative activities.
• Public-private sector collaboration may have
  leveled off.
• Survey research rarely permits a direct link
  between collaboration and the innovation sales
  share.

11
4. The advantage of the KNOW survey

• Includes the question
• What percentage of your firms new or improved
  products (processes) were introduced using any of
  the following methods?
• 1. Buying in ________
• 2. In-house development ________
• 3. Collaboration ________
• 100

12
4. Two research questions

• What factors influence the use of collaboration
  as an innovation method?
• Does collaboration influence the innovation sales
  share?
• In the last year, percentage of sales from
  products that were significantly improved or new
  to your firm in the last three years

13
5. KNOW survey methodology

• Spring 2000 in the UK, Denmark, the Netherlands,
  France, Germany, Italy Greece
• Four sectors relevant here Food beverages,
  Chemicals (excl. pharmaceuticals), Telecom
  equipment, Computer services.
• Random sample of firms with 10-249 and 250-999
  employees
• Crude response rate of 25.3, 507 useable
  responses over 70 adjusted in some countries

14
6. Characteristics of respondent firms
15
6.1 Distribution of innovation methods
16
6.2 Innovation sales share

• No significant difference by country range from
  32.2 in Netherlands to 40.4 in Denmark
• Significant differences by sector
• Food 21.9
• Chemicals 29.9
• Telecom equip 47.9
• Computer services 51.0

17
7. Econometrics

• What factors influence any use of collaboration?
• Logit model where dependent variable 1 when
  collaboration used, zero otherwise
• Does collaboration influence the innovation sales
  share?
• Non-linear logit model where dependent variable
  can range from zero to 1 (Sshaped)

18
7.1 Basic regression form

• yi exp (ßxi ei)
• 1 exp (ßxi ei)

19
7.2 Independent variables

• Sector dummies country dummies
• Firm size (log of employees)
• Innovative capability
• Performs RD on a continuous basis
• RD employment share of all employees
• Outward looking approach regularly reads
  scientific or business journals to find ideas for
  innovation.
• Patents or lead-time advantages as the firms most
  important appropriation strategy

20
7.2 Independent variables

• Potential collaboration
• Independent versus part of a group
• Any expenditure on external RD
• RD projects with public research organisations
• Received of a subsidy for innovation
• Model for innovation sales share includes the
  collaboration product share

21
7.3 Four versions of each model

• Full set of independent variables (358)
• Reduced set to maximize firms (428)
• Independent firms only (195)
• Group firms only (161)
• Only give significant results for 1 2.

22
(No Transcript)
23
(No Transcript)
24
8. Conclusions

• Most innovations developed in-house, but
  collaboration is more important for process than
  for product innovation.
• Buying-in almost as prevalent as collaboration,
  which shows that these two methods need to be
  carefully identified in surveys.

25
8.1 Conclusions

• Probability of collaborating
• Strong country effects
• Increases with external opportunities to
  collaborate

26
8.2 Conclusions

• Innovation sales share
• Collaboration has no effect, in contrast to
  expectations
• External knowledge sourcing is important, but via
  journals (open science?)
• Strong sector effects, weak country effects
• No effect from patents, but lead time advantages
  strongly positive

27
8.3 Study limitations

• Time constraint sales share based on innovations
  developed in previous 3 years
• Only surveyed small and mid-size firms
• Only surveyed four sectors
• No data on quality of the innovations
• No data on cost of in-house development versus
  collaboration
• No data on other outputs, such as profits