Networks in Evaluating Public Investments in Science and Technology

1
Networks inEvaluating Public Investments in
Science and Technology

• Nicholas S. Vonortas
• Center for International Science and Technology
  Policy
• and Department of Economics
• The George Washington University
• EU-WREN Evaluation Workshop
• June 17-18, 2004

2
Science, Technology and Innovation
IndicatorsArguing for Process Indicators
3
STI Indicators

• STI indicators can be grouped into four sets
• Input indicators
• Output indicators
• Innovation indicators
• Process indicators

4
Process Indicators

• Economists would argue that all these indicators
  could fit well to the classic mold of a
  production function, Y f(X), where X is a set
  of ST input indicators and the Y stands for the
  ST output indicators and innovation indicators.
• We are only now starting to unravel the middle,
  the black box, the transformation of one into
  the other (function f).
• Some of the information collected through the
  innovation surveys is heading that way by trying
  to pull in qualitative information on agent
  behavior. Still, nobody can claim we are there
  yet.

5
Knowledge Production Function
6
Network Indicators

• This is exactly where innovation network
  indicators fit in.
• Innovation network indicators account for the
  complex formal and informal relationships among
  economic agents involved in innovation, including
  companies, universities, and government agencies.
• Currently, more than one international research
  projects were building very extensive
  longitudinal innovation network data on a
  subject-based approach that allows their
  complement with other publicly available
  time-series information on the performance of
  individual organizations.

7
NETWORK INDICATORS FOR INDUSTRY APPRAISAL
8
NCRA-RJV DB Overview

• Database of all RJVs registered with the US
  Department of Justice under
• the National Cooperative Research Act (NCRA,
  1984)
• the National Cooperative Research Act (NCRPA,
  1993)
• Based on the announcements in the Federal
  Register (FR).
• Covered period 1985 1999.

9
NCRA-RJV DB Overview

• It records
• Date of announcements on RJVs in the FR.
• Technical area and description of RJVs
• Membership (including the entry and exit dates)
  in RJVs
• Uniquely identified entity (firm, university,
  government agency) which is a member in one or
  more RJVs
• Business/industry characteristics of identified
  entities
• 796 RJVs, 6,517 entities are listed as of the end
  of 1999.

10
NCRA-RJV DB Annual Trends
11
Patent DB Overview

• It consists of
• Patent information.
• Includes 2,923,922 patents in the USPTOs
  database granted from 1963 to 1999.
• Lists patent number, grant year, application
  year, assignee, U.S. patent classification, and
  the technology categories classified by NBER.

12
Patent DB Overview

• Patents
• Includes 2,923,922 patents in the USPTOs
  database granted from 1963 to 1999.
• Lists patent number, grant year, application
  year, assignee, original classification
• Citations
• Includes all U.S patent citations for patents
  granted from 1975 to 1999
• Contains 16,522,438 observations about citing
  and cited patent numbers
• Assignees
• - Lists 175,115 company and assignee names and
  associated assignee number

13
Merged DB Overview

• N-Entities are entities identified in both the
  NCRA-RJV DB and the Patent DB.
• Merged database mirrors the characteristics of
  the two parental databases.

2,435 N-entities
Patent DB
NCRA-RJV DB
175,115 Assignees
6,517 Entities
14
Example Automotive Industry(SIC 371)

• N371-Entities are selected on the basis of the
  following criteria
• N-entities.
• Primary sector is SIC 371 (Motor vehicles and
  motor vehicle equipment).

N371-entities
Entities in Joint Venture DB
Assignees in Patent DB
N-entities
15
Example Automotive Industry(SIC 371)

• N371-RJVs are defined as those having at least
  one N371-entity member.
• 1,635 entities have memberships in N371-RJVs.

1,635 entities
N371-entities
Entities in Joint Venture DB
Assignees in Patent DB
N-entities
16
Patent Network
1
3
10
6
2
11
4
7
5
9
8
E.g. Entity 1s patent cite Entity 3s patent.
5-year span
17
1
3
10
6
2
11
4
Patent Network
7
5
3
9
8
6
10
Inter-organizational relation via patent citation.
1
2
4
7
8
5
11
9
18
RJV Network
R1
R2
R3
R4
N-RJVs
2
3
4
5
6
7
8
9
10
11
1
2
3
N-entities
E.g. Entity 1 has 1 membership in RJV R1.
Entity 3 has 2 memberships in RJV R1.
5 years span
19
R1
R2
R3
R4
N-RJVs
2
3
4
5
6
7
8
9
10
11
1
2
3
N-entities
8
1
RJV network
2
6
9
3
6
4
7
5
11
10
20
EXAMPLE 1 NETWORK CORE
21
Network Level AnalysisQuestion 1 Does Network
Core emerge?
22
Organization Level Analysis

• Question 2 Are organizations in network core
  more likely to be found there?
• Examined the stability of the organizations
  position statistically.
• Result The same group of organizations
  repeatedly appears in network core.

23
EXAMPLE 2ALLIANCE NETWORKSIN COMMUNICATIONS
SERVICES
24
Objectives

• Question 1 Explore the structure of alliance
  networks in the communications industry at the
  U.S. and international level
• Approach From a broader look network level
  to a narrower look - the central players of
  alliance networks in communications industries

25
Data Overview

• Data are drawn from the INNET database, focusing
  on alliances among communications firms
  (participant SIC 48 7375)
• Number of alliances
• 1,965 (International), 631 (U.S.)
• Number of entities
• 2,039 (International), 690 (U.S.)
  
• Covered period 1985-2002

26
Network Indicators

• Visualization of 86-90 (International) Inter-firm
  Network
• (Different color represents different component
  size of circle represents what k-core it belongs)
• (K-core a subgraph in which each node is
  adjacent to at least a minimum number, k,
• of the other nodes in the subgraph. There are 4
  levels of k-cores in this network)

27
Network Indicators

• Visualization of 91-95 (International) Inter-firm
  Network
• (There are 5 levels of k-cores in this network)

28
Network Indicators

• Visualization of 96-00 (International) Inter-firm
  Network
• (There are 6 levels of k-cores in this network)

29
Network Indicators Visualization of 86-90 (U.S.)
Inter-firm Network
30
Network Indicators

• Visualization of 91-95 (U.S.) Inter-firm Network
• (There are 4 levels of k-cores in this network)

31
Network Indicators

• Visualization of 96-00 (U.S.) Inter-firm Network
• (There are 4 levels of k-cores in this network)

32
Central Players

• Central Players the most actively participating
  firms in the communications alliance networks
• Measure Degree Centrality
• Degree of a Node The number of nodes directly
  connected to it
• Degree Centrality the centrality measure for an
  individual node that is based on the degree of
  the node

33
Central PlayersDo central players remain
central in communications networks?

34
EXAMPLE 3CASES OF INDIVIDUALCOMPANY ALLIANCE
NETWORKS
35
PLDT Alliance Network
By reducing PLDT to only a single entity and
reorganizing the network we get
36
PLDT Alliance Network (PLDT at center)
37
PLDT Alliance Network
38
PLDT Alliance Network
39
Pfizers Network(Pfizer in yellow)
By reducing Pfizer to only a single entity and
reorganizing the network we get
40
Pfizers Network(Pfizer at center)
41
Pfizer Alliance Network
42
EXAMPLE 4COLLABORATE TO COLLUDE?
43
Hypothesis
Multi-Market Contact Arguments
44
Consideration
Is Multi-Project Contact through RJVs a Cause of
Concern?
YES
NO
Safeguards against anti-competitive behavior
MPC MMC

• Multi-Project Contact
• Multi-Market Contact

• Foreign Participation
• Technological Market
• Uncertainty
• Porous Constellation

45
EXAMPLE 5SUPPLY AND USE OF SCIENTIFIC
KNOWLEDGE IN THE TRIAD
46
(No Transcript)