Application of Network Analysis Techniques for Japanese Corporate Transaction Network

1
Application of Network Analysis Techniques for
Japanese Corporate Transaction Network

• Osamu Honda, Kouhei Sugiyama
• Hiroyuki Ohsaki, Makoto Imase
• Graduate School of
• Information Science and
• Technology
• Osaka University, Japan

2
Contents

• Background
• Business transaction in Cybersociety
• Related Work
• Objectives
• Analysis corporate transaction in real society
• Corporate Transaction Network
• Network Analysis
• Clarify characteristics of corporate transaction
  network
• Conclusion and Future Works

3
Background

• Several social activities have been shifting into
  network
• Virtual society called Cybersociety
• Several commercial transactions and circulation
  activities will be performed in Cybersociety
• New services are strongly demanded in
  Cybersociety
• Can improve efficiency and productivity of
  activities

4
Business Transaction in Cybersociety

• The structure of business transactions has not
  been sufficiently investigated
• Among several types of entities
• Such as individuals and corporations
• Once it becomes clear
• It is expected that useful services can be
  realized
• Offering business transactions to talented
  individuals
• Predicting growth of corporations

5
Related Work

• Propose a simple network model for reproducing
  characteristics of social network
• A person with many existing human relations is
  likely to acquire a new human relation
• There is an upper-bound in the number of human
  relations that every person can have
• Strength of human relations decays with time

6
Objective

• Clarify the structure of corporate transaction
  network in real society
• Build a corporate transaction network
• Obtain information of transaction among real
  corporations
• Analyze a corporate transaction network
• Detail characteristics by applying
  network-analysis techniques

7
Corporate Transaction Network

• Represents existence of transactions among
  corporations
• Node corporation
• Link existence of transaction(i.e., dealings of
  products or services)

8
Presumption of corporate transaction network

• Utilize the corporate information database
• Published quarterly
• Includes data for approximately 3,700 Japanese
  corporations
• Contains a list of major customers
• Representation as an undirected graph
• Not to contain information on trading direction
• Each corporation is assigned as a node
• Based on the list of major customer, links are
  generated

9
Corporate Transaction Network(Graphical
Representation)
First quarter of 2004
Forms a large cluster
10
Characteristics of Corporate Transaction Networks
and Other Social Networks
the number average of nodes degree
11
Network Analysis (Average Distance)
20

• Average distance

2005 1st quarter
movie actors
math. co-authorship
15
random network
the number of node shortest path length from i
to j
l log(k)
10
Corporations are closely related
5
0
1
100
10000
1e006
Number of nodes N
12
Network Analysis(Clustering Coefficient)

• Clustering coefficient

1
0.1
0.01
C / k
0.001
Degree of node i total number of triangular
cycles
0.0001
2005 1st quarter
movie actors
1e-005
math. co-authorship
random network
1e-006
1
100
10000
1e006
Number of nodes N
13
Network Analysis(Degree Distribution)

• Scale-free network

degree distribution function degree power law
index(gt 0)
Power law index 2.06
Have a scale-free property
14
Network Analysis (Spectrum Density)

• Spectrum density

Eigenvalue of the incidence matrix Delta
function
Has a small peak at 1
15
Structural Superiority of Corporations

• Index showing how a node is advantageous from a
  network structural viewpoint
• Structural superiority index of a corporation i

The input degree of node j is large,the
structural superiority index is large
Approximate value of the structural superiority
index of a corporation j
16
Network Analysis (Relation between Structural
Superiority Index and Price Earnings Ratio or
Market Capitalization)
for 0
weak negative correlation
strong correlation
17
Conclusion

• Analyzed the structure of the corporate
  transaction network
• Average distance is approximately 3.4 5.4
• It has a scale-free property with a power-law
  index of 2.1
• Structural superiority index is greatly related
  to several stock price indices and profit
  financial indices

18
Future Works

• Constructing a network model that can reproduce
  the structure of the corporate transaction
  network
• Realization of a new network service in
  Cybersociety