【2009,science】Economic Networks_ The New Challenges

Economic Networks: The New Challenges

Author(s): Frank Schweitzer, Giorgio Fagiolo, Didier Sornette, Fernando Vega-Redondo,

Alessandro Vespignani and Douglas R. White

Source: New Series, Vol. 325, No. 5939 (Jul. 24, 2009), pp. 422-425

Science,

Published by: American Association for the Advancement of Science

Stable URL: /stable/20536695

Accesd: 02-12-2015 14:14 UTC

Your u of the JSTOR archive indicates your acceptance of the Terms & Conditions of U, available at /page/

info/about/policies/

JSTOR is a not-for-profit rvice that helps scholars, rearchers, and students discover, u, and build upon a wide range of content

in a trusted digital archive. We u information technology and tools to increa productivity and facilitate new forms of scholarship.

For more information about JSTOR, plea contact support@.

American Association for the Advancement of Science

is collaborating with JSTOR to digitize, prerve and extend access to

Science.

This content downloaded from 113.240.234.241 on Wed, 02 Dec 2015 14:14:54 UTC

Pushing

Networks

to

the Limit

remedy government

opened fishery

into than

more

initial this across resource Wade,

failure,

the are

the but to

divided the coastal

50

ctors,

re

area

t of

SES

variables

needed enable test

theoretical models costs

benefits between

individuals and

scholars build

systems

to

and

and

and

quotas, Social

and

required governments,

obrvers onboard all catches to

to C.

record

transferable

assigned

that all have

ships

neutral

(32).

of

heterogeneous

communities,

lead to

improved policies.

19. R.

Economie Conditions

Village Republics:

for

Collective

Action in

South India

(ICS,

San

Francisco, CA,

1994).

20.

].-M. Baland, J.-P. Platteau,

Halting Degradation of

Natural Resources

(Oxford

Univ.

Press,

New

York,

2000).

21. M.

]. Wilson,

Acheson, ].

A. R. S.

Steneck,

in

Linking

and

Ecological Systems,

F.

Berkes,

Folke,

Eds.

(Cambridge

Univ.

Press,

Cambridge,

1998),

pp.

390-413.

22. P. N.

Wilson,

G. D.

Thompson,

Econ. Dev. Cult.

Change

41,

299

(1993).

23.

E.

Mwangi,

Socioeeonomie

Change (2008).

and Land monitoring

U in

New The

York, 3. National Rearch

2007).

MacA/lillan,

Africa (Palgrave

24. E.

Schlager,

W.

Blomquist,

S. Y. Land Econ.

Tang, of effectiveness

70,

294

(1994).

25. A. in

Agrawal, People

and Forests:

Communities, Science

Institutions,

and C. C. M. A.

Governance, 6. E. R.

Gibson,

E.

Ostrom,

Eds.

(MIT

AAcKean,

Press,

Cambridge,

MA, Common-Pool Resources Ann

2000),

pp. scale

57-86.

26. P. B.

Trawick, 7. F.

Hum. Ecol. 1

29,

(2001).

27. E.

Ostrom,

Understanding

Institutional

Diversity

(Princeton

Univ.

Press, Princeton,

N], and

2005).

28.

].

A.

Brander,

M. S.

Taylor,

Am. Econ. Rev.

88, example,

119

(1998).

29. X.

Basurto, ]. (2005).

Soc. Nat. Resour.

18, had been

643

30. H.

Nagendra,

Proc. Nati Acad.

Sei. U.S.A.

104,15218

(2007).

31. E.

Ostrom,

H.

Nagendra,

Proc. Nati. Acad. Sei. U.S.A.

103,

19224

(2006).

32. C. W.

Clark, 11. R.

The

Worldwide Crisis in Fisheries: Economic

Models and Human Behavior

(Cambridge 2001).

Univ. further

Press,

Cambridge,

2006).

33. G. C.

Gibson, ].

T.

Williams, Ostrom, L.

E.

World Dev.

33, 13.

273

(2005).

34. E.

Coleman, Steed, 14. W. A.

B.

Ecol. Econ. findings

68,

2106

(2009).

35. in of

Supported

part

by grants (2007).

NSF coastal

BCS-0624178 and

BCS-0601320. I thank T. K. X.

Ahn, Axtell, Basurto,

R.

E.

B. C.

Fischer, 13286

C.

A.

Gonzalez,

]. Broderick,

Coleman,

Eavey,

E.

Jameson,

B. de D.

Leon,

Porter,

M.

Schlueter,

D.

Sprinz,

and Walker for comments and

].

suggestions.

10.1126/72133

the

long-term sustainability

of

Furthermore,

on

at a focal

SES level

depends

rules

devid

as

well enforcement their not

and

being by

overruled

larger policies. Council,

government

rules has been shown

long-term

in recent

studies of forests in to

depend

other's

harvesting practices Larger (Univ.

governance

countries

multiple

on

urs'

one

an

to monitor

willingness

(75,31,33,34).

either

facilitate

systems may

or

destroy

governance

systems

at Ecological

a

focal SES level.

The colonial in

powers

Africa, Asia,

and Latin

America,

resource

for not

did

recognize

local

institutions that

developed

over York,

centuries

and

impod

their which

own

rules,

frequently

led to

overu

if not destruction

(3, 7,23).

Efforts

are

currently

under

way

to

revi and

develop

the

SES framework

prented

here the 104,

with

goal

of

establishing comparable

databas enhance the

to

about

gathering

of rearch

the sustain

process

affecting

zones,

and water

forests,

pastures, 104,

ability

across Proc. Nati Acad. Sei. U.S.A.

dis Chhatre,

systems

around the world.

Rearch

ciplines questions

rapidly

enhance the

and will

and to

increa the

thus

cumulate

more

knowledge

needed

sustainability complex

of

SESs.

and P.

qualitative

data Science 1907

about the

core

Quantitative

References and Notes

1. E.

Pennisi,

Science

302,

1646

(2003).

2. R. B.

Norgaard,

Conrv. Bio!.

22,

862

The Drama

of

the

Commons

Academies

Press,

(National

Washington,

DC, 2002).

4. L.

Pritchett, Woolcock, 32,

M.

World Dev.

191

(2004).

5. F. Berkes et

ai, 311,

1557

(2006).

Ostrom,

Gardner, ]. Walker, Rules, Games,

and

of

Michigan

Press,

Arbor, Ml,

1994).

Berkes,

C

Folke, Eds.,

Linking

Social and

Systems (Cambridge

Univ.

Press,

Cambridge,

1998).

8. M. A.

Jansn,

Complexity Ecosystem

Management

(Edward

Elgar,

Cheltenham, UK,

2002).

9. a

].

Norberg,

G.

Cumming,

Eds.,

Complexity Theory for

Sustainable Future

(Columbia

Univ.

Press,

New

2008).

10. S. A.

Levin,

Ecology

73,

1943

(1992).

Axelrod, Cohen,

M. D.

Harnessing Complexity

(Free

Press,

New

York,

12. E.

Ostrom,

Proc. Nati. Acad. Sri. U.S.A.

15181

(2007).

]. Wilson,

Yan,

C Proc. Nati. Acad. Sei. U.S.A.

Wilson,

104,

15212

(2007).

Brock,

5. R.

Carpenter,

Proc. Nati Acad. Sei. U.S.A.

15206

15. A.

A.

Agrawal,

105,

(2008).

16. R.

Meinzen-Dick,

Proc. Nati. Acad. Sri. U.S.A.

104,

15200

(2007).

17. G.

Hardin, 162,

Science 1243

(1968).

18. T. E.

Dietz, Stern, 302,

Ostrom,

(2003).

PERSPECTIVE

mutual

Economic Networks:

The

New

Challenges

Sornette,1'3

Fernando

Vega-Redondo,4,5

crisis critical need for and

illustrates

of economic networks.

a

new

fundamental

Economic

of the

understanding

are structure

on and

built

systems dynamics

increasingly

credit and

investment trans-national

networks,

trade

relations,

or

R&D

interactions,

be

it

trade,

ownership,

or

credit-debt

Different

alliances,

relationships.

under the

agents

may

have

different behaviors

interactions

strategic

interactions

can

be

reprented

(/).

The

evolving

that bound in

are

space

and

by

network

dynamics

same

conditions and

have

Frank

Schweitzer,1*

Didier

Giorgio

Fagiolo,2

Alessandro

R.

White8

Douglas

Vespignani,6,7

The current

economic

time and

can

change

with the environment

and

are

formed

coevolve the

with

agents (2).

Networks

or

devolve of the addition

on

the basis

or

deletion

of either

agents

or

the

links between

them.

The socioeconomic

has

empha

perspective

sized

understanding

how the

strategic

behavior

of the

interacting

reciprocally

architectures. One

star-spoke

agents

is influenced

shapes?relatively

common a predict

network,

by?and

simple

network

is that of

example

centralized

or

established in economic

paradigms

theory.

This will facilitate

the

design policies

of conflicts

that reduce

between individual and

global efficiency,

as as

well

reduce the risk interests

of failure

global

by making

economic

networks robust.

more

interdependences,

implemented

through

that

supply

chains that an

have difficult to

proven need, therefore,

and We

control.

approach

stress can

the

systemic

complexity a

of economic

networks

and that to revi and

be ud

extend

like

very

f

I

ihe

I network.

A

economy, sys

tern,

reflects

large

number not

as Institute, Geneva,

any

other be traced University

complex

a

dynamic

interaction of of the

of different

agents,

a

a

few

key

players.

The

resulting systemic

on

obrvable the Institute,

havior,

aggregate

shows

just

be two one

properties

underlying

interaction

Rearch economic

networks has

examining

been studied from

comes

from

perspectives;

economics

and

sociology;

view

the other

can

back

to

the structural

conquences

illustrated cannot

by

the current

crisis,

which be

explained by

Thus,

we

level,

often

that to

are

hard on

predict, systems

as

originated

physics

in rearch

simply reprent

agents. countries,

sight system's

the failure

of few

a

major Plaza, Irvine, 92697,

need fundamental in

a more

dynamics they

and how

into the

and In

computer

individual

agents,

which

the different

can

reprent

firms, banks,

or even

and

where links

between the nodes

reprent

their

complex

in

both,

nodes

science.

XETH Switzerland.

Zurich, D-MTEC, 5,

Kreuzplatz

8032

Zurich,

di Economia

e

Management

(LEM),

Scuola

Superiore

2Laboratorio

Piazza Martin delta Liberta

3Swiss

Sant'Anna, 33,56127 Pisa,

Italy.

c/o

Finance 40 Boulevard Du

of

Pont

d'Arve,

1211 Geneva Switzerland.

4,

4Economics

Department,

Via della Piazzuola

43, Institute,

50133

European University

instituto Valenciano de

Investigaciones

Firenze,

Italy,

Econ?micas,

22 e. 2 no

1,46020 Valencia, Civil,

Spain.

6School

Calle Guardia

of

Informatics and Pervasive

Technology

Indiana

47408,

USA.

University,

919 East 10th

Street,

Bloomington,

IN

10133

Torino,

Italy, Interchange,

institute

for Scientific

institute

of Mathematical Behavioral

Sciences,

University

of

California,

3151

Social Science CA

USA.

*To whom

correspondence

fschweitzer@

should be addresd.

E-mail:

422

24 JULY 2009

VOL 325

SCIENCE

This content downloaded from 113.240.234.241 on Wed, 02 Dec 2015 14:14:54 UTC

All u subject to JSTOR Terms and Conditions

SPECIALSECTION

in which central "hub" channels all conflict between

a

ganization,

In

this "micro"

communication their the overall ef

among

agents. gate

we on

focus the individual in the

system

perspective performance ficiency

elements and their detailed network of rela this is exacerbated if

tions. for environment is

"macro" focus the statis which for

In to

contrast,

large adopts underlying

tups,

one a

that to the

on

perspective volatility,

individual incentives and how

or on on

impact

welfare,

of the network

problem large.

Furthermore,

the on

aggre

at on

examine the

the basis of the

rules be

might grounded

economic incentives

of the

agents.

Thus,

instead of

focusing

understand

behavior of

individual

agents, ing

the

endogenous

the centers

complex-systems subject persistent

approach

standing

how be due

the network-formation

tematically

Networks real

affect innovation

under

rules

sys

the link structure

emerging

(4). regularities ephemerality

stochastic different

generated

with

tical of intrinsic of and if

the network whole. Each

as a

and has its in most of

disadvantages. approach advantages equilibrium,

on was

the micro If this is the or

Previous work world situations. it is reasonable

perspective

strongly oversimplifying assumptions posit

rooted follow that

in

on on

both the of the network and

structure

the micro For behaviors

ap conditions, i.e.,

(3).

example, agents' experiences.

agent agents

incentives have com

proach emphasized

may

and have failed real

may

to

successfully predict ever-changing

outcomes. can

The across istic in

macro

approach changes

for the better

large-scale suggests

system conquences

proper

may,

(9),

as are

agents

out

ca,

simple

bounded world

to

agents

example,

rational mies that modified

are

in of their

light

under

such and infrastruc

However,

are

to attain

unable

efficient road

tions,

despite

an

small structures environmental

their continuous

situation.

configura

to to

adapt

even

Additionally, (1, 10).

efforts internet and mobile

such

as

random,

scale-free small

algorithms,

networks,

have been with

compared

real

networks tho

in

biology, including

complex

genetic

networks;

metabolic

networks and

power

grids;

ture,

i.e.,

munication, i.e.,

phone;

and

interaction, i.e., 2,

collaborations

network

the dif

Comparing volatility

net ferent

that economic

disciplines configu

works reflect

may

also similar

a

universality

(77). [e.g., (Fig. 1)].

an

the connections

of banks in

interbank

Indeed,

network that have been ob

(12, approaches, produce

13), regularities

show the fat

tail,

characteristic

of

a

scale-free that that

system,

indicates

only

a

few banks

interact with In this

many

others.

ex

similar investment behavior with

ample, large-scale

banks for networks.

in the network. made from the of will

Similar cluster

regularities

examples including

also

the

can

be traced affect formation and information

for

many

social

in the

development

of informal links within firms

dynamic

accounts

ties,

but fails in the to the economic

linking

motivation of individual

agents (4).

In recent micro

were as a

often the result of network viewed rved in network

formation and of that

economic networks statistical

cooper game among

include firms

on

job

oppor

have drastic in the overall

ration of the

system

to re

The of

inability previous

approaches

empirically justifies

structures

our a

pursuit competing

complex-systems approach

may

provide predictions ating regard,

The

are

predictions testing

stochastic mies that link workers who share

agents. agents

In this

that collaborate in

joint

R&D

projects (5)

or

tunities their links added deleted

(6);

are or as

the of decisions of

conquence

ing payoffs. degree connectivity

to maximize are

purpoful regional

attempt

their In this their

context,

rely anticipate centrality, binary

on on or as

(and

be able the their measured

to)

that to some sort

take into in addition

account,

randomness,

the characteristic features of the

agents,

such of

as

(num

ber of not

links) agents

basis of the in what others do and

importance weighted according generally

of

a a

node?which,

turn,

can u

be affected its links nodes. under information about in network

by asymmetric

to other

the

complex-systems approach major

However,

pos

tulates mies and does time within bounded

exogenously explicitly problem necessarily (Fig. 2)].

not

international

trade network and

(ITN) (14, 15)

(16).

investment networks

or

ownership

In the

complex-network

context,

"links"

not must

(existing existing),

consideration

of

financial institutions worldwide shown

but

are or

to the

economic interaction

may

(in

imperfect

their

manner); [for example,

environment be frame the

(which limited);

may

horizon;

may

create

a can

biad and their

ilar situations encountered later.

The tended considerations

some

and

learn from the which

past,

if sim

experience

are

to re

a

-,-,-,-,-,-,-,-,-,

in

Furthermore,

links traded

reprent

invested and

capital,

so

on,

volumes,

change weight

over

time.

Distinguishing

levels where consider

undirected sult number of and

networks

we or

at

different

directed

or

unweighted weighted

MMM

1

in

a

dramatically large

that choo from

agents

must

options

on

the of

basis limited information. their

net

The micro of economic When the

works relies which

analysis foreign

on

game

theory,

8LJMjp%

^^^*J%1

lM

f

o/v

=

ioo

-I

=

1000

Q

A/

A/=

5000

1

links

helps

illuminate the evolution of

topological properties.

are a

prented

power-law scaling

direct invest

5

0.6

-

/

M

aims at Nash

identifying equilibria (i.e.,

situations that stable

are

strategically

in

the that has incentive the number

n no an

agent

*sB?Ej^-a

"?

ments

(FDI) European

among

firms

as

directed

scaling depends

on em

network,

is obrved. This

of

of invested and the number

of and have been and

to

deviate).

It also

can on

rely

opera

tions where for firm

rearch, in,

algorithms

optimizing arching incoming

As the number of nodes

developed.

and links scales about the

possible point predictions

up,

however,

to

solve,

and classical the basis of the and

approaches activity

literature The has Similar structural

O 1

0.2 0.4 0.6 0.8

Environmental

volatility

ployees investing

in both the

and the

on

outgoing

investments

both firms This allows time

(16)

single

investments

on or

that will

regions

receive

make,

connec

transitions

can

such become difficult

problems

very

are

unsatisfactory.

game-theoretic

tives induced in terms

of network

an

equilibrium strongly efficiency

Fig.

1. The structure and

of their firms.

can

on the

form links. We

tivity

conditions under which

depend myopic agents

the of their that have

neighbors

neighbors higher centrality,

a

which

creates local shortcuts. Network

measured the basis of is

on

efficiency

the of

aggregate agents.

centrality

Environmental the

volatility

measures

show simulations that that

computer

assume

agents

prefer

to connect to

ing country's endogenous centrality

a

highlighted

the crucial role of incen also be detected in the ITN.

in the

and of

behavior of socioeconomic networks

(3,

7,

8).

However,

macro can

approaches

this micro

By weight

with

has not been links

typically integrated existing

complex systemic weight, changes

out cannot

this

information,

understand such the that

important divergence

that the

identify

forces work.

at

With the relative in

fully

as

issues,

we

approach

the likelihood that additional

any

given

dollar traded in the

world reaches that

risk to an

that if is it will

any exogeneous shock,

single agent expod

force the deletion of the loss of links the network country by following

one

link. If

pushes

down and environmental critical

up

past

some

level,

with to its

a

probability proportional

efficiency volatility

the network structure will break down into

strongly homogeneous

a

central

in network

with breakdown

a

accompanied efficiency.

SCIENCE JULY 2009 423

VOL 325

24

spar, ity

hierarchical similar to a structure

structure,

core-periphery

and is

over

time show trends for different

in

predict

countries

This content downloaded from 113.240.234.241 on Wed, 02 Dec 2015 14:14:54 UTC

All u subject to JSTOR Terms and Conditions

Pushing

Networks

to

the Limit

Royal

Bank Scotland

Sumitomo

0 4^^

Generali

Mitsubishi UFJ

^

g|

?

Lloyds

TSB

*HBOS

Time and

space.

By allowing time-dependent

a

net

resolution of the of economic

properties

works,

we a

will

be able

to move

beyond single

snapshot approach.

Gen.

Electric,

Bear

Intesa-Sanpaolo

~

i

I

Prudential

Fin.

to

identify evolutionary path

the of networks

of the combination

complementary

infor

illustration of this is mation

pro

This will allow the rearcher

through

sources.

A

good

ING a

Aberdeen

vided the R&D networks in the field of human

by

biotechnology (20),

which follow

predictable

life related the

cycle timing

to the

and of

integration knowledge.

Structure

topology reported particular

gregated

identification. Extracting

of

exchange

Commerzbank

Mediobanca

(

Unicredito

network

from in

data,

for

ag

economic is

data,

very

difficult. This is

true

for the

banking

ctor,

where

particularly

detailed of debt-credit relations

accounts

are not

publicly although

theoretical decom

available,

of data

aggregated

have been studied

then,

analys reading Fidelity Mng,

may

remble

and reveal known leaves

only previously

or

Friends Provl

7?

Barclays

positions

(13).

Even

tea

Deutsche

Bank

Franklin Res.

Wellington Mng.

Merrill

Lynch

predicted regularities

information. in

Statistical

eco

can

be identified

nomic networks

through large

in

scale data but difficulties the

ts,

asssing

relevance of the various remain.

measures

In an

evolving require

economic informa

tion about their function and their

agents'

network,

we

roles,

where the nodes

reprent

major

financial of the international

influence New methods needed

(23).

are

to iden

Fig. sample

2. A finan?ai

network,

are

both directed and and the links

weighted

and the relations

reprent strongest existing

institutions

are new

needed and

to

patterns,

concepts

areas:

different

Union members North

(red),

tify

among

them. Node colors

express

European

geographical

quantify

both direct and indirect influence

(e.g.,

with the reduced number of links

America other countries

(blue),

(green).

Even

displayed

in the

figure,

in the ITN

The identification

through ownership).

relative

to the true world the network shows

economy,

a

high connectivity

among

the financial in

in the net

of such roles bad similar

on

positions

veral nodes. This indicates that

stitutions that have

mutual and clod

share-holdings loops involving

work that have

suggests steps

promising begun

to

which affect market and

may

competition strongly interdependent,

systemic

risk

the financial

ctor is

functional roles interactive

agents

identify played by

and make the network

vulnerable

to

instability.

the within networks

global

economy

regional integration

and do better

so to

than traditional international

trade and economic network small

and macroeconomic statistics

becau facilitate

(17). statistically optimal

This is also favor networks

not account in the

the latter do for

the entire economic shocks and that statistical of the

1980 and this endeavor. needed between

huge by cross-cutting experienced

in data Our

of interactions. We then be able of their multirelational interaction network

should

that relate

to structure

specific

patterns

in the link

(24).

predict

propo

that

policies large homologous Mapping

are more

structures

that robust with

to ts of distinctive

can a ca

describe

tion trade. is for New World-Afro-East

or

Below,

we

briefly alignments

to tackle

Massive each of their interconnected

more as

integra gives correspondence

what ITN world

a as a

one,

roles,

Asian North and Central Eurasia rect

versus

alignments

cores,

in world

sys

and and better data will foster the

peripheries, miperipheries,

tem to a

models,

but with countries fell. The Latin American sition

much and from

greater

precision. qualitative quantitative

Beyond simplicity. computational

All

economic networks science. As

are

with their and the astonish network data

respect agents

to both

heterogeneous large-scale

interaction and also ered for different of the levels

strength strongly (e.g.,

can

vary

in

time Previous studies trade and

(25).

of efficient and

(i.e., countries),

not

networks assumed

homogeneity. processing

However,

as

the should data methods

network

topology

but

only

consider bilateral di

trade links. For

example,

2005 East Asian

countries

creas

in their but the

centrality centrality quality

scores,

ranking

of

most

trade statistics

of

the evidence-bad

regions, displayed

however, power

similar

patterns. gath

In other be

words, increas,

records different

were

not well

ingly

development

macroeconom

tracked international and models

by

ics statistics. network-bad

Thus, syn

approaches through generation

may

analysis. ability

to obtain

tran

a

can

firms, industries,

be tested

economy

can

manage, way

monitor, thetic,

provide powerful

a more

to ts. New

economics wide

and data

govern

complex

systems.

a on or can

focus other such be communication

However,

centrality exploited.

can a

only provide prediction possible

first be

of networks

properties

the role of the cific interactions time such

order classification that

emphasizes

and and R&D

randomness fluctuations

cannot

predict

the of

underlying dynamics

the whether become of firm-bank credit market

agents,

they anticipate

are or to

firms We

countries.

that the data tools

next to

generation

of rearch be able

will and the scale of available infor

mea

sure at

any

deviations

from and will mation interactions

universality reflecting

the of

large,

a

range open

of business and internet

wasteful) equilibrium (or strategically stable)

and

differences between and

weighted unweighted

for the ITN indicate

ca

(14),

network

properties

any may spe

of transitions fail under data individualized

pha gather

In

fact, flows,

heterogene simplified assumptions. employee

ities of and within business

agents prevent

can so on a

turn out to

pha

tran

sitions, i.e.,

a source or

stability. large

Simple amplification digest

Systemic feedbacks. require powerful

that It will then

to

on

over as

collaborations,

interactions. The

streams

more

manipulate

huge

mechanisms

(such

as

herding)

can

dominate the

network the best and network

tentions of the Economic information Databas

large, despite dynamics

in

are

networks to

agents.

with individual and their decision

agent subject amplifications

dynamics

making

vidual their

process.

agents (22).

This of rearch

new wave

with obrvations

allow the associated

us

identify idiosyncrasies properties.

agent

this

containing

therefore

complement

both theoretical

may

should

begin

to debt in bank and in credit

merge (20, 21)

the of indi economic network studies

description

strategies coevolving large-scale

from the that

may

result

to

nomic network

experiments

(18,19) (e.g.,

and redistribution of the load if node fails

empirical

one

a

provide electricity grid

a or

power

eco

in real-time If node it force

ing network). single

a

fails,

may

424

24 JULY 2009

VOL 325 SCIENCE

This content downloaded from 113.240.234.241 on Wed, 02 Dec 2015 14:14:54 UTC

All u subject to JSTOR Terms and Conditions

SPECIALSECTION

other

nodes

to fail as

well,

which

lead failure and a

to

cascades the breakdown of

system,

particular

reprent

connected financial

not

well how vances understood

financial a network

denoted

may

eventually appropriate

the their

as

systemic

risk. This in

applies

to

financial

networks where

links are from

debts

affects the

and claims

between a

it is

of

a

institutions.

of

economic

description

agents

and 93,

interactions,

and

systemic

perspective

be

of

global understanding

effects

as

stowing

a new

interactions

coming

varying ]. Schiavo,

network

needed. We (2008).

predict

that such create

studies will

more

unified

field of

economic

networks that ad

our structure

understanding

and leads to further

insight

We still far from

are

satisfactory

identification

and G.

integration

of the Am. Socio!. Rev.

many topical Although

components,

but

the

recent a methods

advances outlined start.

suggest

promising

15. D.

Garlaschelli,

M. I.

Loffredo,

Phys.

Rev. Lett

188701

(2004).

16. S.

Battiston, ].

F.

Rodrigues,

H.

Zeytinoglu,

Adv.

Complex

Syst.

10,

29

(2007).

17.

Reyes,

S.

G.

Fagiolo,

Adv.

Complex Syst.

11,

685

18. M.

Kosfeld,

Rev. Netw. Econ.

3,

20

(2004).

19. S.

Callander,

C.

Plott, ]. 89,

Public Econ.

1469

(2005).

20. W.

Powell,

D. K.

White,

Koput, ].

Owen-Smith,

Am.

J.

Socio!. 1132 probability

110,

(2005).

21. B.

Kogut,

Walker, (2001).

66,

317

22. D.

Sornette,

F.

Deschatres,

T. Y.

Gilbert,

Ageon, Phys. suited

Rev.

Lett.

93, (2004).

228701

23. T. A.

Snijders,

G. G. van de

Bunt,

C. E.

Steglich,

Soc.

Networks, 1. F.

in

press; Social Networks

published

online 26 March

2009

(10.1016/.2009.02.004).

24.

]. Reichardt, White, (2007). 2. A. M.

D.

Eur.

Phys.

]. risk has

B 217

60, A.

25. A.

Kirman, 7,

;. Networks

Evol. Econ.

339

(1997).

26. F.

Allen,

D.

Gale, J. 108,

Polit Econ.

1

(2000).

27. S.

Battiston, Gatti, 3. M. 0.

D. Delli

M.

Gallegati,

B.

Greenwald, Econ.

]. (2007). dynamic 4. R.

Stiglitz,

J. A.-L.

Econ.

Dyn.

Control

31,

2061

28. We would

like C. a

to thank M.

K?nig,

]. Battiston, 477

Tessone, Res.

S.

and S. Vitali

(ETH

for

aid with

figures

and S.

Zurich) White

on

the text. Data for 2 were

for

commentary

Fig. provided

by 1995)

Orbis of

Databa Van F.S.

(end and

2007),

Bureau

Dijk.

D.S. agents

acknowledge

financial

support

from the ETH

Competence

with Cris in

Socio-Economic less

Center,

"Coping Complex

Grant