复杂网络科学导论 (7).pdf

《复杂网络科学导论 (7).pdf》由会员分享，可在线阅读，更多相关《复杂网络科学导论 (7).pdf（6页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、The Network of Global Corporate ControlStefania Vitali,James B.Glattfelder,Stefano Battiston*Chair of Systems Design,ETH Zurich,Zurich,SwitzerlandAbstractThe structure of the control network of transnational corporations affects global market competition and financial stability.So far,only small nat

2、ional samples were studied and there was no appropriate methodology to assess control globally.Wepresent the first investigation of the architecture of the international ownership network,along with the computation of thecontrol held by each global player.We find that transnational corporations form

3、 a giant bow-tie structure and that a largeportion of control flows to a small tightly-knit core of financial institutions.This core can be seen as an economic super-entity that raises new important issues both for researchers and policy makers.Citation:Vitali S,Glattfelder JB,Battiston S(2011)The N

4、etwork of Global Corporate Control.PLoS ONE 6(10):e25995.doi:10.1371/journal.pone.0025995Editor:Alejandro Raul Hernandez Montoya,Universidad Veracruzana,MexicoReceived March 29,2011;Accepted September 15,2011;Published October 26,2011Copyright:?2011 Vitali et al.This is an open-access article distri

5、buted under the terms of the Creative Commons Attribution License,which permitsunrestricted use,distribution,and reproduction in any medium,provided the original author and source are credited.Funding:The authors acknowledge financial support from the ETH Competence Center Coping with Crises in Comp

6、lex Socio-Economic Systems(CCSS)through ETH Research Grant CH1-01-08-2;the European Commission FP7 FET Open Project FOC No.255987.The funders had no role in study design,datacollection and analysis,decision to publish,or preparation of the manuscript.Competing Interests:The authors have declared tha

7、t no competing interests exist.*E-mail:sbattistonethz.chIntroductionA common intuition among scholars and in the media sees theglobal economy as being dominated by a handful of powerfultransnational corporations(TNCs).However,this has not beenconfirmed or rejected with explicit numbers.A quantitativ

8、einvestigation is not a trivial task because firms may exert controlover other firms via a web of direct and indirect ownershiprelations which extends over many countries.Therefore,acomplex network analysis 1 is needed in order to uncover thestructure of control and its implications.Recently,economi

9、cnetworks have attracted growing attention 2,e.g.,networks oftrade 3,products 4,credit 5,6,stock prices 7 and boards ofdirectors 8,9.This literature has also analyzed ownershipnetworks 10,11,but has neglected the structure of control at aglobal level.Even the corporate governance literature has only

10、studied small national business groups 12.Certainly,it isintuitivethateverylargecorporationhasapyramidofsubsidiariesbelowandanumberofshareholdersabove.However,economic theory does not offer models that predicthow TNCs globally connect to each other.Three alternativehypotheses can be formulated.TNCs

11、may remain isolated,cluster in separated coalitions,or form a giant connectedcomponent,possibly with a core-periphery structure.So far,thisissue has remained unaddressed,notwithstanding its importantimplications forpolicymaking.Indeed,mutual ownershiprelations among firms within the same sector can,

12、in some cases,jeopardizemarketcompetition13,14.Moreover,linkagesamong financial institutions have been recognized to haveambiguous effects on their financial fragility 15,16.Verifyingto what extent these implications hold true in the global economyis per se an unexplored field of research and is bey

13、ond the scope ofthis article.However,a necessary precondition to such investi-gations is to uncover the worldwide structure of corporatecontrol.This was never performed before and it is the aim of thepresent work.MethodsOwnership refers to a person or a firm owning another firmentirely or partially.

14、Let W denote the ownership matrix,wherethe component Wij0,1?is the percentage of ownership that theowner(or shareholder)i holds in firm j.This corresponds to adirected weighted graph with firms represented as nodes andownership ties as links.If,in turn,firm j owns Wjlshares of firm l,then firm i has

15、 an indirect ownership of firm l(Figure 1 A).In thesimplest case,this amounts trivially to the product of the shares ofdirect ownership WijWjl.If we now consider the economic value vof firms(e.g.,operating revenue in USD),an amount Wijvjisassociated to i in the direct case,and WijWjlvlin the indirec

16、t case.This computation can be extended to a generic graph,with someimportant caveats 17,Appendix S1,Sections 3.1 and 3.2.Each shareholder has the right to a fraction of the firm revenue(dividend)and to a voice in the decision making process(e.g.,voting rights at the shareholder meetings).Thus the l

17、arger theownership share Wijin a firm,the larger is the associated controlover it,denoted as Cij.Intuitively,control corresponds to thechances of seeing ones own interest prevailing in the businessstrategy of the firm.Control Cijis usually computed fromownership Wijwith a simple threshold rule:the m

18、ajorityshareholder has full control.In the example of Figure 1 C,D,this yields Cijvj1vjin the direct case and CijCjlvl0 in theindirect case.As a robustness check,we tested also moreconservative models where minorities keep some control(seeAppendix S1,Section 3.1).In analogy to ownership,the extensio

19、nto a generic graph is the notion of network control:cnetiXjCijvjzXjCijcnetj.This sums up the value controlled by ithrough its shares in j,plus the value controlled indirectly via thenetwork control of j.Thus,network control has the meaning of thetotal amount of economic value over which i has an in

20、fluence(etivjzvkin Figure 1 D).Because of indirect links,control flows upstream from manyfirms and can result in some shareholders becoming very powerful.PLoS ONE|www.plosone.org1October 2011|Volume 6|Issue 10|e25995However,especially in graphs with many cycles(see Figures 1Band S4 in Appendix S1),t

21、he computation of cnet,in the basicformulation detailed above,severely overestimates the controlassigned to actors in two cases:firms that are part of cycles(orcross-shareholding structures),and shareholders that are upstreamof these structures.An illustration of the problem on a simplenetwork examp

22、le,together with the details of the method areprovided in Appendix S1,Sections 3.23.4.A partial solution forsmall networks was provided in 18.Previous work on largecontrol networks used a different network construction methodand neglected this issue entirely 11,Appendix S1,Sections 2 and3.5.In this

23、paper,by building on 11,we develop a newmethodology to overcome the problem of control overestimation,which can be employed to compute control in large networks.ResultsWe start from a list of 43060 TNCs identified according to theOECD definition,taken from a sample of about 30 millioneconomic actors

24、 contained in the Orbis 2007 database(seeAppendix S1,Section 2).We then apply a recursive search(FigureS1 and Section 2 in Appendix S1)which singles out,for the firsttime to our knowledge,the network of all the ownership pathwaysoriginating from and pointing to TNCs(Figure S2 in AppendixS1).The resu

25、lting TNC network includes 600508 nodes and1006987 ownership ties.Notice that this data set fundamentally differs from the onesanalyzed in 11(which considered only listed companies inseparate countries and their direct shareholders).Here we areinterested in the true global ownership network and many

26、 TNCsare not listed companies(see also Appendix S1,Section 2).Network TopologyThe computation of control requires a prior analysis of thetopology.In terms of connectivity,the network consists of manysmall connected components,but the largest one(3/4 of all nodes)contains all the top TNCs by economic

27、 value,accounting for94.2%of the total TNC operating revenue(Table 1).Besides theusual network statistics(Figures S5 and S6 in Appendix S1),twotopological properties are the most relevant to the focus of thiswork.The first is the abundance of cycles of length two(mutualcross-shareholdings)or greater

28、(Figure S7 and Section 7 inAppendix S1),which are well studied motifs in corporategovernance 19.A generalization is a strongly connected component(SCC),i.e.,a set of firms in which every member owns directlyand/or indirectly shares in every other member.This kind ofstructures,so far observed only in

29、 small samples,has explanationssuch as anti-takeover strategies,reduction of transaction costs,risksharing,increasing trust and groups of interest 20.No matter itsorigin,however,it weakens market competition 13,14.Thesecond characteristics is that the largest connect componentFigure 1.Ownership and

30、Control.(A&B)Direct and indirect ownership.(A)Firm i has Wijpercent of direct ownership in firm j.Through j,it hasalso an indirect ownership in k and l.(B)With cycles one has to take into account the recursive paths,see Appendix S1,Section 3.1.(C&D)Thresholdmodel.(C)Percentages of ownership are indi

31、cated along the links.(D)If a shareholder has ownership exceeding a threshold(e.g.50%),it has fullcontrol(100%)and the others have none(0%).More conservative model of control are also considered see Appendix S1,Section 3.1.doi:10.1371/journal.pone.0025995.g001The Network of Global Corporate ControlP

32、LoS ONE|www.plosone.org2October 2011|Volume 6|Issue 10|e25995contains only one dominant strongly connected component(1347nodes).Thus,similar to the WWW,the TNC network has a bow-tiestructure 21(see Figure 2 A and Appendix S1,Section 6).Itspeculiarity is that the strongly connected component,or core,

33、isvery small compared to the other sections of the bow-tie,and thatthe out-section is significantly larger than the in-section and thetubes and tendrils(Figure 2 B and Table 1).The core is also verydensely connected,with members having,on average,ties to 20other members(Figure 2 C,D).As a result,abo

34、ut 3/4 of theownership of firms in the core remains in the hands of firms of thecore itself.In other words,this is a tightly-knit group ofcorporations that cumulatively hold the majority share of eachother.Notice that the cross-country analysis of 11 found that only afew of the national ownership ne

35、tworks are bow-ties,and,Figure 2.Network topology.(A)A bow-tie consists of in-section(IN),out-section(OUT),strongly connected component or core(SCC),and tubesand tendrils(T&T).(B)Bow-tie structure of the largest connected component(LCC)and other connected components(OCC).Each section volumescales lo

36、garithmically with the share of its TNCs operating revenue.In parenthesis,percentage of operating revenue and number of TNCs,cfr.Table 1.(C)SCC layout of the SCC(1318 nodes and 12191 links).Node size scales logarithmically with operation revenue,node color with network control(from yellow to red).Li

37、nk color scales with weight.(D)Zoom on some major TNCs in the financial sector.Some cycles are highlighted.doi:10.1371/journal.pone.0025995.g002Table 1.Bow-tie statistics.TNC(#)SH(#)PC(#)OR(%)LCC154914781939969694.17IN28252051292.18SCC2950102318.68OUT6488031807359.85T&T8426426148047113.46OCC27569296

38、37802965.83Percentage of total TNC operating revenue(OR)and number(#)of nodes inthe sections of the bow-tie(acronyms are in Figure 2).Economic actors typesare:shareholders(SH),participated companies(PC).doi:10.1371/journal.pone.0025995.t001The Network of Global Corporate ControlPLoS ONE|www.plosone.

39、org3October 2011|Volume 6|Issue 10|e25995importantly,for the Anglo-Saxon countries,the main stronglyconnected components are big compared to the network size.Concentration of ControlThe topological analysis carried out so far does not consider thediverse economic value of firms.We thus compute the n

40、etworkcontrol that economic actors(including TNCs)gain over theTNCs value(operating revenue)and we address the question ofhow much this control is concentrated and who are the top controlholders.See Figure S3 in Appendix S1 for the distribution ofcontrol and operating revenue.It should be noticed th

41、at,although scholars have longmeasured the concentration of wealth and income 22,there isno prior quantitative estimation for control.Constructing aLorenz-like curve(Figure 3)allows one to identify the fraction g?of top holders holding cumulatively 80%of the total networkcontrol.Thus,thesmallerthisf

42、raction,thehighertheconcentration.In principle,one could expect inequality ofcontrol to be comparable to inequality of income acrosshouseholds and firms,since shares of most corporations arepublicly accessible in stock markets.In contrast,we find that only737 top holders accumulate 80%of the control

43、 over the value ofall TNCs(see also the list of the top 50 holders in Table S1 ofAppendix S1).The corresponding level of concentration isg?10:61%,to be compared with g?24:35%for operatingrevenue.Other sensible comparisons include:income distributionin developed countries with g?3*5%10%22 and corpora

44、terevenue in Fortune1000(g?4*30%in 2009).This means thatnetwork control is much more unequally distributed than wealth.In particular,the top ranked actors hold a control ten timesbigger than what could be expected based on their wealth.Theresults are robust with respect to the models used to estimat

45、econtrol,see Figure 3 and Tables S2 and S3 in Appendix S1.DiscussionThe fact that control is highly concentrated in the hands of fewtop holders does not determine if and how they are interconnect-ed.It is only by combining topology with control ranking that weobtain a full characterization of the st

46、ructure of control.A firstquestion we are now able to answer is where the top actors arelocated in the bow-tie.As the reader may by now suspect,powerfulactors tend to belong to the core.In fact,the location of a TNC inthe network does matter.For instance,a randomly chosen TNC inthe core has about 50

47、%chance of also being among the topholders,compared to,e.g.,6%for the in-section(Table S4 inAppendix S1).A second question concerns what share of totalcontrol each component of the bow-tie holds.We find that,despiteits small size,the core holds collectively a large fraction of the totalnetwork contr

48、ol.In detail,nearly 4=10 of the control over theeconomic value of TNCs in the world is held,via a complicatedweb of ownership relations,by a group of 147 TNCs in the core,which has almost full control over itself.The top holders within thecore can thus be thought of as an economic super-entity in th

49、eglobal network of corporations.A relevant additional fact at thispoint is that 3=4 of the core are financial intermediaries.Figure 2D shows a small subset of well-known financial players and theirlinks,providing an idea of the level of entanglement of the entirecore.This remarkable finding raises a

50、t least two questions that arefundamental to the understanding of the functioning of the globaleconomy.Firstly,what are the implication for global financialstability?It is known that financial institutions establish financial Figure 3.Concentration of network control and operating revenue.Economic a