Social Times of Network Spaces

David Stark and Balazs Vedres

to model, from its inception, network formation across an entire epoch of economic transformation

Processes of network evolution

Embeddedness of foreign capital?

analytic move from how a national economy is integrated into the world economy to whether and how FDI is integrated into local networks

Methodological innovation

We modify analytic tools from DNA sequencing to reconcile the structural focus of social network analysis with the temporal orientation of historical sociology

Structure as topology and temporality

Emergence of domestic networks

Massive decline of state ownership

Extraordinary institutional uncertainty

Ambiguity about the rules of the game

Foreign investment

Did massive FDI eradicate networks?

Which forms were open or closed to FDI?

Do foreigners build domestic networks?

Our question

Can networks of global reach coexist and entwine with those of local embeddedness?

Restated, can FDI be integrated into national networks? And, if so, how?

Data Largest 1,800 firms of the period by revenue between 1987-2001

Ownership data from registry courts

names of top 25 owners and their shares all changes recorded for the whole life of the

firm

A tie is a direct ownership stake by one of our 1,800 firms in one of the other firms in that same population (i.e., not an “affiliation network”)

The ‘network movie’

Animation of network emergence

Month 1 December, 1987

Month 2 January, 1988

Month 3 February, 1988

Month 4 March, 1988

Month 5 April, 1988

Month 6 May, 1988

Month 7 June, 1988

Month 8 July, 1988

Month 9 August, 1988

Month 10 September, 1988

Month 11 October, 1988

Month 12 November, 1988

Month 13 December, 1988

Month 14 January, 1989

Month 15 February, 1989

Month 16 March, 1989

Month 17 April, 1989

Month 18 May, 1989

Month 19 June, 1989

Month 20 July, 1989

Month 21 August, 1989

Month 22 September, 1989

Month 23 October, 1989

Month 24 November, 1989

Month 25 December, 1989

Month 26 January, 1990

[Continues to 169th month ]

For historical network analysis

from a kind of aerial sociology to the network histories of 1,800 firms.

To move from system-level properties to historical processes at the level of firms ...

For historical network analysis Network analysis: topology Historical analysis: temporality Synthesis: find structures in social space and

social times Methodological innovation: Sequence analysis

of network positions to identify pathways through local network topologies

From time as a variable to time as variable

Probe for differences in types of embeddedness Different local network topographic

properties reflect different organizing practices

Firms can use network properties, for example, to hide assets, to restructure assets, to gain access to knowledge, to increase legitimation, to secure access to supplies and markets, and so on

Structure as topology and temporality

Studying variation in the sequences of local structures is a way to identify distinctive pathways of network evolution

1990

1989 1990 1991

7. Member of a strongly cohesive group

6. Member of a cohesive group

5. Star center

4. Large star periphery

3. Small star periphery

2. Dyad component member

1. Isolate

GraphColorName

From 1,696 firm histories we need to find similar sequences. We use optimal matching analysis to find the distance of each sequence from all others.

Finding sequential equivalence

To the resulting matrix we then apply hierarchical clustering that groups sequences so that within-cluster distances are as low as possible and between-cluster distances are high.

The combination of these two algorithms, yields – not unlike the concept of structural equivalence in network analysis – sequential equivalence.

Sizeable foreign ownership

in 2001 (Yes = 1)

model 1

model 2

Star-periphery recombinants

1 (I-S) -5.513** -5.781** 2 (P) -.422** -.785**

Cohesive recombinants

3 (I-P-C-P) -.065** .622** 4 (C-G-C) .485** 1.112** 5 (C-G-I) 1.327** 2.047** 6 (I-L-C-G) -1.091** -1.341**

Startups

7 (P-I) 1.565** 2.087** 8 (D-I) .342** 1.076** 9 (P-D) 1.419** 2.756**

Second wave networks

10 (I-D-P) 1.218** 1.752**11 (D-P) 1.184** 1.717**

Foreign ownership in 2001: Logistic regression estimates

Independent variablesPathways

Industry

Agriculture -2.973** Food 2.779** Energy and mining .996** Chemical 4.756** Heavy industry 1.768** Light industry and textile

.378** Construction -.517** Wholesale .391** Retail 3.695** Finance .359**

Local network position in 2001

D (dyad) -.720** P (small star periphery)

-.097** L (large star periphery)

1.892** S (star center) .140** C (cohesive cluster) -.039** G (strongly cohesive group)

-2.737**Early foreign ownership (1990) 4.326**

Constant .205** -.935**N 1286..….... 1286..…....-2LL 1709.03…. 1326.78….R-squared .249... .498...Percentage correctly classified 66.7…... 74.8…...χ2 (df) 302.45 (11) 684.71 (28)p-value .000… .000…

Hungary is not a segregated dual economyGlobalization is compatible with local

embeddingsForms of recombinant property are robust Cohesive forms are adaptive

An internationalized market economy emerged in Hungary not despite but, instead, because of inter-organizational ownership networks.

Developing economies do not necessarily face a forced choice between networks of global reach and those of local embeddedness.

High levels of foreign investment can be integrated into processes of inter-organizational ownership network formation in a developing economy.