Social networks of corporate citizens

The name of this blog, Memetix, reflects my interest in the propagation of memes¹. A related idea is that of social networks, about which much theoretical material has been written. Memes propagate from node to node (person to person) along the “edges” (technical term for links between people) of the network, and spread faster if they hit a node which is “central” in some sense.

As you can imagine, sociologists are quite interested in social networks. However, the study of social networks is also a big deal in business circles. A recent Wharton Business School article describes how and why companies are becoming interested in studying social networks:

Mapping social networks can be useful in many ways, but [Wharton management professor] Rosenkopf says there are at least two reasons why corporate interest in the subject is growing: Companies want to be able to identify key performers and get a better understanding of the nature of the interaction among employees.

"Hopefully, you have organized your company the best way to get the job done," she says. "But mapping out a network will give you a sense of whether actual work flow and communication flow match what you hope to achieve. Maybe there are bottlenecks where one person is managing all interactions. If you expect two groups to work together closely, and you don't see them doing this, you might want to create liaison roles or other relationships to make information flow better. On the other hand, you may see groups talking to each other too much. When managers see network diagrams, they often realize they need to reconfigure their organizational chart."

The internet is a gigantic social network, and has been studied by Albert-Laszlo Barabasi, an important mathematician. An MIT researcher, C. Marlow, is interested in studying how news travels through the blogosphere:

We have constructed a system to track memes within the weblog community. As a meme diffuses through social ties, our system documents the time and location for each posting that is observed. Based on these data, webloggers can be categorized by their adoption characteristics, ranging from early- to late-adopters.

Yes, dear friends, you are being studied.

I am starting to learn some of the techniques used by social network theorists. As an example (which has its own semi-enlightening results), in this article I will present a preliminary analysis of the network formed by our “corporate citizens”. As UCSC sociology professor G. William Domhoff states,

Interlocking directorates -- defined as the linkages among corporations created by individuals who sit on two or more corporate boards -- have been a source of research attention since the Progressive Era at the turn of the 20th century, when they were used by famous muckraking journalists, and future Supreme Court Justice Louis Brandeis, to claim that a few large commercial and investment banks controlled most major corporations.

Drawing inspiration from the site They Rule, which provides a graphical tool for researching corporate board memberships and interlocking directorates, I consider two corporations to be linked if they share a board member (or if an officer in one company sits on the board of another). Unfortunately, the data at They Rule is from 2004 and not available in a format necessary for social network analysis. Consequently, I spent the day downloading board information from Reuters. The data used for this analysis consists of the top 100 companies (in gross revenue), minus six companies for which data were difficult to obtain and including an additional 27 companies that are involved in defense, media, or pharmaceuticals. Clearly the 27 extra companies reflect the bias of my own personal interests. Eventually I plan to expand the analysis.

A few screen shots of the network (excluding isolated companies) appear above. Table 1 below shows the non-isolated companies, sorted by centrality degree (number of ties to other companies) and showing betweenness (how much of a bridge the company is) and closeness (how close the company is to others in the network). Unsurprisingly, Citigroup is a huge bridge and very linked, while General Electric, Verizon, and Honeywell are well-linked and represent bridges in the network. Oddly, Pepsico and Target are at the top of the table as well.

A statistical test for betweenness produces a significant P-value (~0.01), demonstrating that this network is more centralized by betweenness than the typical random network of the same size and density, i.e. there are a few key corporations. The corresponding test for degree centrality not significant (with respect to network size, P~0.6), indicating that this network is typical in terms of linkage when compared with other networks of its size.

Thus, it would appear that this social network of corporate citizens seems to have a very large conduit of corporate governance memes in the form of Citibank and perhaps few other companies. It is interesting to compare the results to Domhoff's article, where he uses Citibank as an example, and to the statistics mentioned in this 2002 USA Today article.

This is a work in progress. Please let me know if you have comments, any interesting ideas (or you would like to help download Reuters data!!)

¹Thought of more informally than in the most technical academic sense.

Table 1: Social Network Measures for Individual Companies

	Centrality	Between	Close		Centrality	Between	Close
Company	Degree	ness	ness	Company	Degree	ness	ness
Citigroup	34	1089	0.42	Dow Chemical	8	182	0.34
General Electric	20	455	0.38	Edison International	8	102	0.32
Pepsico	20	435	0.38	General Motors	8	52	0.33
Verizon	20	471	0.37	Hewlett Packard	8	74	0.31
Honeywell	18	337	0.35	Lockheed Martin	8	126	0.30
Target	18	400	0.38	New York Times	8	48	0.32
Du Pont	16	169	0.35	Northrop Grumman	8	65	0.30
Pfizer	16	123	0.34	Sears	8	112	0.31
Procter & Gamble	16	262	0.37	United Parcel	8	109	0.32
Walt Disney	16	322	0.33	Wachovia	8	123	0.32
Boeing	14	191	0.34	Yahoo	8	108	0.31
Chubb	14	261	0.34	ADP	6	58	0.33
Coca Cola	14	159	0.35	AIG	6	38	0.28
ConocoPhillips	14	183	0.36	Albertson's	6	27	0.27
Dell	14	166	0.34	CVS	6	129	0.25
Exxon	14	137	0.34	Federated Dept	6	102	0.28
Ford Motor	14	296	0.35	Google	6	114	0.27
Goldman Sachs	14	130	0.34	International Paper	6	22	0.31
IBM	14	236	0.35	Kroger	6	39	0.30
United Technologies	14	262	0.36	Lowe's	6	131	0.31
American Express	12	187	0.34	Merck	6	19	0.31
Bristol Myers Squibb	12	189	0.37	News Corp	6	5	0.30
Deere	12	135	0.35	TimeWarner	6	46	0.33
Dow Jones	12	111	0.34	United Health	6	21	0.31
FedEx	12	123	0.33	Viacom	6	38	0.28
Home Depot	12	241	0.34	Wellpoint	6	61	0.27
Merrill Lynch	12	147	0.32	WeyerHaeuser	6	29	0.28
Metlife	12	123	0.31	Wyeth	6	5	0.31
Motorola	12	127	0.36	Borders	4	15	0.26
Sprint Nextel	12	68	0.32	Calpine	4	0	0.30
Walmart	12	138	0.34	Costco	4	5	0.26
Wells Fargo	12	241	0.35	HCA	4	0	0.31
Abbott Labs	10	87	0.35	Ingram Micro	4	19	0.27
Alcoa	10	64	0.34	Lehman Brothers	4	8	0.30
Allstate	10	92	0.32	McKesson	4	29	0.28
AT&T	10	87	0.30	Raytheon	4	14	0.31
Bank of America	10	192	0.28	St. Paul Travelers	4	0	0.28
Cardinal Health	10	114	0.32	Sysco	4	0	0.23
CBS Corp	10	166	0.30	Washington Mutual	4	19	0.27
Chevron	10	152	0.32	Aetna	2	0	0.25
Cisco	10	210	0.34	Altria Group	2	0	0.26
Haliburton	10	84	0.35	Amazon.com	2	0	0.21
Intel	10	130	0.31	Barnes & Noble	2	0	0.24
Johnson & Johnson	10	117	0.34	Coca-Cola Bottling	2	0	0.24
Marathon Oil	10	99	0.34	Comcast	2	0	0.30
Microsoft	10	92	0.34	General Dynamics	2	0	0.24
Prudential Financial	10	70	0.33	Genzyme	2	0	0.27
Berkshire Hathaway	8	77	0.30	Hartford Financial	2	0	0.27
Caterpillar	8	18	0.30	Monsanto	2	0	0.23
Choicepoint	8	139	0.29	Safeway	2	0	0.22
Coca-Cola Enterprises	8	25	0.32	Sunoco	2	0	0.20
Consolidated Edison	8	199	0.32	Walgreen	2	0	0.26