Tuesday, August 23, 2011

the network of global corporate control

fighting the blogosphere;-)
(Edit: this post is a response to an on-line article, for more insights into our study and additional details on the concerns and misconceptions - especially the often voiced critique about inferring control from ownership and what control means in the financial sector -  please read this post.)

A recent news article about the network of global corporate control (Vitali, Glattfelder and Battiston) is floating around in the blogosphere. Ms. Yves Smith, from Naked Capitalism, was not amused. She choose an appropriate title expressing her disdain:
Study Asserts World’s Stocks Controlled by "Select Few" (Bad Studies That Confirm Conventional Wisdom Refuse to Die Edition) 

Especially as this is the second ownership network study, where the first one, the Backbone of complex networks of corporations: The flow of control, was already fiercely declared a deplorably bad study by Ms. Smith.

In her recent critique, we can read:
  • This paper is a garbage-in, garbage out analysis.
  • The problem is that the authors never bothered to understand how shares are held and how voting behavior varies based on institutional arrangements.
  • But the authors didn’t bother thinking about how [some issue] complicate their tidy story of big powerful institutions controlling the world.
and in the comments:
  • They used the wrong data. 
  •  [...] but the authors are physicists. They want to apply their network methodology to a large data set with lots of nodes. So they will be utterly unreceptive to the notion that the data they used is bad and their study is bunk.
  • This is classic drunk under the streetlight behavior 
As a co-author of the study, see the response, epilogue and conclusion sections I provide below...

(Note: somehow the comment section seems broken on their blog. You need to click the "No Comments" link at the end of the post to see the comments.)


Thanks for covering the article. I fully agree, inferring control from ownership is a tricky issue. I also think the conspiracy theory spin some media coverage gave the story was unnecessary.

However, I do believe the study has some merits. The fierce critique appears to stem from two issues (apart from the obvious fact that Ms. Smith thinks the whole enterprise of ownership network analysis is total and utter crap;-): the quality of the database and the notion of control, i.e., shareholder voting.

Obviously, everything in the study depends on Bureau van Dijk's (commercial) Orbis database:
http://www.bvdinfo.com/Products/Company-Information/International/ORBIS.aspx
It should perhaps be noted, that the database contains also natural persons and government entities next to corporations. So this is the garbage-in part.

Let's look at the garbage-out bit. It is maybe worth mentioning, that ownership is an objective and measurable quantity. So without even mentioning control, the study empirically uncovers that the global topology of transnational corporations is, similarly to technological networks, organized as a bow-tie, with a very small core of corporations, where 3/4 of the direct ownership stays in the core. This could be interesting for two reasons. Firstly, the obvious being that apparently no one has looked at this global network before. (Which, in itself, is also noteworthy, as many other empirical and interdisciplinary fields have successfully embraced network analysis to deal with the increasing accumulation of mountains of data. In economics, this is only starting slowly. Oh, but wait, physicists can only understand physics and any interdisciplinary effort is necessarily futile;-) Secondly, the economics literature would have us think that ownership relations of US and GB financial institutions should not organize as a tightly-knit group owning the majority in each other (see cross-shareholdings and business groups). The existence of the core could have implications for global market competition and financial systemic risk. The study does not claim anything else. An interesting open question at this point is: is this an emergent feature or the result of a direct organizing principle of the involved corporations? So, no need to discuss control up to here.

What about this worrisome control? Issues like "but control cannot really be assessed from ownership", "comparing ownership, and hence control, in different countries is like comparing apples with oranges", "funds don't exert control" and "are the results really new" are extensively discussed in Chapter 6.2 here:
http://e-collection.library.ethz.ch/eserv/eth:2007/eth-2007-02.pdf
next to also being mentioned at the end of the study and in the Supporting Information Section 8.1. In a nutshell, yes it is problematic, but perhaps not too much so. The results for the control are at an aggregated level, and it is very encouraging that fundamentally different models used to infer control from ownership (from the scholarly literature) give very similar results.

In any case, indirect ownership (yes, the network;-) and the corresponding control can be quite dramatic. Recall the pyramidal group of indirect ownership around Marco Tronchetti Provera, allowing him to control Telecom Italia, one of the world's largest telecom companies, with a disproportional small amount of equity. Indeed, as the study mentions, control can effectively be covert:
"For example, a mutual fund owning some percent of a large corporation may try to impose job cuts because of a weak economic situation. This can happen: (i) without voting and (ii) although the fund does not plan to keep these shares for many years. In this case, the influence of the mutual fund has a direct impact on the company and its employees. Furthermore, mutual funds with shares in many corporations may try to pursue similar strategies across their entire portfolio."

The study also mentions, that "control" should be taken with a grain of salt:
"In this sense, our notion of control can be related to Max Weber’s definition of 'power', i.e. the probability of an individual to be able to impose its will despite the opposition of the others."
It is up for question, if economic agents, having large potential control, would in fact choose not to exert it.

I had to smile when I read the sentence: "But the authors didn't bother thinking about how [some issue] complicate their tidy story of big powerful institutions controlling the world." Of course, it's not about empirical data being analyzed for the first time. No, it's all about a story to be sold to the uncritical media and the conspiracy theory camp;-) Also, one of the tags of the post, the label "dubious statistics", is funny. Especially, as the network analysis of the study has nothing to do with the methods of classical statistics (it's a lot about linear algebra and algorithms).

I also like the appeal to authority. Physicists, by default, cannot understand anything about economics. It's a law of nature;-) By the way, the lead author, Ms. Vitali, is in fact an economist and Mr. Battiston published with an economics Nobel Laureate. And the two studies, so hated by Ms. Smith, also had real economists comment kind of favorably:

Bruce Kogut: http://j-node.homeip.net/wfb/jbg%20archive/media/sciencenews.pdf
"The results nicely show how structure emerges from an otherwise weak signal, revealing the ownership backbone within and across countries."

Matthew Jackson: http://www.livescience.com/9704-world-stocks-controlled-select.html
"It's clear, looking at financial contagion and recent crises, that understanding interrelations between companies and holdings is very important in the future," he said. "Certainly people have some understanding of how large some of these financial institutions in the world are, there's some feeling of how intertwined they are, but there's a big difference between having an impression and actually having ... more explicit numbers to put behind it."

Gerald Davis: http://www.sciencenews.org/view/generic/id/333389/titl /Financial_world_dominated_by_a_few_deep_pockets
Because interpreting and analyzing these kinds of data is difficult, he says, the analysis serves more as "an impression of the moon’s surface you get with a telescope. It’s not a street map."

Although, to be fair, there are the legitimate concerns also mentioned in Ms. Smith's post:

Matthew Jackson: http://www.sciencenews.org/view/generic/id/333389/title/Financial_world_dominated_by_a_few_deep_pockets
Ownership can be difficult to study internationally because holding shares in a mutual fund doesn’t necessarily mean the same thing in the U.S. as it does in communist China. And even within a single country ownership can be hard to tease out, says economist Matthew Jackson of Stanford University. For example, when an individual invests in a mutual fund or even purchases shares through an institution like Merrill Lynch, the firm is often still the official owner of the assets. And even when shareholders do have voting rights, they may not exercise them. "This becomes worrisome if everyone is like me and says I’ll let Vanguard do the voting," says Jackson. "Maybe we should be a little bit worried. I don’t know if we should be."

Finally, the question raised in Ms Smith's post of 2009, where she allegedly debunked the first deplorable ownership network study, namely: "Conspiracy theorists will have to wait until the article described in Inside Science is published to determine whether it delivers on its claims." has been answered. Yes, it was published: Phys. Rev. E 80, 036104 (2009). Oh, but wait, that's a physics journal;-)

---
More here: Ownership Networks and Corporate Control: Mapping Economic Power in a Globalized World



Just for fun, an ownership network...


-----------------

 25th of August 2011


Dear authors and readers of Naked Capitalism,

Thank you for your interest in our study. To cut a long discussion short, please let me take a step back and focus on some general issues.

I understand that you categorically disapprove of the value of economic network analysis. Although I don't think the fierce and hostile tone this sentiment was often expressed in was really appropriate, I am fine with that opinion. Especially as this kind of analysis represents a new paradigm in dealing with complex systems and vast amounts of data. (However, other people have thought that network analysis of economic data has been worth their time, see for instance the chapter:
"The Structure of Financial Networks"  in Springer's "Network Science: Complexity in Nature and Technology".)

The mentioned study is currently in the peer-reviewing process for it to be accepted for publication. This not only means that the journal chooses independent (anonymous) reviewers it thinks fit for the job (yes, the experts who should find all the flaws in a paper), but, crucially, that every claim we make we had to try and back with understanding found in the scholarly literature. For what it's worth, this is the academic game to play if you want to publish studies. But this is also where an objective discourse can begin.

I don't want to be condescending when I allege that what is being expressed here are just opinions. But without the same rigor in argumentation, i.e., pointing to studies backing ones claims with existing knowledge, the whole discussion just becomes an endless string of assertions. Of course, one can always claim that this focus on published studies is one of the things that is wrong with academia and its ivory towers...

The bottom line:

A.) Please, for 1 minute, just forget that the notion of control even exists! This was the main focus of all the critique. So we are now simply talking about ownership networks. This begs the question: why bother?

Some scholars have noted, that ownership networks can be understood to exhibit (the references can be found in http://e-collection.library.ethz.ch/eserv/eth:2007/eth-2007-02.pdf):
"the historical bargains struck by labor, the state and holders of capital regarding who gets to own and control the economic assets." (Kogut and Walker, 2001)

That, their study has implications for individual firms:
"A network of ownership ties represents a unique opportunity to examine how an economy-wide structure of relations affects individual firm diversification events." (Kogut and Walker, 1999)
"The relationship of capital to the firm is also shaped by the structure of interfirm networks, which influences firm behavior through access to critical resources and information." (Aguilera and Jackson, 2003)

Finally, they are relevant for policymakers:
"The analysis of networks of business enterprises has grown to be one of the leading perspectives in the study of business policy, organizational behavior, and public economic policy." (Corrado Zollo, 2006)

Analyzing (national) ownership networks has been done before:
-B. Kogut and G.n Walker, "The Small World of Germany and the Durability of National Networks", American Sociological Review, 2001
-R. Corrado and M. Zollo, "Small worlds evolving: governance reforms, privatizations, and ownership networks in Italy", Industrial and Corporate Change, 2006

With our study, we add more sophistication in the methods and extend the scope. And indeed, we find the novel features I mentioned many times not documented in the existing literature (this is a main criterion for a study to be published in a journal).

B.) So, let's look at control now.

There is an extensive body of literature in economics on the issue of control and ownership, which does, in no way, give such a clear cut picture some people here would like to imply and the whole discussion is a lot more complex (again, the references can be found in http://e-collection.library.ethz.ch/eserv/eth:2007/eth-2007-02.pdf). For instance:
(Brioschi et al., 1989; La Porta et al., 1998, 1999; Claessens and Djankov, 2000; Nenova, 2003; Chapelle and Szafarz, 2005; Chapelle, 2005; The Deminor Group, 2005; Almeida and Wolfenzon, 2006; Almeida et al., 2007). Some of the mentioned authors infer control from ownership using the same models as the study under scrutiny. Others discuss the role of mutual funds:
(Santos and Rumble, 2006; Davis and Kim, 2007; Davis, 2008).

With respect to control, the claim of the study is simple: it is a first approximation of (potential) global control structures. And its distribution turns out to be unprecedentedly skewed (hello conspiracy theory camp;-).

C.) The study only claims to have opened the door to further discussions. "This remarkable finding raises at least two questions that are fundamental to the understanding of the functioning of the global economy": global market competition and global financial systemic risk. To what extent this is really relevant and the control approximation is true, is, of course, up for discussion. But we believe this to be a valuable and fruitful starting point for further research.

So, to be honest, I don't really get the whole fuss that's being made here and the general sense of animosity.

Finally, without wanting to sound patronizing, Ms. Smith should be aware that she is offering her opinions and not "the truth", and hence it is maybe not very pertinent when she writes sentences like "I pre-debunked a report on an earlier version of this paper", "this paper is a garbage-in, garbage out analysis", "you might do some very soft proxies with qualitative data, but the authors are physicists". But then again, in a blog post such utterances help make a gripping read;-)

Anyway, all the best,
j


-----------------

 31st of August 2011

Conclusion

Ms. Yves Smith chose a winning strategy:

i.) write a scathing sensationalist critique of a study, frothing with animosity, sprinkled with ad hominem attacks, keeping the readers happy and entertained;
ii.) stubbornly ignore any subsequent clarifications, rebuttals, explanations and context-giving efforts offered by the authors of the study.

This surely works spectacularly well on the Internet, but, in my opinion, it does raise serious questions about Ms. Smith's sincerity, integrity and professionalism...


-----------------

 3rd of October 2011

Update

New blog post: The network of global corporate control - revisited 

1 comment:

Eric said...

Yves' reaction was unfortunate. Not all of her readers reacted the same. I like your work. I also like her blog and can forgive her reaction. The financial crisis that is hurting so many people was, to a large extent, the result of "quants" building bigger and more complicated structures. She's developed a forgivable distaste for anything quant related.

I understand this is a new area of research and we shouldn't expect the first paper to be a home run and solve all our problems. I think the ideas hold a lot of promise.