Monday, August 6, 2012


My latest Bloomberg column can be read here. The title sounds rather weird, I think, because the research I've written about has nothing to do with Google helping anyone. The hazards of writing and the production process. But ignore that -- I think the subject matter is important, and I wanted to offer some further information on what I wrote about there, and a link to the original research.

The basic topic is a new measure of systemic risk known as DebtRank, which was just introduced a few days ago in this new paper in Nature Scientific Reports. This is the work of physicist Stefano Battiston and a team of other physicists and economists, and takes its inspiration from the famous PageRank algorithm invented by Google founders Sergey Brin and Larry Page. DebtRank could do for the analysis of global financial risks what Google did for web search -- make it really possible to determine which elements in the network matter most.

The essential insight of the PageRank algorithm is that in any network of things that make references to each other -- web pages through hypertext links, or scientific papers through citations, for example -- each element effectively votes for other importance of other elements by linking to them. Hence, the most important web pages, Page and Brin reasoned, should be those drawing links from many other pages, especially from other really important pages. The best scientific papers should be those that get those most references, especially from other profoundly important scientific papers (rather than from papers no one ever reads).

To calculate the PageRank of a web page (or a scientific paper), you have to look at all the web pages that link to it. The page gets a higher PageRank in so far as many other pages link to it, especially if those other pages are important, i.e. are pages that have many other web pages linking to them, especially other important web pages. The definition is obviously circular -- you have to know which pages are important in order to be able to calculate which pages are important. That may seem useless, but this problem is easy to sort out mathematically (it's just some linear algebra).  That's what Google algorithm does, and it can be done with lots of computing power, which makes Google so powerful (this is of course only part of the recipe of the search engine). 

Now, what about finance? The analogy for DebtRank is quite direct -- those institutions that present the greatest risks to the financial system are those that, if they fail, would cause the widest spread of economic distress. Naturally, you would tend to have a high DebtRank if you are linked by loans and other financial ties to other firms with high DebtRank -- the same circularity again.

I won't go into more detail except to say that an algorithm can calculate the DebtRank using real data (a partial network of financial ties between institutions based on public information). This is what Battiston and colleagues have done, and it shows some surprises. At the peak of the financial crisis, for example, DebtRank measures for the largest 20 or so banks show that simple bank size just isn't as important as we've come to think. Institutions such as Barclays, Bank of America, JP Morgan and Royal Bank of Scotland presented more systemic risk than did Citigroup or Deutsche Bank, despite being significantly smaller in total assets. Wells Fargo stands out even more: It presented as much systemic risk as Citigroup, despite having only a quarter of the assets. 

This is really only a proof of principle, as the network used is indeed very incomplete. But it shows quite clearly how the phrase "too big to fail" is slightly misleading. We need to worry about which institutions are too central to fail, or really a mixture of too big and too central, and this is what something like DebtRank lets you get at.

Most importantly, the study shows how any assessment of global financial risks will require much more public knowledge of the links between institutions, most of which are currently not public. For more information, I suggest looking at the web site of the European project FOC (Forecasting Financial Crises), in which this research took place. In particular, you may want to have a look at the widget produced by the researchers that shows some of the key institutions and reflects their systemic risks at the height of the crisis as estimated by DebtRank.



  1. The New Scientist article about DebtRank shows a circular graph populated by dots.
    Have you figured out what the coordinates of the graph are?

    DebtRank analysis sounds like it is based on reading a the detailed daily balance sheet for 22 banks. I am puzzling about what are the next steps for performing the analysis.

    I get this far, Equity=Assets-Liability. A liquidity crisis is when quick Assets are less than current Liability.
    So they read the list of liabilities for one bank and match a sum to the list of other institutions. The item appearing as a liability on the first bank is matched to an item appearing on the list of assets of another bank on the list. Right?
    Then they run a little calculus type delta event on the specific financial instrument linking the two banks. Right? That is how they spot instability.
    The question is, what do you see as how the DebtRank authors actually do the analysis with a specific row of data?

  2. Hi Mark,

    I wrote a short response to your article looking at why banks should welcome transparency. It would be great to get your thoughts:

  3. Surely this is a conflict of interest, why would a bank disclose ip on interconnections and thus its all a Hirsch conjecture?

  4. Excellent content - as you always provide and inspires me to come again and again. ppi compensation reclaim

  5. Debt rank is something I hear for the first time about. The idea is great and I admire Google inventing such great products. I still cannot imagine how it is working but I suppose that this I completely different from the process of finding out your credit score. No credit check loans online also surprise me as a perfect financial solution.

  6. This is a great post. I like this topic.This site has lots of advantage.I found many interesting things from this site. It helps me in many ways.Thanks for posting this again.
    Kredit for house

  7. This is an excellent post I seen thanks to share it. It is really what I wanted to see hope in future you will continue for sharing such a excellent post.
    Jakarta Wedding Photography

  8. I recently found many useful information in your website especially this blog page. Among the lots of comments on your articles. Thanks for sharing.
    free credit report gov

  9. Curly Hair Products for Women: Now we have a wide collection of curly and styling hair products for women. Now women can change their appearance easily as they want.

  10. I have been meaning to write something like this on my website and you have given me an idea.ColdFusion Tutor online

  11. I can't say too much about this idea, yet it's pretty interesting discussion indeed.SEO Writing Service

  12. I want to encourage that you continue your great job, have a good day.C# Project Help C# Homework Help

  13. Another great post, I appreciate all the work you put into this site, helping out others with your fun and creative works.homework help economics

  14. I should say that I hear about debt rank for the first time myself but what I know is that it is good to find out something new no matter what. I am more than just sure that for most of you guys, the following info will be new as well, thus Why Proxima B Can Be Potential Habitable Planet might be very interesting to know for those who are curious by nature.


  15. Yeah, the title do sound weird. But if you say that the research you found has nothing to do with Google helping anyone, that means something. Essay About Earthquakes As Main Topic For Humanity is a good choice for writing an essay. In case you don`t know what to begin with, how to proceed and what conclusion to make, writing service online is a way out.

  16. Shokugeki No Soma Season 3 release is expected in 2017 and here are exclusive updates and plot details about our favorite anime series

  17. Love the idea thank you so much for sharing this the idea. Your ideas inspire me and always prove helpful for me.golden globes channel

    2018: Deadline for nomination ballots to be mailed to all HFPA members from Ernst

  18. 121,544 qualified Republicans that this year. The service includes itself as being
    the initial awards show to acknowledge the work of the first and also union members to ever present awards for picture casts and television ensembles.

    Succeeding awards have been held at the Shrine Exposition Center in
    Sag Awards Live