Monday, November 7, 2011

ISDA: Stop Making Sense

The following is a response I just posted on Bloomberg to some criticism last week coming from the International Swaps and Derivatives Association. They took issue with some things I had written in my latest Bloomberg column. I think their comment was partially fair, and also partially misleading, so I thought some clarification would be useful. The text below is identical to what appears (or will very shortly) in Bloomberg:

*********************************

My most recent Bloomberg column on the network of credit default swaps contracts provoked a comment from the International Swaps and Derivatives Association, Inc. The group objected to my characterization of the network of outstanding CDS contracts as "hidden" and potentially a source of trouble. I'd like to address their concerns, and also raise some questions.

Contrary to the association's claim, I am aware of the existence of the Depository Trust & Clearing Corporation. I'll admit to having underestimated how much their project to create a warehouse of information on CDS contracts has developed in the past few years; my statement that these contracts are not "recorded by any central repository" was too strong, as a partial repository does exist, and the DTCC deserves great credit for creating it.

However, it is not clear that this repository gives such a complete picture of outstanding CDS linkages that we can all relax.

For example, DTCC's repository covers 98 percent of all outstanding CDS contracts, not 100 percent. Asking why may or may not be a quibble. After all, a map showing 98 percent of the largest 300 cities in the U.S. could leave out New York, Los Angeles, Chicago, Houston and Philadelphia. Moreover, the simple number of contracts tells us nothing about the values listed on those contracts. In principle, the missing 2 percent of contracts could represent a significant fraction of the outstanding value of CDS contracts.

More importantly, when thinking about potentially cascading risks in a complex network, fine details of the network topology -- its architecture or wiring diagram -- matter a lot. Indeed, the CDS contracts that put American Insurance Group Inc. in grave danger in 2008 represented a tiny fraction -- much less than 1 percent -- of the total number of outstanding CDS contracts.

Hence, it would be interesting to know why the repository holds only 98 percent rather than 100 percent. There may be a very simple and reassuring answer, but it's not readily apparent from DTCC's description of the repository.

Also, there is another issue which makes "fully transparent" not quite the right phrase for this network of contracts, even if we suppose the 98 percent leaves out nothing of importance.

The DTCC commendably makes its data available to regulators. Still, it appears that the full network of interdependencies created by CDS contracts may remain opaque to regulators, because DTCC, according to its own description, enables...
"... each regulator to access reports tailored to their specific entitlements as a market regulator, prudential or primary supervisor, or central bank. These detailed reports are created for each regulator to show only the CDS data relevant to its jurisdiction, regulated entities or currency, at the appropriate level of aggregation."
This would imply, for example, that regulators in the U.S. can look and see which of their banks have sold CDS on, say, a big German bank. But the health of the U.S. banks then depends directly on the health of that German bank, which may in turn have sold CDS on Greek or Italian debt or any number of other things. The DTCC data on the latter CDS contracts would, apparently, not be available to U.S. regulators, being out of their jurisdiction.

The point is that a financial institution is at risk not only from contracts it has entered into, but also from contracts that its many counterparties have entered into (this is the whole idea of systemic risk linked to the possibility of contagion). Credible tests of the financial network's resilience require a truly global analysis of the potential pathways along which distress (particularly from outright counterparty failures) may spread. It's not clear that any regulator has the full data on which such an analysis can be based.

None of this, by any means, is meant as a criticism of DTCC or what it has done in the past few years. The 98 percent figure is impressive, and let's hope the 98 percent soon becomes 100 percent and the DTCC finds a way to make ALL information in the repository available to regulators everywhere. Even better would be full disclosure to the public.

Of course, nothing in the comment from the International Swaps and Derivatives Association changes the main point of my column, which was that it is incorrect to believe that more CDS contracts -- or, more generally, more financial interdependencies of any kind, including links created by other derivatives such as interest-rate swaps -- automatically lead to better risk-sharing and a safer banking system. More apparent risk-sharing can actually mean more systemic risk and less overall banking safety.

(Mark Buchanan is a Bloomberg View columnist.)

2 comments:

  1. Mark,

    Pardon me if these questions were answered earlier in this saga, I hsve been traveling and had little access to things for awhile.

    My understanding of the process to establish a CDS would be:

    a) establish an ISDA contract with the counterparty that covers CDS activity.

    b) execute CDS contracts under this ISDA, which are confirmed according to the ISDA

    c) hold contracts until expiry, unwinding, trade, or settlement.

    Is it a or b that is held at the DTCC?

    If it is a, that is not very meaningful data. I remember a company that only had 10% of its total ISDA's active, there were no trades under the other 90%.

    Second, if it is b, there is a lot of sophisticated technology, costing tens to hundreds of millions of dollars, required to stay on top of current and future potential risks related to these activities. Even if it is b, is there anything sophisticated enough to mine the DTCC's database in a manner to make informed current assessments?

    Thanks.

    ReplyDelete
  2. Just as Mark points out, there are no guarantees, even at 98%, that there will be any sort of predictability given the unknown network topology. The dynamics of the network can't be understood without that, and even that information might not be enough.

    These are non-linear, discrete, iterated dynamical systems. They may very well be chaotic. If so, they are unpredictable with anything less than perfect knowledge: of all the transactions; their size; the exact transaction time... and so on. As I said, a complete understanding. That's not possible in this universe.

    So the danger of a transition to a very different set of orbital trajectories will always remain a real possibility. This seems to has happened over and over again since the country began, with periodic catastrophes causing market crashes. But now it's happening at a much faster pace, thanks to automation.

    Attempts to regulate these markets will be doomed to failure if this isn't brought under control. That's because the advent of unlimited high-speed trading presents a very real barrier to any sort of long term stability. As currently construed, the market system will keep crashing and burning without circuit breakers - and fast ones. It appears that the SEC is finally prepared to taken this problem on. It's none too soon, in my opinion.

    ReplyDelete