Streetwise Professor

Linking to a speech by Benoit Coeuré of the ECB (who gave the keynote at the Paris conference I spoke at a month ago), ISDA’s Derivativiews blog points out how the reporting of derivatives transactions has been totally botched due to jurisdictional fragmentation:

Mr. Coeuré asks an important question about transparency in his speech: “Does…the supervisor responsible for the supervision of a large cross-border financial institution at a consolidated level have direct and immediate access to information on OTC derivatives transaction that encompass all transactions entered into by all entities of this group? Is the information accessible, in other words can it be easily aggregated across trade repositories and jurisdictions?”

He then goes on to say: “My answer would be a clear no!”

And we agree.

Mr. Coeuré went on to discuss privacy laws, blocking statutes and indemnification clauses in several jurisdictions which restrict access to the detail of OTC derivatives transactions; the inability to aggregate data across trade repositories and jurisdictions; and differences in the type and level of information required for reports across jurisdictions. He ultimately broke the problem into three main issues: information gaps, data fragmentation across trade repositories and jurisdictions and obstacles impeding authorities’ access to data.

ISDA notes that this mess was predictable, and predicted.  I agree: I was one of those who predicted this years back, based in part on my experience with attempting to create an energy data hub in 2003-2004.  Both commercial and political pressures have led to the proliferation of repositories.  Market participants and regulators have beavered away erecting a derivatives Tower of Babel, a collection of databases that serve as a barrier to the sharing of information, rather than a means of sharing information.

I would further note that there are other, deeper problems that call the entire exercise into question.  It may do more harm than good.

First, OTC derivatives represent only a subset of exposures, and therefore even if one had a single, unified database of all derivatives trades one would still have an incomplete picture of interconnections in the system as a whole.

Second, and more fundamentally,  even if regulators were to know the totality of interconnections with only a slight degree of imprecision, what could they do with that information?   The financial system is complex, and it is impossible to know how the system will react to the failure of one node (or several nodes) in that system, or even the threatened failure of some nodes.  Indeed, the actions of market participants depend on their beliefs; the stability of the system depends on how market participants behave; and the beliefs of these market participants is unknowable.  Meaning that even an accurate map of all interconnections is not sufficient to understand the susceptibility of the system to crisis: one has to know about the expectations and beliefs of market participants.

Moreover, if one views a crisis as a period when the financial system goes chaotic (in the technical sense of the term), even the slightest imprecision in measuring the state of the system (where the state includes actual interconnections/exposures and beliefs about these interconnections and the beliefs of others and beliefs about how people will act on those beliefs and on and on) means that one cannot predict how the system will behave: the defining characteristic of a chaotic system is extreme sensitivity to initial conditions, so if you measure those conditions with even the slightest error, you will not be able to predict its evolution, or its response to a shock.  Moreover, it means that one cannot predict the effect of interventions in the system when (a) the system is chaotic or on the cusp of chaos, and (b) one cannot measure the state of the system with near perfect precision.

In other words, even if the Tower of Babel problem is corrected, and one measures derivatives exposures with considerable accuracy, one can have little confidence that regulators will have the information necessary to identify an incipient systemic event, or how their interventions will affect the system.  Indeed, in crisis situations, a little knowledge can be a dangerous thing: interventions in a chaotic system based on an imprecise measurement of the state of the system, let alone an incomplete understanding of the dynamics of the system, can make things worse, not better.

This suggests that the entire enterprise of attempting to map the system is futile, and perhaps even dangerous.   This enterprise is predicated on a mechanical view, a view that characterizes the financial system as stable and predictable if you have enough information.  If instead the system is complex and on the edge of chaos, this view is completely misguided because no amount of information is enough.  Moreover, this view encourages hubris and can result in interventions that destabilize instead of stabilize.

This is the knowledge problem on krokodil.  Centralized intervention in a complex system based on imperfect knowledge is a very dangerous thing indeed.