An Elegant Answer to the Wrong Question (or an Incomplete One)
Rodney Garrett and Peter Zimmerman of the NY Fed have produced a paper studying the effect of clearing on derivatives counterparty risk exposures. It is basically an extension of the Duffle-Zhu paper from a few years ago. It studies more realistic networks and a more diverse set of scenarios than D-Z. It demonstrates that with a variety of network structures, clearing actually reduces netting efficiency and increases counterpart risk exposures. This is especially true with “scale free” or “core-periphery” networks, which are more realistic representations of actual derivatives markets than the all-to-all structure in D-Z. They show that when the system relies on relatively few crucial nodes, as is the case in most dealer structures, clearing reduces netting efficiency. This, as Garrett and Zimmerman note, could explain why clearing has not been adopted voluntarily. It also raises doubts about the advisability of clearing mandates, inasmuch as the alleged benefit of clearing is a reduction in counterparty credit exposures.
A few comments. The results, taken on their own terms, make sense. In particular, networks connecting dealers and customers via derivatives transactions, are endogenous. And although network structures are not necessarily efficient due to network externalities and path dependence, there are forces that lead to minimizing credit exposures. thus, although it would be Panglossian to assert that existing structures minimize these exposures, it should not be surprising that interventions that lead to dramatic alterations to networks increase counterparty risk exposures as existing networks are configured at least in part to reduce these exposures.
More importantly, though, there is the issue of whether counterparty risk exposure in derivatives transactions is the proper metric to evaluate the effect of clearing mandates. As I have noted for years (as has Mark Roe), when participants in derivatives transactions have other liabilities, changes in netting efficiency in derivatives primarily redistribute wealth to or from one group of creditors (derivatives counteparties) from or to other creditors (e.g., unsecured lenders, commercial paper purchasers, deposit insurers). Netting and offset essentially privilege the creditors that can use them, at the expense of others who cannot. So telling me policy A reduces counterparty risk exposures by netting provides me very little information about the systemic effects of the policy. To understand the systemic effects, you need to understand the distributive effects across the full set of creditors impacted by the change in derivatives netting efficiency induced by the policy–and that would be every creditor of derivatives market participants. This paper, like all others in the area, does not do that.
Put another way. Papers in this literature say little about systemic risk because they only analyze a piece of the system. The derivatives-centric approach is of little value in assessing systemic risk. To analyze systemic risk, you need to analyze the system, not a piece of it.
What’s more, shuffling around credit risk is probably not the most important effect of clearing mandates, even though it receives the vast bulk of the attention. As I’ve written repeatedly in the past, including here, clearing reduces credit risk by increasing liquidity risk, most notably, through variation margining which results in the need to obtain cash in a hurry to meet margin calls, which can be large when there are large market shocks. The expansion of clearing to OTC markets which dwarf listed derivatives potentially leads to orders-of-magnitude increases in liquidity needs.
It is these liquidity demands which create huge potential systemic risks. Financial crises are usually liquidity crises: mechanisms such as clearing increase demands for liquidity in stressed market conditions, and do so in a way that increases the rigidity and tightness of the coupling in the financial system. This is extremely dangerous. Tight coupling in particular is associated with system failure in a variety of real world systems including both financial and non-financial systems.
Indeed, all of the attempts to make CCPs invulnerable all tend to exacerbate these problems. This raises the possibility that CCPs could be bastions surviving in the midst of a completely rubbled financial system.
In sum, papers like the Garrett-Zimmerman work are very elegant and technically sophisticated, and help answer a question: Does clearing reduce derivatives counterparty risk exposures? But all the elegance and technical sophistication is likely for naught given that the question is the wrong question, or a very incomplete one.