Streetwise Professor

November 21, 2012

It’s Contagion, Stupid-Not Interconnectedness.

It is hard to overstate the importance of this report by Hal Scott and the Committee on Capital Markets Regulation, titled “Interconnectedness and Contagion.”  It evaluates the issue of systemic risk through the lens of the Lehman Brothers failure (and to a lesser degree, the AIG failure).

One way to summarize the report is “everything you knew about Lehman was wrong.”  Certainly, everything Gensler and Timmy! and Frank and Dodd and legislators and regulators around the world knew was wrong.

That conventional narrative is that Lehman’s failure brought the world financial system to its knees because it was so interconnected to the rest of the financial system.  That losses on exposures to Lehman-an most notably through derivatives exposures-threatened to propagate through an interconnected set of large financial institutions.   That the crisis was the result of a “daisy chain” or “domino” effect via what Scott refers to as “asset interconnectedness” (e.g., derivatives exposures) or “liability interconnectedness” (e.g., a big supplier of credit/liquidity fails, thereby forcing those dependent on it for credit or liquidity to fail).

Scott thoroughly debunks the interconnectedness narrative.  Instead, he shows persuasively that a contagion caused the crisis.  The Lehman shock dramatically shifted beliefs about whether large financial institutions would be bailed out, and also conveyed information about the severity of losses on real estate assets that were held not just by Lehman, but by a wide array of financial institutions.  This raised doubts about the solvency of all financial institutions, resulting in a run on their liabilities and a refusal to roll over these liabilities.  Given the dependence on short term credit, this run caused a major crisis.

In other words, it’s all about the funding mechanism.  It’s about liquidity.   It’s about runs.

Scott notes that the configuration of the financial network does affect its vulnerability to contagion, but that attempting to mitigate systemic risk through policies intended to affect the degree of interconnection is unlikely to reduce substantially the risk arising from contagion.

This is incredibly important because Frankendodd and EMIR and all of it is largely predicated on the belief that interconnection is the source of systemic risk.  But the diagnosis is wrong, meaning that the prescription is almost certainly wrong too.

Scott looks at CCP mandates in particular, and gives them 1.5 cheers.  I think that’s one or 1.5 too many, but it is gratifying to see that he does forthrightly downplay the potential benefits of CCPs as a means of preventing future crises.

The reason that I think he is not negative enough on this point is that he only looks at the first order impacts of CCPs, through their effect on derivatives netting, for instance.  He does devote some attention to the effects of CCP mandates on funding and liquidity, but more focused attention on the indirect effects of clearing and collateral mandates would, in my view, raise more serious concerns, especially in light of the primacy of funding/liquidity contagion channel in creating systemic risk.

Scott, citing Duffie’s work, claims that CCPs may reduce the incentive to run on big financial institutions.   I can see that in some scenarios, but I can also envision others in which a rigorous, highly-time sensitive, no-credit system like that will result from CCP and collateral mandates can lead to runs, either on the CCP, or on firms connected to the CCP.  That’s my take on what happened on Black Monday, 1987.

Clearing and collateral mandates will lead to a whole series of changes to the demand for liquidity, and the mechanisms for supplying it.  A cleared system is more tightly coupled; increases the likelihood in big shocks to liquidity demand; potentially ties up liquidity (especially under rigorous segregation regimes); and on and on.  Given the nature of cleared systems (and the imposition of rigorous daily variation margin based on mark-to-market for non-cleared derivatives), market users will make arrangements to secure contingent liquidity.  As a result, the entire financial network topology will change.  How it will change, and how these changes will affect systemic risk, are impossible to divine at present.  But it is certain that these changes will be profound and it is not difficult to imagine scenarios in which the new topology is as fragile in the face of large shocks as the old one.  Indeed, it’s not hard to imagine scenarios in which the new topology is more fragile than the old.

Clearing and collateral mandates should also be viewed as regulations of capital structures.  (“No credit extended through derivatives transactions” and “derivatives are at the top of the priority queue.”)  Capital structures will respond to this regulation, again in unpredictable ways, and in ways that affect the vulnerability of the system to contagion.

I am giving a talk on Monday at a joint Bundesbank-University of Frankfort conference about central banking.  My talk is about systemic risk and CCPs, and I will explore all of these themes.  The overarching theme is that you cannot view an intervention into the structure of financial markets, like a CCP mandate, in isolation.  You have to consider the endogenous responses throughout the system. The entire structure of financial contracts will change as the result of such a big intervention, and this structure will have its own special vulnerabilities.

When evaluating these vulnerabilities, one should pay special attention to liquidity issues, and liquidity contagion.  This was a theme of my JACF article “Clearing and Collateral Mandates? The New Liquidity Trap?”  Scott’s report adds special force to this contention that liquidity and contagion, rather than interconnection, should be the focus of any regulatory and legislative efforts.

It also suggests that much, and arguably virtually all, of the regulatory response to the 2008 Crisis was misdirected, and hence will either not reduce systemic risk, and may increase it.   Given the costs that these regulations impose, this is a sobering conclusion.

It finally suggests that it all comes down to central banks, as lenders of last resort.  If it is a liquidity problem (and in this Scott is closely aligned with Gary Gorton), the institutional means for supplying liquidity in times of crisis-which now means central banks-is of paramount importance.

Print Friendly, PDF & Email


  1. Good stuff, Prof. Here’s a good discussion re the repo market, and the issues surrounding rehypothecation, especially re liquidity constraints:

    The authors note: “The financial strains experienced by several dealers, including Bear Stearns and Lehman Brothers, during the financial
    crisis of 2007-09 highlighted the fact that the two tri-party clearing banks are not only agents, but also the largest creditors in the tri-party repo market on each business day.” (p. 6 -7)

    Comment by markets.aurelius — November 21, 2012 @ 10:46 am

  2. Thanks, @markets, both for the props and the link. Triparty repo is a huge deal, and could well become more important to the extent that clearing and collateral mandates drive even more utilization of secured lending.

    Note that the two settlement banks-JPM and BONY-are looking to expand their collateral management services and secured lending businesses (through collateral transformation, for instance) in order to accommodate increases in clearing. So here is another channel, another set of linkages. Given the extreme systemic importance of these settlement banks and the triparty repo market, these linkages are particularly problematic.

    The ProfessorComment by The Professor — November 21, 2012 @ 11:31 am

  3. Nice summary. Corporate capital structures will undoubtedly evolve as non-financial corporates continue to understand increasingly precious capital – which traditional funding mechanisms encumbered little of- can be used better within the funding process than their current advisors suggest. Cash funded risk retention mechanisms involving non-bank suppliers of contingent liquidity are already being built in many corporates. They utilise regulated insurance vehicles like captive insurance companies/ ICC’s to assume credit risk for example, but are outside the skill-set of traditional credit intermediaries. Such mechanisms offer treasurers callable contingent liquidity held on their own balance sheet. Such liquidity pools help reduce cash flow volatility that external funding can no longer be guaranteed to “smooth”. This will be important as external auditors increasingly demand greater clarity regarding going concern assessment processes that claim to mitigate the diminishing availability of externally sourced liquidity and, potentially, a growing dependence on CCP’s. Cash funding contingent liabilities whilst transferring the timing risk can also protect corporates better from the contagion risk Hal Scott rightly highlights. It isn’t too strong to suggest the contagion was a man-made virus created by credit intermediaries and policymakers who built a “highly capital efficient” just-in-time liquidity model.

    Comment by creditplumber — November 21, 2012 @ 12:51 pm

  4. @creditplumber. Fascinating. Have any links/reading suggestions regarding the regulated insurance vehicles?

    One of the concerns about clearing is it puts tremendous stress on just-in-time liquidity models. There is no margin for error, no pun intended.

    The ProfessorComment by The Professor — November 21, 2012 @ 2:40 pm

  5. You’re right. There is no margin for error in JIT liquidity models. So the answer is look beyond the traditional opaque models which discouraged risk sharing. For years, we’ve all been comfortable with banks telling corporates the equivalent of “i can insure your $1m home on the Florida cost against all perils for $500. You only need to carry a $50 excess. By the way, here’s a $100,000 overdraft, just in case…. ” Those days have gone.

    Very little on what I wrote about in the public domain, I’m afraid. FYI, my background is AIG so straddle both reg & un-reg “insurance/finance”

    I’ve attached one link above that I believe offers a structural taster of a funding structure that cash funds a contingent liability. The UK govt uses the insurance/PCC structure for its scheme to engender much needed new housebuilding in the UK. However, house builders weren’t keen given house price volatility. So the govt essentially said “we’ll guarantee a minimum sale value to housebuilders” . They did this by cash funding the insurance vehicle to cover potential default events. This allows the House builders to build new homes confident of a floor to selling prices they can achieve

    On the corporate side, the use of captives to assume the corporates operational credit risk is one way of enabling counterparty risk to be adjusted depending on external market conditions. This might mean retaining more risk when risk premia are high and transferring more risk when risk premia are low. Until now, it’s a “structure” that corporate treasurers had no need to adopt because of the technical risk tracking challenges (now overcome with technology) and the abundance of external liquidity. Times have changed. Hurricane Sandy has also highlighted the challenge for many corporates .They probably retained significant unfunded contingent liabilities such as contingent business interruption, wind-driven water, denial of access losses, etc. Historically, they had sufficient external sources of liquidity to pay for these crystallizing exposures. They could appropriate future profits confident these shortfalls could be made good by one of their sources of external liquidity. This is no longer a certainty. An answer? Use technology and insce vehicles to cash fund contingent liabilities out of future cash-flow, not current cash, giving current value to future cash flow.

    An overlooked contingent liability is bad debt reserves that are rarely funded to cash. By cash funding such risks – and having an insurer take the timing risk so a $10m fund to be built over 3 years is available to call on Day One – you can see you begin to reduce exposure to external contagion, reduce liquidity risk and make the going concern assessment process a little more transparent if you can introduce daily observable event data into the funding feedback loop.
    In respect of the trade receivable asset alone, the captive may purchase credit insurance legally drafted to comply with banks eligibility criteria for Credit Risk Mitigants. The insurers PD can substitute for that of the corporate, and a bank can provide liquidity secured on individual assets backed by granular and daily observable event data. The banks RWA reduce – reducing the margin they need to charge – but with a captive taking a 5-10% retention/Equity tranche, advance rates banks are willing to extend increase from say 65-70% of a total portfolio, (even less for exporters) to 90-95%. More working capital liquidity.

    Point is there are plenty of ways of integrating risk-based solutions into corporate funding structures that can reduce corporate liquidity risk, counterparty risk, reputation risk & contagion risk. But all you hear from those who don’t understand insurance regulation is “there are no solutions.” It’s understandable. They face career risk as I did. However, there are solutions. However, the insurance industry is ill equipped to sell funding solutions given they competed historically with more sophisticated banking advisors. However, times have changed.

    Hope this helps develop the picture a little.

    Comment by creditplumber — November 21, 2012 @ 3:52 pm

  6. […] Streetwise Professor » It’s Contagion, Stupid-Not Interconnectedness.. Share this:EmailPrintTwitterLinkedInMoreGoogle +1FacebookDiggLike this:LikeBe the first to like this. […]

    Pingback by Streetwise Professor » It’s Contagion, Stupid-Not Interconnectedness. « The OTC Space — November 22, 2012 @ 7:55 am

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress