CCPs & RTGS: Devil Take the Hindmost?
The frantic sewing of parachutes in a plane that is 30,000 feet in the air continues apace. Last week the Office of Financial Research (a Son of Frankendodd) released its 2015 Annual Report. This tome received attention mainly because it raised alarm about potential systemic risks arising from central clearing mandates. An improvement, I guess, but like most official evaluations of the systemic risks of CCPs, it misses the real problems.
It gets off on the wrong foot by misstating the real benefits of CCPs. Its top two benefits are related to the ability of CCPs, and via them regulators, to get more complete, accurate, and timely information on derivatives positions and trading prices. But these can be achieved by transaction reporting alone, without going the full monty to clearing, which also entails collateralization (including both initial and variation margining) and mutualization of default risk. (Trade reporting has turned into a nightmare, which I will write about further soon. But the point is that you don’t need clearing to get the benefits OFR touts.)
But the main problem, yet again, is that OFR focuses on the “single point of failure”/interconnectedness/default loss contagion channel for CCP systemic risk. This is not immaterial, but it is not the main thing. The main thing is that CCPs create potentially massive contingent demands for liquidity, where the liquidity contingency is likely to occur precisely at the worst time–when the system is undergoing a financial crisis.
Further, OFR gets it wrong when it states that CCPs “reduce the risk of counterparty default.” CCPs redistribute the risk of the insolvency or illiquidity of a large financial institution away from its derivatives counterparties towards its other creditors. It protects one group of creditors at the expense of others.
It is very much open to question whether this reallocation is systemically stabilizing, or is instead a means whereby one relatively concentrated group of market participants can advantage themselves at the expense of others.
Reading Izabella Kaminska’s excellent FT Alphaville post on Real Time Gross Settlement (RTGS) mechanisms makes plain that this phenomenon of substituting liquidity risk for credit risk, and redistributing credit risk away from core banks, is not limited to derivatives clearing. RTGS replaced deferred net settlement (DNS) because of banks’ and central banks’ concern that in the latter, interbank credit balances could accumulate, resulting in a default loss to settlement banks in the event that an net payer bank failed before the next netting cycle. RTGS eliminates interbank credit exposure.
But, of course, this doesn’t make credit exposure go away. It redistributes it to settlement banks’ other creditors. To a first approximation, the total losses from the inability of a bank to meet its obligations are the same under RTGS and DNS. The difference is who gets a chair when the music stops. Settlement banks–and crucially, the central banks–like RTGS because they almost always are going to get a chair.
Furthermore, as even its proponents acknowledge, RTGS is much more liquidity intensive. To be able to make every payment in real time, a settlement bank either has to have the cash on hand, or the ability to borrow it on demand intraday from the central bank. Liquidity needs scale with gross payments, which are substantially larger than net payments. Thus, like CCPs, RTGS substitutes liquidity risk for default risk.
This risk is exacerbated by the fact that a prisoner’s dilemma problem exists in RTGS. Participants concerned about the creditworthiness of other banks have an incentive to delay payments and hoard liquidity, since once a payment goes into the system, it is final and the payer is at risk to loss of the entire gross amount if a bank that owes it fails before it pays.
Since holding cash in sufficient amounts to meet all payment obligations is extremely expensive, RTGS has evolved to permit central banks to lend intraday on a collateralized basis. But as was seen in the 2008 crisis, collateralization poses its own risks, including ballooning haircuts that can set off price spirals due to collateral fire sales. Further, due to the potential for the breakdown of large collateral chains, this creates an interconnection risk, and represents a further coupling of the system. And it is coupling, remember, that is at the root of most catastrophic accidents. Secured lending can create a false sense of security.
Izabella’s post also points out another problem with RTGS, which is common to central clearing. It creates a much more tightly coupled system that is very vulnerable to operational risk. This risk crystalized in October, 2014, when a seemingly innocuous change to the system (deleting a member bank) caused the failure of the UK’s CHAPS settlement system for a day. Ironically, this was the result of an interaction between one part of the system, and another part (the Liquidity Savings Mechanism) that was intended to economize on the liquidity demands of RTGS, and essentially created an RTGS-DNS hybrid. As in most “normal accidents”, unexpected interactions between seemingly unrelated parts of a complex system led to its failure.
There is another way to see all of this. Both central clearing and RTGS are intended to create “no credit” systems. That is true only in a very limited sense–a profoundly unsystemic sense. Yes, CCPs and RTGS are designed so that participants in those arrangements don’t have credit exposure to one another. But those participants aren’t the entire system, just a part of it: the exposure is pushed away from them to others. Further, the method for reducing credit exposure among the participants is to require extensive reliance on liquidity mechanisms that are prone to breakdown in stressed market conditions. Further, these liquidity mechanisms are based on credit: banks (or the CCP) borrow from other banks, or from central banks.
Not to sound like a broken record, but things like CCPs and RTGS redistribute and transform risks, rather than eliminate them altogether. Unfortunately, these transformations do not necessarily reduce the risk of a systemic crisis, and arguably increase it in some cases. The failure of officialdom, and large swathes of the banking sector, to recognize or address this reflects in large part a failure to take a systemic perspective. Perhaps cynically, this can be explained by the fact that the central banks and banks that drive these reform efforts mistake their own interests for the interest of the system as a whole: le système, c’est nous. As a result, “Devil take the hindmost” could well be applied as the motto of RTGS and central clearing.
This illustrates a broader problem in public policy. Government is too often invoked as a deus ex machina that internalizes externalities. But the fact that most regulatory change efforts are driven by, or ultimately controlled by, a small subset of interested parties who have the most concentrated stake in an issue. Given the diffuseness of other impacted parties this is inevitable. But it means that in practical terms internalizing externalities via regulation of something as complex as the financial system is a chimerical goal. The externality hot potato gets tossed from one segment of the financial sector to another. Government regulation, as opposed to self-regulatory initiatives, mainly affect the makeup of the subset of participants who are involved in influencing the process, and the distribution of the bargaining power. This works through the entire process, from the crafting of legislation, to the writing of regulations, to their implementation. This is why we get things like RTGS or CCP mandates, which make a certain set of participants better off, but which it is heroic indeed to believe are truly welfare increasing.