6. Collateral and the OTC Derivatives Market

Manmohan Singh
Published Date:
June 2014
  • ShareShare
Show Summary Details

This chapter provides an overview of the over-the-counter (OTC) derivatives market and the associated drawbacks in the regulatory initiatives that propose to move these contracts to central counterparties (CCPs). As cross-border issues continue to unfold, this chapter proposes an alternative that would make this market safe with relatively little by way of additional collateral costs. Furthermore, it is argued that there may not be need for creating more too-big-to-fail institutions.


As part of the extensive regulatory reform proposals, the new rules will warrant a significant increase in the use of collateral across the financial system. Estimates by markets and research and policy institutions suggest that the Dodd–Frank Act, Basel III and European Market Infrastructure Regulation (EMIR) may warrant US$2–$4 trillion in additional unencumbered collateral that will span margins for OTC derivatives at CCPs; liquidity ratio(s) under Basel III; and related needs stemming from parallel developments under the EMIR and Solvency II. At the same time, due to the global financial crisis and efforts at quantitative easing (QE) in the US and Europe, significant amounts of collateral have been drained out of the financial system and siloed at central banks. Furthermore, due to counterparty risk in dealing with large banks and the risk-aversion of clients, collateral reuse (or velocity) has also been decreasing rapidly. In fact, the bilateral pledged market, which offers a genuine market clearing price for collateral has shrunk from about US$10 trillion to about US$6 trillion in recent years. More importantly, many of the proposed regulations (eg, moving OTC derivatives to CCPs) have not yet been in force and some key start dates postponed on several fronts.

The financial crisis following Lehman Brothers’ demise and the American International Group’s (AIG) bailout provided the impetus to move the lightly regulated OTC derivative contracts from bilateral clearing to CCPs. The debate about the future of financial regulation has heated up as regulators in both the US and the European Union (EU) seek legislative approval to mitigate systemic risk associated with systemically important financial institutions (SIFIs), which include large banks and nonbanks. In order to mitigate systemic risk that is due to counterparty credit risks and failures, either the users of derivative contracts will have to hold more collateral (or equivalent capital) from bilateral counterparties, or margin will have to be posted to CCPs. Studies (Singh, 2010) have shown that this US$600 trillion OTC derivatives market is seriously undercollateralised and thus contributes to systemic risk (Panel 6.1 and Table 6.1). Also, recent work has shown that the associated demand for additional collateral to satisfy the envisaged regulatory efforts will be onerous (ISDA 2012; Bank of England 2012; Bank for International Settlements 2011).

Table 6.1Under-collateralization in the OTC derivatives market
Gross market value
H2 2008H1 2009H2 2009H1 2010H2 2010H1 2011H2 2011H1 2012H2 2012
GRAND TOTAL35,28125,31421,54224,67321,29619,51827,28525,39224,740
A. Foreign exchange contracts4,0842,4702,0702,5242,4822,3362,5552,2172,304
B. Interest rate contracts20,08715,47814,02017,53314,74613,24420,00119,11318,833
C. Equity-linked contracts1,112879708706648708679645605
D. Commodity contracts955682545457526471487390358
E. Credit default swaps5,1162,9871,8011,6661,3511,3451,5861,187848
F. Unallocated3,9272,8172,3981,7881,5431,4141,9771,8401,792
GROSS CREDIT EXPOSURE*5,0053,7443,5213,5783,4802,9713,9123,6683,626

Gross market values have been calculated as the sum of the total gross positive market value of contracts and the absolute value of the gross negative market value of contracts with non-reporting counterparties. Gross credit exposure is after taking into account legally enforceable bilateral netting agreements.

Gross market values have been calculated as the sum of the total gross positive market value of contracts and the absolute value of the gross negative market value of contracts with non-reporting counterparties. Gross credit exposure is after taking into account legally enforceable bilateral netting agreements.

Panel 6.1:Undercollateralisation in the OTC Derivatives Market

While a much-cited figure, the notional value of contracts of about US$600 trillion overstates the importance of this market. More relevant are the “in-the-money” (or gross-positive-value) and “out-of-the money” (or gross-negative-value) derivative positions, which are further reduced by “netting” of related positions. From a collateral demand/supply framework, undercollateralisation is the more relevant metric for policy discussions. While, typically, collateral – both initial and variation-margin – is posted by hedge funds, asset managers and other clients, large banks active in this space do not have a two-way margin agreement with some clients (eg, sovereigns, quasi-sovereigns, large pensions and insurers, and AAA corporations), so collateral may not be forthcoming when due and, as a quid pro quo, the banks may not be posting collateral, either, to such clients. Interestingly, regulatory proposals also exempt foreign-exchange swaps from central clearing. A key incentive for moving OTC derivatives to CCPs is higher multilateral netting, ie, offsetting exposures across all

OTC products on SIFIs’ books – intuitively, the margin required to cover the exposure of the portfolio would be smaller in a CCP world. However, if there are multiple CCPs that are not linked, the benefits of netting are significantly reduced, because across-product netting will not take place (since almost all CCPs presently offer multilateral netting in the same asset class and not across products).

At present, there is undercollateralisation within the OTC derivatives space that stems from several privileged investors within the financial system – sovereigns, sovereign wealth funds, central banks, corporate, multilateral institutions, etc. There has been reported undercollateralisation in recent years of about US$3–5 trillion, according to the Bank for International Settlements (BIS) semi-annual surveys (see Table 6.1) and Singh (2010). However, these figures may not pick up the full extent of collateral shortfall since they may not include initial margin that is not posted; this is because only variation margin is typically captured in financial statements. Even if we consider half of the total positions (ie, when SIFIs are out-of-the-money) that are risks to taxpayers, these estimates are sizable. Furthermore, although BIS and International Swaps and Derivatives Association (ISDA) sources indicate about US$1.9 trillion of collateral dedicated to this market, this collateral is fungible and includes a reuse factor of about 2.5 to 3.0. So dedicated collateral may be only US$700–900 billion. Most recent estimates of a reuse factor (or collateral velocity) are around 2.0, so significant additional collateral will be required to be posted (see Chapter 3).

Moving (Some) OTC Derivatives to CCPs

By way of background, prior to the momentum to move OTC derivatives from SIFIs’ books, CCPs were viewed under the rubric of payment systems. In the aftermath of the Lehman crisis, the G20 Pittsburgh meetings in 2009 decided that a critical mass of SIFIs’ derivative-related risks would be moved to CCPs. Regulators are forcing, en masse, considerable OTC derivatives to CCPs. This is a huge transition, primarily to move this risk outside the banking system. These new entities may also be viewed as “derivative warehouses”, or concentrated “risk nodes” of global financial markets. There are many proposals on trying to unwind SIFIs; it is a difficult (if not an impossible) task. So creating new SIFIs such as CCPs should be backed by sound economics. Figure 6.1 illustrates that, on average, each of the top 10 SIFIs carries about US$100 billion of derivative-related tail risk. This is the cost to the financial system from the failure of a SIFI (where tail risk is measured by residual derivative liabilities at a SIFI, after netting and collateral); the underlying economics holds under both International Financial Reporting Standards (IFRS) and the Generally Accepted Accounting Principles (GAAP). For example, Barclays’ annual report indicates derivative liabilities of US$527 billion; after netting of US$427, the risk is US$100 billion of contracts if Barclays fails. This is reflected in GAAP accounting but not under IFRS (as it does not allow netting). However, the economics of US$100 billion residual risk to the market if Barclays fails holds true even under IFRS accounting.

Figure 6.1.How SIFIs offload most of their OTC derivative book to CCPs

Yet, instead of addressing the derivatives tail risk, the present regulatory agenda is focused on offloading most of the derivatives book to CCPs. Past and present market practices result in residual risk in the form of derivative liabilities (and derivative assets), based on International Swap and Derivatives Association’s (ISDA) netting agreements, because:

  • sovereigns, AAA insurers, corporates, large banks, multilateral institutions (eg, EBRD), and the Berkshire Hathaway types of firms do not post adequate collateral, since they are viewed by banks and regulators as privileged and (presumably) safe clients; and

  • SIFIs (ie, banks and dealers) typically post no initial margin/default funds to each other for these contracts.

It was envisaged that CCPs will require collateral to be posted from all members and thus offer a transparent ground for the regulatory overhaul. In essence, all parties should post collateral to CCPs; no exceptions or exemptions. This is also called two-way CSAs (credit support annexes) under ISDA. However, this is not happening as envisaged. As stated above, there will be exemptions to some end-users, and many central banks, sovereigns and municipalities are not required to post collateral.

Not surprisingly, the regulatory efforts(s) are meeting resistance from the financial industry, including the large banks, asset managers (such as pension funds) and insurers. Another market that has lobbied to avoid posting collateral is the “end-users” such as airlines and non-financial corporates, who, presumably, are genuine hedgers but will nevertheless contribute towards the systemic risk stemming from the use of OTC derivatives if they pass the buck to their bank by not posting their share of collateral.

Some issues relevant for discussion under the proposed regulations are detailed below and include the onerous collateral requirements, a central bank backstop for CCPs, the fallacy of the utility comparison, and reduced collateral reuse rate (velocity). There are still many other impediments to the successful implementation of the proposed reform agenda (eg, lower overall netting, no interoperability between CCPs, demand for segregated collateral, extraterritoriality and regulatory arbitrage).


Interoperability, or linking of CCPs, will increase each CCP’s clearing fund in line with the net open positions between them. So CCP a may hold or have access to collateral from CCPb, which may go bankrupt in the future, so that losses involved in closing out CCPb’s obligations to CCP a can be covered. However, legal and regulatory sources indicate that cross-border margin access is subordinate to national bankruptcy laws (such as Chapter 11 of the US Bankruptcy Code). It is unlikely that CCPa in one country would be allowed access to collateral posted by CCP b registered in another country. Nor is it in the interests of CCPs to change their business model and lose their niche market(s). The sheer collateral arithmetic to support interoperability is daunting.

Sizable collateral requirements

The 10 largest SIFIs will continue to keep systemic risk from OTC derivatives on their books since only standard OTC derivatives are mandated by regulation to move to CCPs. Regulatory efforts will introduce more new entities (ie, CCPs) that will also hold systemic risk from OTC derivatives. This goes against the intuition that suggests the need to minimise the number of CCPs (and benefit from additional netting), rather than increasing their number. Thus, collateral needs will be higher in the proposed world. Most of the major SIFIs’ derivatives books are largely concentrated in one “business” (a legal entity) to run the derivatives clearing business so as to maximise global netting. Some clients, such as sovereigns and US municipalities, are presently not in a position to post collateral. Due to exemptions, a significant part of this market will not reach CCPs. A 2011 Oliver Wyman and Morgan Stanley study also finds considerable additional collateral needs. The ISDA has also acknowledged the significant collateral needs resulting from moving derivative positions to CCPs, despite their (earlier) margin surveys indicating that most of this market is collateralised. The considerable collateral needs along with the exemptions imply that CCPs may not inherit all the derivative positions from SIFIs. As discussed above, interoperability of CCPs (on a cross-border basis) is unlikely, so overall global netting will not increase. Netting is the flipside of collateral needs and is discussed in Annex 6.1 of this chapter. The large banks active in the OTC derivative space are reluctant to unbundle “netted” positions on their books, as this results in deadweight loss and increases collateral needs.

Central-bank backstop

A CCP may face a pure liquidity crisis if it is suffering from a massive outflow of otherwise solvent clearing members, in which case the risk is that it will have to realise its investment portfolio at low prices. Assuming an external shock where everyone is trying to liquidate collateral simultaneously, this will lead to a problem if the CCP has repo’d out the collateral it has, cannot get it back, and for whatever reason does not want to pay cash to the members (effectively purchasing the securities at that price). In these circumstances, a central bank would be repoing whatever collateral the CCP would ultimately get back. In such instances, it would be more sensible to require the bank members (eg, JP Morgan, Credit Suisse) of the CCP to access the central bank and then provide the CCP with liquidity.

The CCP may also need central-bank support if it has suffered a series of member defaults and is subject to a run because of credit concerns. In this case, the CCP’s book is not balanced (since the trades of the defaulting members have fallen away) and if the central bank provides liquidity support it will be taking credit/solvency risk on whatever the net CCP position is (Panel 6.2). In this regard, the report “Principles for financial market infrastructures” from the Committee for Payments and Settlement Systems (CPSS) and the International Organisation of Securities Commissions (IOSCO) (FMIs) is a good background. A CCP failure should not be ruled out (Panel 6.3). As CCPs begin to clear more complex, less liquid and longer-term instruments, their potential need for funding support in extremis will rise. In the most extreme scenario, where a temporary liquidity shortfall at a CCP has the potential to cause systemic disruption, or even threaten the solvency of a CCP, it is likely that a central bank will stand ready to give whatever support is necessary. However, such an arrangement would create moral hazard. For example, in the US, under the Dodd–Frank Act, the Federal Reserve cannot bail out any derivatives dealer. More generally, there is no complete clarity on whether nonbanks would have access to central-bank liquidity. Sections 802 through 806 of Dodd–Frank generally authorise the Fed to provide liquidity support under unusual or exigent circumstances to CCPs that have been designated as systemically important. (The EU has similar language.) A taxpayer bailout is not ruled out but will be handled not by the Fed but by the US Treasury. Regulators are keen in avoiding a CCP bailout and thus the thrust to use variation-margin (VM) haircuts before a CCP defaults (Panel 6.3 and chapter 10)

Panel 6.2:Central Bank Backstop With and Without Interoperability

CCPs might need a central-bank backstop even if they take adequate collateral. Central-bank support for interoperable CCPs that are in distress could well span several jurisdictions, due to the associated contagion. Previous analytical work suggests that if a critical mass (about two-thirds) of OTC derivatives does move to CCPs, then about US$200 billion will need to be contributed to initial margin and default funds at CCPs (Singh 2010). As per Figure 6.1, augmenting default funds would imply that the four linked CCPs hold more than US$200 billion. In the context of this chapter, the focus is on CCPs relevant to OTC derivatives such as ICE Clear UK, ICE Trust US, LCH.Clearnet’s Swapclear, CME and Eurex (some of the newer names are not included). If these four CCPs are linked, they will augment their default funds, and additional netting will result from consolidation of exposures by participants at their chosen CCP. A back-of-the-envelope calculation, assuming the four key CCPs are about equal in size (each with about US$50 billion in default funds), suggests that each default fund may need to be augmented by another US$75 billion, midway between the envisaged US$50 billion and the US$200 billion maximum.

Increased netting results in lower residual risk across all SIFIs’ books. In an interoperable world, if international legal challenges are overcome, the increased netting benefits may exceed the extra funds needed to augment default funds at linked CCPs. In the more likely scenario (ie, under “no interoperability”), central-bank support will presumably be limited to the failed CCP in one jurisdiction, assuming there is no contagion. However, the present non-linked CCP world (and a realistic scenario, going forward), results in lower multilateral netting that will require much higher collateral costs for all users of derivatives.

CCPs and utilities

The revenue and benefits from OTC derivatives come from three sources: the origination fee plus netting on books plus the clearing fee. Banks will still keep all of the origination fee plus some of the netting (from OTC derivatives that do not clear). A utility has two characteristics: (a) government backstop but (b) at negotiated “economic rents”. So for CCPs to be utilities, all three revenue pieces mentioned above (which comprise the total economic rent) should be negotiable. But banks will never let go of the origination or structuring fee – this is the biggest piece. The negotiation between regulators and banks is such that this fee will remain undisclosed – especially since this fee straddles several items in their annual reports that fall under fixed income, currency and commodities (FICC). The comparison of CCPs as utilities is not apt unless it spans the full spectrum of “economic rents”.

Panel 6.3:CCP Resolution and Recovery: VM “Gain” Haircuts

Some circles of regulators continue to focus on CCP resolution and recovery that avoids taxpayer bailout of CCPs. The Bank of England emphasises that all CCPs, including those in the US and the eurozone, must have plans to allocate uncovered credit losses. VM haircutting is the obvious solution once the default fund, initial margins and other assessment rights are exhausted. There is an increasing consensus on avoiding taxpayer bailout of CCPs. To elaborate, if a clearing member (Deutsche Bank, say) defaults, it would contribute its share of the initial margin and the default fund. If losses are not fully recouped, then the CCP would dip into the mutualised loss in the overall default fund. After that, there would be haircuts on the VM. Most haircut rules do not directly hurt clients of banks (hedge funds, pensions, insurers and so forth), but market practice is such that in a clearing member–client relationship a clearing bank will pass on to the client only what it gets from a CCP.

LCH.Clearnet’s December 2012 letter to the European Commission says, “Open to questions is whether VM (and to some extent Initial Margin) haircutting should also be applied to client balances. Our rules currently allow this.” Also, see Chapter 10 for details on VM haircuts.

This microeconomics of variation-margin haircuts can also be viewed from a macro, too-big-to-fail angle. The UK is home to LCH.Clearnet, the largest CCP for interest-rate swaps by far. The net open interest position (or approximate risk metric for likely losses) of LCH.Clearnet may be too big for the Bank of England balance sheet. Although liquidity will not be a constraint to banks or nonbanks in the UK, it will be more difficult for the UK to bail out a large CCP such as LCH.Clearnet than the US or the eurozone (both of which are relatively bigger economies compared to the UK). So a VM haircut does provide an additional buffer to the taxpayers. As an example, if a pension fund has a swap/futures hedge and has a gain on its swap position but a loss on its futures position, then there may be a VM haircut on its swap receivable (as this is one route being considered), but this would ignore the loss on the pension fund’s futures position. This also translates into more asymmetry between those exempted from clearing and those that are mandated to clear, since clients of CCPs (such as a hedge fund) may have to contribute towards CCP resolution and recovery to avoid a CCP default. (The former FDIC chair, Sheila Bair (2013), testified to being “surprised at the lack of concern over the designation of ‘financial market utilities’ and particularly Section 806, which permits the Fed to provide safety net access to designated FMU”. Under US law, only the Treasury could provide funds for a bailout.)

Decrease in collateral velocity (or reuse rate)

The decrease in the “churning” of collateral may be significant since there is demand from some SIFIs and/or their clients (asset managers, hedge funds and so forth) for legally segregated/operationally commingled accounts (LSOCs) for the margin that they will post to CCPs. Also, the demand for bankruptcy-remote structures – another form of siloing collateral, which stems from the desire not to legally post collateral with CCPs in jurisdictions that may not have the central bank’s lender-of-last-resort backstop (ie, liquidity and solvency support) – will reduce rehypothecation. (Chapter 2 discussed collateral velocity, now much lower at 2.0 relative to 3.0 before Lehman.)

There are still other issues that may be important to consider. For example, regulatory arbitrage is likely due to the staggered implementation of this ambitious international agenda. Under Dodd–Frank, SIFIs’ banking groups can keep relatively safe OTC derivatives such as interest-rate, foreign-exchange and investment-grade credit default swaps (CDSs) on the bank’s book; the rest have be to “pushed” outside the banking entity (although this is not the case in Europe and Asia). Extraterritoriality issues that are being discussed in the US may also lead to regulatory divisions and possibly for booking of OTC derivative books to another jurisdiction (such as Asia) to “accommodate” and adhere to the final definition of extraterritoriality.

An Alternative to the CCP Route: Taxing Derivative Liabilities

A relevant metric that captures derivatives risk to the financial system is the exposure of the financial system to the failure of a SIFI that is dominant in the OTC derivatives market. This is captured by the SIFI’s total “derivative liabilities” (and not “derivative assets”). Derivative liabilities are also called negative replacement value (NRV) in Europe, and derivative assets are called positive replacement value (PRV). At present, a SIFI’s derivative liabilities are not directly targeted with a regulatory capital charge and are not reflected in risk assessments (see Annex 6.2). It is important to recognise that the ISDA master agreements allow SIFIs to net (or offset) their derivative asset-and-liability exposure on an entity. Thus, if, say, Goldman has a positive position with Citi on an interest-rate swap and a negative position with Citi on a credit derivative, the ISDA allows for netting of the two positions.

By using residual (ie, after netting, and after whatever collateral is posted) derivative liabilities as a yardstick, we thus provide a readily available metric to measure systemic risk from derivatives. The past is important to reflect on: the five largest European banks had about US$700 billion in undercollateralised risk in the form of (after netting) derivative liabilities as of December 2008. The US banks had, similarly, around US$650 billion exposure as of end-2008, as dislocations were higher then. The key SIFIs active in OTC derivatives in the US are Goldman Sachs, Citi, JP Morgan, Bank of America and Morgan Stanley. In Europe, the SIFIs that dominate this business are Deutsche Bank, Barclays, UBS, RBS, Credit Suisse and BNP Paribas.

Regulations have not mandated reducing residual derivative payables. There are other avenues to removing OTC derivative risk from the large banks’ books with similar underlying economics and perhaps lower collateral needs. For example, a levy on residual derivative liabilities (ie, after netting and after whatever collateral is posted) is a more transparent approach than moving OTC derivatives to CCPs, especially if the costs of bailing out CCPs are to be funded by taxpayers (although there is now an increased focus on VM haircuts that may avoid taxpayer bailout). If a levy is punitive enough, then large banks will strive to minimise their residual derivative liabilities – this, not the levy, is the primary objective. This will minimise systemic risk via the OTC derivatives markets if a large bank fails.

More importantly, as a by-product of the above levy, the residual derivative assets will also go towards zero. This will happen since the large banks typically have matched books, ie, the sizes of the derivative liability and asset positions at each bank are, on average, roughly the same. From a legal angle, this is an important point. Due to national bankruptcy laws, there is an asymmetry when a bank fails at T0 – the residual derivative assets cannot be used at time T0 as they go under receivership. Assets get stuck. The levy route (see Panel 6.4) brings into consideration the collateral stuck in residual derivative assets. The levy suggestion attempts to make those exempted from clearing also pay for their use of OTC derivatives (or have their banks pay on their behalf if banks want their business). So, if every OTC derivative user (including sovereigns, quasi-sovereigns and end-users) posted their share of collateral, there might be enough collateral within the OTC derivative markets that, if “reshuffled” appropriately, might not warrant large additional collateral. This would also go a long way in breaking the sovereign–bank nexus’s umbilical cord (see Chapter 9).

So, if exempted clients pay a levy such that residual derivative liabilities of the SIFI go towards zero, and those not exempted clear via a CCP, then the risk from derivatives on the SIFI’s book will be minimal. Furthermore, the SIFI may find it optimal not to unbundle sets of transactions that net (but comprise a combination of cleared and uncleared sets). This is where the mandatory requirement to clear only some transactions results in economic inefficiency. Transactions used to clear even before these proposed regulations – the market opted to clear without any clearing mandate.

More recent regulations are now addressing margin requirement for non-cleared trades; it is uncertain whether exempted clients (or their banks) will prefer bilateral clearing over central clearing.

Note that, in a “no mandatory clearing” world, there would be no exempted clients and no arbitrage, however, in this bifurcated cleared and non-cleared world (and also the legacy trades that will remain uncleared as they are not addressed by the new regulations), there will likely be arbitrage as banks and clients will try to minimise their overall costs of using OTC derivatives.

Panel 6.4:Analytics of the Tax on Derivative Liabilities

Under no interoperability, tail risks are less likely to decline. Let p denote the probability of a bailout in a CCP world, and let P measure the probability of the bailout of a SIFI in the (status quo) non-CCP world.

For p < P, overall tail risks in the CCP world would be lower than the tail risks in the non-CCP world. Increased multilateral netting via interoperability is one way this could happen, but is unlikely because the required legal conditions are not in place. Furthermore, no CCP offers cross-product netting, so contracts that net at a SIFI book may need to be “unbundled” when moved to two non-linked CCPs. Similarly, between-product netting may also lead to collateral inefficiencies since a standard–nonstandard combination would have to be unbundled: the standard contract would move to a CCP along with the associated collateral, while the nonstandard contract would stay with the SIFI and attract a regulatory charge. Such unbundling decreases overall netting. Thus, exante, it remains unclear whether the overall netting due to CCPs (primarily between products and not across products) will be higher than that from the unbundling of netted positions or other issues associated with moving derivatives to CCPs (eg, reduced rehypothecation of collateral due to the ‘siloing’ of collateral at CCPs, or demands for segregated collateral accounts by certain clients).

Another way to reduce tail risks is to take collateral from those who are not posting collateral. This can be done in the CCP world by regulatory incentives. But it could also be done in the status quo world by placing a levy or tax on derivative liabilities (which would result in revenue that could be used if a SIFI needs to be bailed out in the future). Now, let p1 and P1 denote the probability of a bailout when the present undercollater-alisation is reduced. Note that p1 < p and P1 < P. Moreover, p1 is largely exogenous due to regulatory uncertainty, while P1 is endogenous since the tax, T, can be calibrated to reduce the risk metric (ie, residual derivative liabilities in the non-CCP world). Thus, P1 can be less than p1 and further strengthens the tax argument analytically. However, we will make the “worst-case” assumption here that they are equal, ie, p1 = P1.

To summarise, the tail (or bailout) risk in the envisaged CCP world and the present SIFI-only world might well remain the same. However, the CCP world would have a bailout cost of C. The status quo world without CCPs may well have a similar bail-out cost C when a SIFI goes under, but this can be paid by the revenue T via the tax/levy that will be imposed on the large residual derivative liabilities of SIFIs who want to “carry” this systemic risk.

CCP worldStatus quo with tax
At present, probability of bailoutpP
Ex-post, probability of bailoutp1P1
Ex-post, cost, C, of bailout in n yearsp1CP1C – Σ nT=1T

As argued above, since p1 is not less than P1, the status quo with tax is economically more efficient. To be technical, if VM haircuts reduce or eliminate a taxpayer bailout, the C in CCP world may be lower than CCP in a “status quo with tax” scenario. However, a large T can change the arithmetic in favor of “status quo with tax” scenario.

Should Every Country Have a CCP? The Cases of Canada and Australia

Canada’s decision not to have its own CCP is based on sound economic analysis that many other key jurisdictions have not yet undertaken. For example, Canadian banks deal in non-Canadian currencies so will get higher netting benefits only if they access a global CCP. Since netting will not be substantially less for Canadian participants if they used a domestic Canadian-dollar-only CCP (putting these participants at a competitive disadvantage relative to their global peers), Canada has decided that it is not in its interests to foot the infrastructure cost of having a Canadian CCP, which may need to be bailed out (Canadian Securities Administrators 2012).

Generally speaking, large losses stemming to a bank from its OTC derivative positions – if it leads to bailout – will typically be picked up by the taxpayer from the jurisdiction in which the bank is located. Also, for example, derivative losses at branches of a Canadian bank in a foreign jurisdiction (eg, London) will also become a Canadian taxpayer liability. Moving OTC derivative positions from, say, a Canadian bank to a foreign CCP that is owned/incorporated in, say, the UK could shift some of the Canadian taxpayer liability related to cleared OTC contracts to a UK taxpayer liability if the UK had to bail out the CCP.

In particular, accessing an international CCP such as LCH Swapclear was deemed to be satisfactory since the following safeguard provisions identified by the Financial Stability Board (FSB) were viewed as being sufficiently in place (since resolution regimes for CCPs are not fully in place yet):

  • fair and open access by market participants to CCPs;

  • cooperative oversight arrangements for CCPs between relevant authorities;

  • resolution and recovery regimes that aim to ensure that the core functions of CCPs are maintained during times of crisis; and

  • appropriate emergency liquidity arrangements for CCPs in currencies in which they clear.

Australian banks, on the other hand, do not deal much in non-Australian-dollar OTC derivatives. Thus, netting benefits from cross-currency derivatives may not justify going to a global CCP (although, in 2013, Australia approved direct access for Australian bank interest-rate swap users to LCH Swapclear). Local CCPs may suffice and thus their present position is open to hosting a domestic CCP. In such a case, regulatory oversight of Australian dollar (AUD) derivative positions along with netting benefits of banks dealing in AUD derivatives will accrue to Australia. On the flipside, since CCPs will be systemically important, any cost of bailing out a domestic CCP will be an Australian taxpayer liability.

From a global systemic risk angle and collateral perspective, consolidation of CCPs will be welcome since risk will be less fragmented and collateral per unit of clearing will decrease.

Policy Issues

The only residual actor left in the financial system to bridge demand and supply will be the 10–15 banks that specialise and span the global cross-border collateral market. This would entail “connecting” clients (such as a pension fund) that have good collateral to lend to clients (such as a hedge fund) that do not have good collateral but need to post collateral acceptable to a CCP. In general, central banks, sovereign wealth funds and long-term asset managers desire collateral that is of low volatility, but not necessarily highly liquid. These entities should be net providers of liquidity in the financial system. On the other side are banks, hedge funds and mutual funds that have a dramatically shifting need for liquid and good collateral. So a market for collateral upgrades and transformations – in theory – could work.

However, many banks may not be able to transform collateral if the (final version of) leverage ratio encompasses all off-balance-sheet pledged collateral items. If the proposed definition stays the same, then banks will have to trade off balance-sheet leverage constraints against profitability from collateral trades. It should be noted that collateral transformation will further “interconnect” the financial system (and moving derivatives to CCPs was supposed to break the interconnectedness). If a pension fund finds collateral very expensive to post, it may not hedge its position – that will be risk in the financial system. Alternatively, the pension fund may skirt the use of OTC derivatives and use futures, as they are much cheaper than posting “expensive” collateral, and get away with it, but futures cannot mimic a 30-year hedge as an OTC (custom-made) derivative.

Alternately, some central banks may supply collateral directly to the nonbanks. But this would not only weaken the financial plumbing between banks and nonbanks (which in essence determines the repo rate), but also provide “puts” to nonbanks such as money market funds. The Fed’s 2013 reverse-repo programme is a step in this direction.

In summary, the proposed route to removing OTC derivatives from banks’ books creates new SIFIs, destroys the economics of netting on the books of the banks, silos collateral and decreases collateral velocity, and increases the interconnectedness of the financial system. Alternately, if every user of OTC derivatives contributed their share of margin(s) when using OTC derivatives (relative to the proposed bifurcated “clearing” and “non-cleared” worlds), the risk from derivatives at SIFIs would be eliminated. There would be no need for CCPs (Panel 6.5). In fact, some countries have opted not to have a CCP.

Panel 6.5:Taxpayers Should not be Made Accountable for Systemic Failure

Financial Times column by Manmohan Singh, October 17, 2012

Since the Lehman bankruptcy and AIG bailout, there has been increased momentum to move OTC derivatives from the books of large banks to clearing houses – or central counterparties (CCPs) – in effect moving the risk off large banks’ balance sheets. But moving counterparty risk from banks to CCPs does not eliminate it. It means risk will instead be shifted from individual banks to new institutions similar to concentrated “risk nodes” in the financial system. This may work in normal times. But what happens during the next crisis?

Little progress has been made on crisis resolution frameworks for unwinding large banks, let alone huge new institutions called CCPs that would house trillions of dollars in financial derivatives. Thus, the underlying economics of having more “too big to fail entities” needs to be justified. Financial statements show that each of the large banks active in the OTC derivatives market in recent years carries an average of $100bn of derivative-related tail risk; that is, the potential cost to the financial system from its collapse after all possible allowable “netting” has been done within the bank’s derivatives book and after subtracting any collateral posted on the contracts. Past research finds that the 10–15 largest players in the OTC derivatives market may have about $1.5tn in under-collateralised derivatives liabilities, a cost taxpayers may have to bear unless some solution to the “too-big-to-fail” question can be proffered.

Housing derivatives in one single global CCP backstopped and regulated by the leading central banks would have been an ideal “first-best” solution as it would enhance netting, reduce collateral cost and “house” overall risk in one place. A “second best” solution would have involved a few linked CCPs scattered around the globe. However, local politics has resulted in the least-best outcome. A plethora of CCPs are being created because countries such as Australia and Singapore do not want to lose oversight to an overseas entity incorporated in a foreign country.

The proposals to create these institutions also have several exemptions for key derivative users that dilute the intended objectives of such a move and increase the overall collateral requirements for the financial system because of the fragmentation that will follow in this new CCP world. Even then, the large banks will keep the more complex derivatives on their books. At present, collateral is fungible and the large banks do an excellent job in reusing it.

In a CCP world, a decrease in the reuse of collateral may be significant as there is increasing demand from some derivative clients for “legally segregating” margin that they will post to CCPs. The MF Global and Peregrine sagas have resulted in increased demand for segregation, so the collateral velocity or reuse within the OTC derivatives market will fall further and may lead to a cottage industry in “collateral transformation” reminiscent of the mid-2000 securitisation in the housing market. The result could be more, not less, moral hazard. In the most extreme scenario, a temporary liquidity shortfall at any of these CCPs would immediately cause systemic disruption. It is likely central banks, and governments, would have to give whatever support was necessary at taxpayers’ expense. In essence, this is a roundabout way for derivatives risk to be picked up by taxpayers.

What should be done instead? A levy on derivative liabilities is a more transparent approach given that the costs to bail out CCPs will ultimately fall on taxpayers. If the levy is punitive enough, large banks will strive to minimise their derivative liabilities, which could eliminate the systemic risk in the derivatives market should a large bank fail. This proposal addresses the source of the problem – under-collateralisation in this market – and does not bury it in technical jargon such as SEFs, FCM, DCM, DCO, DCE, MSP, LEI, portability, interoperability, non-cleared trades and extraterritoriality. The levy will force banks to take (and give) collateral with clients when it is due on the derivatives.

Also, some by-products of the levy will be most welcome. First, a fund from levy revenues could be used to bail out banks that prefer to keep OTC derivatives on their books and thus pay the levy. Second, when all derivative users including sovereigns post their fair share of collateral, banks will not need to hedge positions where they are in-the-money but with default risk. Demand for hedging leads to higher credit default swaps spreads that may increase the cost of debt issuance. The CCP proposal that regulators are so enamoured with is a sleight of hand that instead of resolving the “too-big-to-fail” problem deflects it back to taxpayers.

Annex 6.1: Netting Fragmentation – Cleared and Non-Cleared OTC Derivatives

Before the regulatory proposal to move OTC derivatives to CCPs, there were n big banks with n-1 netting sets between them. Those netting sets were fully cross-product within each bilateral relationship. The beneficial effect of netting on risk (and thereby capital and margin) was around 80–90% of the in/out-of-the-money positions (let us say X).

If there was one global “CCP” that became the counterparty to all OTC derivative (including across products and currencies) transactions among the n big banks, this would result in n netting sets each between a bank and CCP (so netting sets increase from n-1 in the world before regulations). This would not only preserve the cross-product nature of the netting set but also make the overall risk of the CCP multilateral; this would be good and the beneficial effect would be greater than X, let us say Y.

However, a global CCP is unlikely due to political, legal and business model constraints. There will be multiple CCPs p (including CCPs focused on certain OTC derivative product categories such as CDS-only, or IRS-only, due to their specialisation and niche business models). Thus, the netting sets proliferated to be n x p plus of course the netting sets associated with the remaining bilateral trades, which were still n-1 in number but much less diversified because many trades had been moved to the relevant CCP.

Algebraically, the original n-1 netting sets will become np+n-1 or n(p+1)-1 netting sets. The unbundling of the original netting sets will create more sets (numerically) but smaller and less diversified in content, until a CCP can offer to clear all OTC derivatives (unlikely so far). So the netting benefit is likely Z, which will be much less than X or Y. (There have been studies to show that, initially, Z will be less than X, but Z will overtake X only when a sizable part of the OTC D market will be offloaded to CCPs, and when the number of CCPs will consolidate from, say, p to about q(where q < p). For illustration, n is about 10–15; p is envisaged to be (initially) between 20 and 30, since many countries want their own CCP. It remains to be seen if the number of CCPs will consolidate where q is a single-digit number (eg, LCH, ICE, CME, Eurex). Note that ICE Europe and ICE US are two CCPs from a netting perspective. Also, if LCH UK has a branch in the US for US clients, then netting will be fragmented netting since LCH UK will net independently of LCH US

Non-cleared trades will continue to remain on the books of the banks, but after netting bundles are broken. It is like breaking a Ming vase by dropping it and then picking up one of the pieces and saying, “Well at least this bit’s not broken.” The n-1 remaining bilateral netting sets are proposed to be subject to different margin rules, and so the number of netting sets will become np plus at least (2n-1) or, n (p+2)-1, since each bank’s book will not net linearly as in the past.

Thus, the netting benefit from here will be even smaller than Z. -Netting-set fragmentation is real and increases risk. Thus, in line with the economics of ISDA agreements, the non-cleared trades should be allowed to net, to limit fragmentation and collateral silos.

Annex 6.2: Typical OTC Derivative Position from a Sifi’s Financial Statement
March 2009




Derivative contracts for trading activities
Interest rates1,171,8271,120,430
Derivative contracts accounted for as hedges under SFAS No. 133(1)
Interest rates24,347(4)1
Gross fair value of derivative contracts1,938,7541,803,033
Counterparty netting(2)(1,685,348)(1,685,348)
Cash collateral netting(3)(149,081)(27,065)
Fair value included in “Trading assets, at fair value”104,32 5
Fair value included in “Trading liabilities, at fair value”90,620

    Other Resources Citing This Publication