This ubiquity has led to concern as to whether there is enough collateral to meet all the roles it must play. In this article we review the key features of secured transactions, setting out the principal risks that collateralisation brings both to the parties involved and to the system. This discussion is illustrated by two of the key failures during the crisis, AIG and Lehman Brothers. We also examine the structural features of the secured financing market, noting in particular an important lesson to be gleaned from the size of large US dollar tri-party repo portfolios.

We then turn to the post-crisis financial system in detail, highlighting the role of collateral in the principal reforms made to wholesale markets. The paper ends with a discussion of the risk profile of the financial system that will result once these reform efforts are complete.

How collateral works

Financial collateral is typically used when one party has (or may in the future have) a credit exposure to another which is not mutually acceptable. The lender may not wish to have substantial exposure to non-performance by its counterparty. The exposure is mitigated by means of a financial asset. This asset, the collateral, is posted for the benefit of the party owed.

If we assume that the collateral is effective – that it does indeed reduce the amount owed on default – then the reduction in exposure depends on the value of the collateral posted:

  • Net current exposure = gross current exposure – current collateral value


Common situations

There are two cases which together cover much of the use of financial collateral:

  • Exposures arising from derivatives portfolios between the parties; and
  • Exposures arising from secured lending such as repo.


Note that while collateral is used more broadly than this, for instance in residential or commercial mortgages, we confine our attention to the situation where the asset pledged is a financial instrument such as cash, bonds or equities. We will also assume in what follows that it is clear how the exposure before collateral is determined. In particular a key issue is whether close-out netting applies to the portfolio of exposures between the counterparties which is being collateralised. Often it will, especially under the common forms of documentation for OTC derivatives contracts (ISDA) and repo (GMRA), but the legal analysis here may be subtle.

The dynamics of exposure

It is clear from the simple formula above for net exposure that if the gross exposure goes up and the collateral value remains static, then the net exposure goes up too. If we want the net exposure to remain constant, more collateral must be posted. Thus there can be collateral calls, whereby the posting party has to post an additional value of collateral assets. These have an impact on the poster’s liquidity position: they have to fund the additional value which must be posted. The value of a derivatives portfolio will often vary significantly over time. Thus if the parties to it aim for a constant (perhaps zero) exposure, collateral amounts must be recalculated frequently. Daily collateral calls equal to the change in value of the exposure from one day to the next are therefore common. This practice is known as the posting of variation margin (or VM). VM can clearly flow in either direction as the exposure may either increase or decrease in value, so two-way agreements, where each party agrees to collateralise their exposure, are widespread. VM acts like a settlement of the value of the portfolio, so it is almost always exchanged in cash.

The ultimate exposure the postee bears is not the net current exposure, for if the poster fails to perform:

  • There may be a grace period under which this non-performance can be cured;
  • Once this has expired, appropriate notification may have to be made;
  • Only then may it be possible for the collateral to be liquidated and the exposure closed-out.


During this period the value of the exposure, or the value of the collateral, or both may change. Indeed, if the non-performing party is a large financial institution, then it is likely that the markets are in turmoil as a result of their default, and so volatility is to be expected. The period between the last time there was performance on a margin call and close-out of both the exposure and the collateral is known as the margin period of risk or “MPOR”. Regulations typically require a MPOR of five to ten days for liquid OTC derivatives, with shorter periods sometimes being set for exchange-traded portfolios and longer ones for portfolios containing inter alia less liquid positions.

In order to protect themselves against changes in exposure over the MPOR, market participants sometimes take an additional amount of collateral known as initial margin (IM). Figure 2 illustrates the use of IM and VM to cover aspects of the exposure between two parties created by a derivatives portfolio.

IM should be big enough to cover likely changes in portfolio value over the MPOR. Thus for instance if the MPOR is estimated to be no more than five days, IM might be estimated as the 99th percentile largest move in portfolio value over a five day period.

The analogue of IM for securities financing transactions is the haircut: often, a greater value of the security is posted than the finance provided, and this overcollateralisation provides some protection against the risk of falls in the value of the collateral between default and close-out.

The dynamics of collateral value

Turning back to the expression for net exposure, it is clear that if the collateral value goes down and the gross exposure value remains static, then the net exposure goes up. This can happen if the collateral posted includes assets subject to market risk such as bonds or cash in a different currency from that of the exposure. Collateral haircuts are often agreed to mitigate this risk. A collateral value fall (or haircut increase) will often cause a collateral call, and hence a liquidity burden on the poster.

This means that assets become less desirable as collateral as their credit quality declines. For instance, if a government bond falls in value, perhaps due to sovereign risk, then all those parties who have used that bond as collateral have to find extra funds to meet the resulting collateral calls. A systemic liquidity drain is possible if many bonds from the same government are widely used as collateral. (A related effect – artificially low term interest rates due to the excess demand for government bonds from collateral posters – is also potentially an issue.)

Collateral eligibility can be a problem, too. To see this, suppose that two parties have agreed that only a certain class of assets can be posted, such as bonds rated single A or better. If a given bond is downgraded, it may become ineligible as collateral. The poster then has to sell the bond and post a different eligible asset. These sales, if sufficiently widespread, can exacerbate the price falls caused by declining credit quality.

Moving collateral

We have used the neutral term “post” for the action of collateralisation. It is now time to address what this means. There are two classes of mechanism:

  • Title to the collateral may be transferred; or
  • A security interest may be granted over the collateral.


These are known, respectively, as title transfer and security interest. In title transfer, because the postee holds the collateral asset, there is (at least absent any other contractual provisions) no restriction on how it can be used. The postee can sell it, post it on, or encumber it in other ways. (Of course, the contract under which the asset is posted – or the law – may well require the return of the posted asset, or perhaps another asset to equivalent value, and the postee must be cognisant of this requirement.)

Security interest, in contrast, does not necessarily lead to the postee having title to the asset. This can only be acquired under certain circumstances. The postee should for instance ensure that it can gain title to the asset rapidly should the poster fail to perform, so that the asset can be liquidated against the exposure. It may also wish to have the right to re-use the asset. Thus it may be agreed that the postee can rehypothecate the collateral asset, posting it on against its own exposure to another counterparty.

It should be noted that granting a security interest over cash is problematic, so cash margin is typically title transferred.

We do not discuss the legal complexities of these two mechanisms in detail: they are considerable, and depend on numerous factors including the domiciles of poster and postee, the contractual provisions, and the law of the contract. It is also worth observing here that in some countries it is uncommon for either party to actually hold a security any more, in the simplesense: rather securities remain registered at a central securities depository (CSD), and claims over these are recorded in accounts at a custodian and possibly subcustodian(s). Thus a practical posting may well involve poster, postee, their respective custodians, and one or more CSDs. Any complete legal analysis must involve all of these parties, their various segregation arrangements and legal framework(s), and the contractual provisions between them.

The poster’s risk to the postee

The postee may well have their risk to the poster reduced by collateral: the inverse is less clear. A key problem here is over-collateralisation: if the exposure is 100, and collateral worth 120 has been title-transferred, then the postee has no exposure to the poster, but the postee has a claim for the return of the excess, 20. This is often an unsecured claim, so in effect the exposure has flipped: now the postee owes the poster money. Collateralisation is in this sense a Goldilocks technique – it needs to be just right.

This issue has led to considerable interest in the use of third parties to hold collateral. To  illustrate with one example from many, the poster’s custodian may be in a good position to grant an interest to the postee over a securities account whose contents can vary with the poster’s business needs, subject only to the requirement that sufficient eligible securities to meet the contractuallyagreed collateralisation level are present.

Specifying collateralisation

Historically, collateral provisions could be privately negotiated between parties within wide bounds. Thus in OTC derivatives, the credit support annex to a given ISDA master agreement would set out, inter alia, whether collateral was required, what assets could be posted, how and how often the exposure was determined, how the amount of collateral to be posted was determined based on that exposure, how soon the call must be met, and so on16.

Similarly standard documentation arose to specify securities financing transactions. One key feature of both of these standards is that failure to perform on the collateral requirements is an event of default known as credit support default.


In this section we look at how the dynamics outlined above led or can lead to financial stress. This will then illuminate our discussion of the reform agenda in the next section.


The story of AIG’s distress and rescue is relatively well-known, so we highlight only the main points of the narrative:

  • AIG took substantial amounts of risk by writing a large portfolio of credit default swaps, mostly referencing asset backed securities (ABS);
  • When this business began, AIG was a highly-rated company, so AIG’s counterparties were willing to enter into these trades without collateral;
  • However, to guard against future possible falls in AIG’s credit quality, these counterparties negotiated that AIG would have to post collateral were AIG’s rating to decline sufficiently;
  • AIG’s rating did decline past the threshold, forcing it to post collateral.
  • Moreover, the value of the ABS that AIG had written protection on had fallen substantially, resulting in the counterparties’ exposure to AIG – and hence the collateral required – being substantial.
  • AIG was not sufficiently liquid that it could meet the resulting collateral calls, and hence it would likely have suffered a credit support default had it not been bailed out.


This is an extreme example of a phenomenon that has increasingly been of concern to authorities: that of the procyclicality of collateral requirements.

If collateral requirements rise excessively in stress, then the requirement to meet them can cause financial stress. The argument in short is that AIG was rescued to prevent its collateral support default from endangering its counterparties and causing an even more disorderly market in the underlying ABS.


There were numerous reasons for Lehman’s downfall including overleverage, injudicious investment (and in particular imprudent failure to decrease risk-taking as the crisis developed), and legal standing. The Valukas report sets these out in detail.

The proximate cause of Lehman’s failure, though, was liquidity. As the report says, Lehman was “heavily reliant upon wholesale financing sources” to fund “a substantial portion of its balance sheet every 24 hours using overnight repos”.

This meant that the confidence of Lehman’s repo counterparties in the firm “was critical. The moment that repo counterparties were to lose confidence in Lehman and decline to roll over its daily funding, Lehman would be unable to fund itself and continue to operate.” In effect, then, Lehman suffered the modern version of an old-fashioned bank run. Liability holders withdrew funding faster than the firm could meet their claims, and so, unable to access sufficient liquidity, it failed. What this example makes clear is that collateral – such as the bond in a repo – is not always sufficient protection to keep the confidence of counterparties. Rather than run the risk that a firm may fail and that the collateral posted may not prove sufficient to meet the exposure, counterparties sometimes decline a transaction. Given the intense market turbulence after Lehman’s failure, and the resulting uncertainty in the liquidation value of collateral assets, this is understandable. It does however call into question the efficacy of collateralisation, at least with market haircuts, as a credit risk mitigation mechanism in stress.

American tri-party repo

The tri-party repo market is an important source of funding for financial institutions. For historical reasons, the US market evolved with a structure which gave rise to significant systemic concerns. Many of these have been or are, at the time of writing, being addressed. Here we focus on one of the most intractable of those issues, that of the liquidation period which might be required for large collateral portfolios.

The difficulty is this:

  • The USD tri-party market is sufficiently large that some firms finance big (multi-hundred-billion dollar) portfolios using it;
  • The MPOR of these big portfolios is likely to be large, perhaps twenty days or more for some parts of the portfolio;
  • And the haircut typically taken on tri-party portfolios is not large enough to prudently cover these extended MPORs.


This is one of the complexities of collateralisation: taking some collateral reduces counterparty credit risk, but as the portfolio of collateral grows, so a concern develops that it may not be possible to liquidate it speedily. This is especially the case in the febrile markets likely after the default of a large financial institution, with numerous counterparties liquidating collateral portfolios at the same time. A mechanism to reduce the risk of market disruption caused by fire sales of large collateral pools may enhance market stability here.

This situation represents a common micro- vs macro-prudential conflict. In the small, collateral protects the postee, reducing direct credit risk. But in the large, the actions of all collateral takers may create contagion and instability in asset markets either through procyclical margin calls (causing forced asset sales to generate the cash needed to make the call) or through fire sales of collateral portfolios after a default.


After AIG, Lehman, and the other events of the 2008 crisis supervisors rightly wanted to address direct interconnectedness. Systemic institutions should not be so exposed to each other’s performance that the failure of one would endanger others. Collateral had a key place in the resulting reform efforts as a mitigant of counterparty credit risk.

For OTC derivatives, the key initiatives are:

  • The G-20 commitment to centrally clear standardised OTC derivatives at CCPs (“Central Counterparties” otherwise known as clearing houses), developed in legislation such as EMIR and the Dodd-Frank Act. OTC derivatives CCPs require both IM and VM on cleared portfolios both as a commercial matter and under the relevant Principles for Financial Market Infrastructure. Moreover, bonds posted as IM to a CCP are not freely reused, and cash must be reinvested according to the CCP’s (necessarily constraining) investment policy.
  • The requirement for many institutions to post IM and VM to each other on bilateral OTC derivatives should exposure exceed a de minimis threshold, and (under the same regulation) to limit rehypothecation of collateral assets.


Together these reforms mean not just that more collateral is needed, but also that it is more difficult for it to flow through the system. Meanwhile, reforms aimed at addressing liquidity risk have also affected the demand for collateral. For instance:

  • Basel 3 contains provisions requiring banks to hold large liquid asset buffers to meet its liquidity coverage ratio and net stable funding ratio requirements. The assets which are eligible for inclusion in these buffers are typically just those which are most in demand to collateralise derivatives or funding transactions.


Many other reforms also impact on collateral demand, notably:

  • Entity structuring requirements, such as ring-fencing and “pushing out” derivatives trading from deposit takers, will increase the need for collateral (as exposures that used to be within one netting set will now be split into two or more);
  • Changes to the definition of large exposures may also increase bank collateral needs;
  • Funding markets continue to incentivise banks to use secured instead of unsecured transactions.



The rise of collateralisation has intensified a number of financial stability concerns. In this section we review these concerns, illustrating some of the behaviours of the system that is emerging from the reforms detailed above. We organise this discussion around the themes offinancial system structure, transaction design, and potential over-reliance on collateralisation.

Financial system structure

The sheer amount of collateral that must flow around the financial system is enormous. For individual firms, this posts operational challenges. How can they keep track of available and required collateral? Timing is paramount here, as many postings have to be made in a short window: CCPs typically demand that the previous night’s margin calls have been met by 9am, for instance.

There are three aspects to the question “is there enough collateral to support the post-reform financial system?”. These are:

  • Is there enough collateral in the broad, across the system, during ordinary times? Official sector work indicates that there is.
  • Is there enough collateral to meet collateral requirements in stress? Here the answer is less clear, and further work may be needed to reduce the procyclicality of collateral requirements.
  • Will particular classes of market participant, or particular currencies, suffer collateral shortages? It seems likely that the answer to this is yes, and indeed some central banks have already developed policy tools to address collateral scarcity.


Those parties who find themselves asset rich but eligible-collateral poor will have to transform the assets that they have into ones that they can post: a fully invested fund, for instance, will not have cash to post as VM. The repo market is the obvious tool to address this.

As a related matter, current collateral practices are often inefficient: excess collateral is sometimes not recalled, and collateral users may not have the systems needed to allocate their collateral optimally across central banks, bilateral and tri-party repo counterparties, bilateral derivatives counterparties and CCPs. Collateral optimisation is therefore a growing trend. The net result of both collateral transformation and collateral optimisation is increased complexity: more trades, larger gross exposures, additional flows on tighter deadline, and new roles for some market participants. It is to be hoped that these changes do not cause the collateral equivalent of a motorway pile-up.


Cash collateral typically earns a return, often the overnight indexed swap rate, or OIS. That reflects a cost to the poster, as few parties fund more cheaply than OIS: but it also poses a challenge to the postee, who must earn at least OIS on cash collateral. This is not entirely straightforward especially without taking credit risk, so postees increasingly need sophisticated investment risk control frameworks to manage these issues.

The haircut creates another tension in collateral agreements. For many transactions, market standard haircuts exist, and it can be difficult to agree increases over these standards in particular instances. However, as we saw with the tri-party repo example above, these standards are often imprudent for large collateral pools where the MPOR may be extended. Postees should have analytics available to monitor these issues, and to appropriately capitalise residual credit risk arising from haircuts which do not prudently cover the likely liquidation period of illiquid exposures or collateral pools.

Market participants also need to be cognisant of the systemic implications of the definition of collateral eligibility. It may seem prudent to only take a restricted range of high quality liquid assets as collateral, but if everyone else does that too, then the resulting shortage of those assets could distort the market and make the liquidation of collateral after a default more difficult.

It may be that in stressed conditions a wider range of collateral, prudently haircut, may better protect postees than a narrower one.

Over-reliance as a single mechanism?

Our brief survey of the role of collateral in post-crisis financial reform indicates its ubiquity. The question naturally arises as to whether we have focused too much on a single risk – direct exposure – and a single technique – collateralisation – at the expense of other forms of contagion and other risk mitigants. In particular collateral reduces risk-based exposure and hence capital requirements, so the capital required to support counterparty credit risk in the post reform system will drop substantially. Collateral is therefore not an additional protection for the financial system, but also a partial replacement for existing safeguards.

The global legal framework for collateralisation is fragmented and complex, and it interacts with bankruptcy law and with the evolving recovery and resolution framework for financial institutions. There has even been discussion as to whether the current safe harbours that exist for netting and collateral in, for instance, US bankruptcy law are socially optimal27. Collateral does have the advantage that, unlike capital, it is an unmutualised resource provided by the defaulter, but the shorthand description here might perhaps be “defaulter’s unsecured creditors pay” rather than the better known “defaulter pays”. This perhaps suggests that neither the statutory framework nor its benefits are currently sufficiently well-organised and well-agreed that they should necessarily form the predominant tool for counterparty credit risk mitigation.

The liquidity risks created by collateralisation are coming into greater focus, and we expect further discussion of them to feature in the policy agenda in coming years. Meanwhile market participants would be well advised, when negotiating the contractual provisions of collateral agreements, to consider the risk of forcing their counterparty into a credit support default. A risk mitigant that creates defaults is of dubious value so the management of collateral-related liquidity risk is a key issue for both posters and postees.


David Murphy is founding principal of rivast consulting, a leading risk management and derivatives consultancy. He has written extensively on financial stability issues, and his latest book, OTC Derivatives: Bilateral Trading and Central Clearing was published by Palgrave Macmillan in August 2013. Email: