7
Margin (Collateral) and Settlement

Long-dated transactions such as derivatives suffer from the problem that, whilst the current exposure might be relatively small and manageable, the future exposure could increase to a relatively large, unmanageable level. An obvious way to mitigate this problem is to have a contractual feature in the transaction that permits a risk-mitigating action to reduce a large exposure.

Future exposure reduction can most obviously be achieved by having the contractual right to demand some form of security or margin (collateral) as a mitigant against that exposure.¹ Margin-posting is a periodic process that ensures continuity of the underlying transaction(s) but acts as a potential mitigant for both parties in a transaction. On an exchange, margin-posting represents a daily settlement, whereas in over-the-counter (OTC) derivatives it represents a collateralisation process.

Whilst margin has become a standard feature of exchange-traded derivatives and other simple products such as repos, it is not a standard component of OTC derivatives, especially those involving end users. Regulation is promoting more standard margin practices in OTC derivatives markets (Section 2.5), although some end users will be exempt from such regulation.

In OTC derivatives, other contractual features sometimes exist which can be seen as a more basic form of margining because they may be less periodic (such as resets) or simply cause a transaction to be terminated early.

7.1 TERMINATION AND RESET FEATURES

7.1.1 Break Clauses

A break clause allows a given transaction to be terminated, either mandatorily, optionally, or contingent on some defined event (such as a rating downgrade). Such clauses may apply to one or both parties in a transaction. Prior to the financial crisis, break clauses were typically required by banks trading with certain (often uncollateralised, i.e. non-margined) counterparties. More recently, it has become common for counterparties such as asset managers and pension funds to require break clauses linked to banks' own credit ratings due to the unprecedented credit quality problems within the banking sector during the global financial crisis (GFC).

For example, an International Swaps and Derivatives Association (ISDA) Master Agreement allows the specification of an Additional Termination Event (ATE), which permits a party to terminate transactions in certain situations. The most common ATE is in relation to a rating downgrade (e.g. below investment grade). For unrated parties such as hedge funds, other metrics such as market capitalisation, net asset value, or key person departure may be used.

ATE provisions are typically defined at the counterparty level in the ISDA schedule. However, they can be specified in a transaction confirmation. More commonly, individual transactions may reference similar terms and these are typically referred to as ‘mutual puts’. Such clauses often do not reference specific events but are either mandatory (meaning the transaction will terminate at the specified break date) or optional (where the party has the right to break the transaction). It may be considered advantageous to attach such a clause to a long-dated transaction (e.g. 10 years or above), which carries significant counterparty risk over its lifetime. For example, a 15-year swap might have a mutual put in year five and every two years thereafter.

Sometimes mandatory breaks are used because a party is unable (due to internal policy such as credit limits, for example) to execute a long-dated transaction. In such cases, there may be an expectation or ‘gentleman's agreement’ that a new transaction will be executed at the break time. This is similar to a resettable transaction discussed in Section 7.1.2.

The exercise or triggering of a break clause does not necessarily cause a transaction to terminate, and the party against which it is exercised may have to comply with one or more of the following options:

• provide a replacement counterparty;
• provide a counterparty offering a third-party guarantee;
• terminate the transaction(s) at their current replacement value; or
• post some sort of margin.

In the event that a break clause results in a termination of a transaction(s), this is typically done at the prevailing mark-to-market (MTM) value, as determined in the documentation, typically involving pricing via a panel of banks. This is typically a base valuation (Section 5.3.6) and does not, therefore, share some of the problems with the determination of the valuation that is associated with close-out netting and xVA (Section 6.3.4), although problems sometimes still exist.²

An ATE is obviously designed to mitigate against counterparty risk by allowing a party to terminate transactions or apply other risk-reducing actions when their counterparty's credit quality is deteriorating. This may be considered particularly useful when trading with a relatively-good-credit-quality counterparty and/or long-maturity transactions. Over such a time horizon, there is ample time for both the value of the transaction to become significantly positive and for the credit quality of the counterparty to decline.

It is worthwhile comparing the different types of break clause with the following considerations (Table 7.1):

• Counterparty risk reduction. This refers to a firm's view of the counterparty risk reduction benefit, such as in consideration of credit limits. It is likely that all types of break clause will be deemed beneficial here, with some caution over the usefulness of optional and rating-based triggers.
• Credit value adjustment (CVA) reduction. This probably depends on the view of the CVA desk on the benefit of the break clause and their willingness to incorporate it in the price, and the associated view of whether it is appropriate to include it in CVA accounting assumptions. It is likely that such a desk may only see the benefit of mandatory breaks since they have no control over optional and rating-based triggers.³

  Table 7.1 Illustration of the risk-mitigation benefit of different types of break clause.

  	Counterparty risk reduction	CVA reduction	LCR	Capital relief
  Mandatory	✓	✓	✓	✓
  Optional	(✓)	✕	✓	✕
  Rating-based	(✓)	✕	✕	✕

• Liquidity coverage ratio (LCR). The LCR forces banks to pre-fund the requirements of rating-based triggers (Section 4.3.3).
• Capital relief. It is likely that capital relief will only be achievable for mandatory breaks. Even here there may be problems if a bank has a history of waiving such breaks, which may cause a regulator to question the risk-reducing benefit of the break clause. It is difficult to achieve capital relief through optional breaks, although if a bank can show that such breaks are rigorously exercised, then this might be possible. Finally, Basel capital rules explicitly rule out any capital relief from rating-based triggers (BCBS 2011b).

The problem with break clauses as a risk mitigant is that they potentially damage client relationships, and therefore salespeople in banks would be against exercising them (especially if the customer has not been downgraded and has a good credit standing), whereas the CVA desk (xVA desk) and risk management department would be in favour of making use of them wherever possible. This is essentially part of a moral hazard problem, where front-office personnel may use the presence of a break clause to get a transaction done, but then later argue against the exercising of the break to avoid a negative impact on the client relationship. Banks should have clear and consistent policies over the exercising of optional break clauses and the benefit they assign to them from a risk reduction point of view. There must not be any lack of internal clarity around who in a bank is empowered to exercise a break clause.

Behaviour around breaks has changed in recent years, with banks becoming much more willing to exercise optional break clauses and clients accordingly expecting such behaviour. This has occurred because banks are much more sensitive to the valuation impact of breaks on xVA terms and also capital costs and may put maximising profitability and reducing capital ahead of client relationship concerns.⁴ Hence, it might be considered that even optional breaks should now offer CVA reductions (Table 7.1).

Recent years have highlighted the potential dangers of ATEs and other break clauses, in particular:

• Risk-reducing benefit. Whilst an idiosyncratic rating downgrade of a given counterparty may be a situation that can be mitigated against, a systematic deterioration in credit quality is much harder to nullify. Such systematic deteriorations are more likely for larger financial institutions, as observed in the GFC.
• Weaknesses in credit ratings. Breaks clearly need to be exercised early before the counterparty's credit quality declines significantly and/or exposure increases substantially. Exercising them at the last minute is unlikely to be useful due to systemic risk problems. Ratings are well known for being somewhat unreactive as dynamic measures of credit quality. By the time the rating agency has downgraded the counterparty, the financial difficulties may be too acute for the clause to be effective. This was seen clearly in relation to counterparties such as monoline insurers (see Section 2.4.4) in the GFC.
• Cliff-edge effects. The fact that many counterparties may have similar clauses may cause ‘cliff-edge’ effects, where a relatively small event such as a single-notch rating downgrade may cause a dramatic consequence as multiple counterparties all attempt to terminate transactions or demand other risk-mitigating actions. The near failure of AIG (Section 2.4.4) is a good example of this.
• Modelling difficulty. Breaks are often very difficult to model since it is hard to determine the dynamics of rating changes in relation to potential later default events and the likelihood that a break would be exercised. Unlike default probability, rating transitions probabilities cannot be implied from market data. This means that historical data must be used, which is, by its nature, scarce and limited to some broad classification. This also means that there is no obvious means to hedge such triggers. One exception is where the break is mandatory, where the model may simply assume it will occur with 100% probability.

7.1.2 Resettable Transactions

A reset agreement is a clause to periodically reset product-specific parameters determining the value of a transaction and therefore avoid either party becoming strongly in-the-money. Reset dates may coincide with payment dates or be triggered by the breach of some market value.

For example, in a resettable cross-currency swap, the MTM value of the swap (which is mainly driven by FX movements on the final exchange of notional, especially in floating-floating structures⁵) is cash-settled at each reset date. In addition, the FX rate is reset to (typically) the prevailing spot rate. The reset means that the notional on one leg of the swap will change. Such a reset is similar to the impact of terminating the transaction at its MTM and executing a replacement transaction at market rates, and consequently reduces the exposure to zero at each reset date (typically they are every quarter). An example of the impact of such a reset is illustrated in Figure 7.1, which shows the exposure over one year and the impact of quarterly resets. Figure 7.2 shows the impact of the reset on exposure over the lifetime of the transaction in a scenario where the value of the transaction would become significantly positive (in-the-money) due to a relatively large move in the FX rate.

Since resets are mandatory and not linked to any external event, they are much easier to assess and model than break clauses. One simply needs to apply the logic of the reset at each contractual date, as shown in Figure 7.1.

A reset can be seen as a weaker form of margining because the margin is usually posted on a daily basis, whilst refixes are much less frequent. Another difference is that a refix is a settlement, whereas margin (in OTC derivatives) is normally posted as security against an exposure.

Figure 7.1 Illustration of the impact of reset features on the exposure of a long-dated cross-currency swap over the first year. Resets are assumed to occur quarterly and are shown by dotted lines.

Figure 7.2 Illustration of the impact of reset features on the exposure of a long-dated cross-currency swap. Resets are assumed to occur quarterly.

7.2 BASICS OF MARGIN/COLLATERAL

7.2.1 Terminology

Historically, ‘collateral’ has been a term used within OTC derivatives markets and contractual agreements such as ISDA Master Agreements. However, ‘margin’ has been used in exchange-traded markets to represent a similar concept. Accordingly, there are terms such as ‘independent amount’ (OTC markets) and ‘initial margin’ (exchange-traded markets) that are, to some extent, interchangeable.

The industry is generally converging on the use of the term margin over that of collateral. For example, relatively recent regulatory rules for OTC derivatives are known as the ‘bilateral margin requirements’ (Section 7.4). The term margin will, therefore, be used from now on, with occasional references to collateral and similar terms. The terms ‘collateralised’ and ‘uncollateralised’ will still be used to define the presence, or not, of margin in OTC derivatives.

7.2.2 Rationale

Margin represents a requirement to provide collateral in credit support or settlement. In exchange-traded markets, margin acts as a daily settlement of the value of a transaction. The fundamental idea of OTC derivatives margining is that cash or securities are passed (with or without actual ownership changing) from one party to another as a means to reduce counterparty risk. This different treatment in exchange-traded and OTC markets is well-known and forms the basis of the difference between futures (exchange) and forward (OTC) contracts. ISDA (2015) estimates that there is about $5trn in margin supporting bilateral derivatives transactions, with cash becoming increasingly common.

The difference between margin as settlement (exchange-trading) and margin as a form of security (OTC markets) is important. The benefit of the latter is that it is flexible and different currencies of cash and various securities (and, in theory, even non-financial collateral) can be used. However, margin as security is not economically owned and has associated risk, such as legal challenges regarding ownership. Margin as settlement is more convenient as its value is realised immediately, but it must, therefore, be a cash payment, most obviously in the same currency as the underlying transaction. This chapter will be largely concerned with margin in bilateral OTC derivatives, with centrally-cleared OTC derivatives being the subject of Chapter 8.

Whilst break clauses (Section 7.1.1) and resets (Section 7.1.2) can provide some risk-mitigating benefit in these situations, margining is a more dynamic and generic concept. A break clause can be seen as a single payment of margin and termination of the transaction. A reset feature is essentially a periodic (infrequent) settlement payment, similar to margin, made to neutralise an exposure.

A margin agreement reduces risk by specifying that margin must be posted by one counterparty to the other to support an exposure. In OTC derivatives markets, the margin receiver typically only becomes the permanent economic owner of the margin (aside from any potential legal challenge) if the margin-giver defaults. In the event of default, the non-defaulting party may retain the margin and use it to offset any losses relating to the positive value of their portfolio. As with netting, this needs to be legally enforceable. Like netting agreements, margin agreements may be two-way, which means that either counterparty is required to post margin against a negative value (from their point of view). One or both counterparties will periodically mark all the relevant positions to market and calculate the net value. They will then check the terms of the underlying agreement to calculate whether they are owed margin and vice versa.

The basic idea of margining is illustrated in Figure 7.3. Parties A and B have one or more transactions between them and therefore agree that one or both of them will exchange margin in order to offset the exposure that will otherwise exist. The rules regarding the timings, amounts, and type of margin posted should naturally be agreed before the initiation of the underlying transaction(s). To keep operational costs under control, posting of margin will not be continuous and will sometimes occur in blocks according to pre-defined rules. In the event of default, the surviving party may retain some or all of the margin to offset losses that they may otherwise incur.

Figure 7.3 Illustration of the basic role of margin.

Note that, since margin agreements are often bilateral, margin must be returned or posted in the opposite direction when exposure decreases. Hence, in the case of a positive valuation, a party will call for margin, and in the case of a negative valuation, they will be required to post margin themselves. Posting margin and returning previously received margin are not materially very different. One exception is that, when returning, a party may ask for specific securities back, whereas when posting outright, optionality may exist.

Margin posted against OTC derivatives positions is, in most cases, under the control of the counterparty and may be liquidated immediately upon an event of default. This arises due to the laws governing derivatives contracts and the nature of the margin (cash or liquid securities). Counterparty risk, in theory, can be completely neutralised as long as a sufficient amount of margin is held against it. However, there are legal obstacles to this and issues such as rehypothecation. Bankruptcies such as Lehman Brothers and MF Global have provided evidence of the risks of rehypothecation (see Section 7.5.4). It is therefore important to note that, whilst margin can be used to reduce counterparty risk, it gives rise to new risks, such as market, operational, and liquidity risk.

Margin also has funding implications. Consider the bank and end user viewpoints described in Section 2.2.5 with margin included (Figure 7.4). An end user will be directional and so when posting margin may have to source the cash or assets to post and may not naturally hold liquid resources that could be used. An end user hedging another financial instrument such as a loan will not receive margin from the loan to offset that required on the derivative.

The above is why some end users have historically been unable or unwilling to post margin. However, this creates an associated problem for a bank, since hedging an uncollateralised transaction with a collateralised one will expose them to asymmetric margin-posting (Figure 7.5). These aspects require consideration of funding and funding value adjustment (FVA) (Chapter 18).

Figure 7.4 Illustration of the classic end user counterparty risk setup shown previously in Figure 2.9 with margin included.

Figure 7.5 Illustration of the classic bank setup shown previously in Figure 2.10 with margin included.

7.2.3 Variation Margin and Initial Margin

There are two fundamentally different types of margin. Primarily, margin would most obviously reflect the valuation of the underlying transactions, which can generally be positive or negative from each party's point of view. This idea forms the basis of ‘variation margin’. Typically, the base valuation (i.e. without xVA, see Section 5.3.6) is typically used for variation margin calculations because it is the most obvious and easy way to define a valuation that can be agreed upon by both parties on a frequent basis. Although this base valuation is not the actual valuation, it has the benefit that it is probably symmetric (i.e. both parties can potentially agree on the valuation) and relatively easy to calculate. Furthermore, since margin has an impact on xVA, any inclusion of xVA in the valuation for margining purposes would create a recursive problem (this is similar to the discussion on close-out and xVA in Section 6.3.4).

In an actual default scenario, the variation margin is likely to be insufficient to cover the costs experienced by the surviving party. This is primarily a result of two factors:

• the inherent delay in the process between the time the party last posted margin and the time they were declared in default; and
• the associated costs in replacing and/or rehedging defaulted transactions.

The above components are often considered together as the margin period of risk (MPoR).

Due to the imperfection of variation margin in relation to the above two components, additional margin is sometimes used in the form of ‘initial margin’. Historically, the term ‘independent amount’ has been used in association with OTC derivatives transactions under ISDA agreements. However, the term initial margin – which originated on exchanges and in central clearing – is becoming more common and will be used here. Figure 7.6 conceptually shows the roles of variation and initial margins.

ISDA (2015) reports that cash makes up around three-quarters of variation margin and just over half of the initial margin, with government securities being the next most common margin type.

Figure 7.6 Illustration of the difference between variation and initial margins as forms of collateralisation. Variation margin aims to track the value of the relevant portfolio through time, whilst initial margin represents an additional amount that may be needed due to delays and close-out costs in the event of a counterparty default.

Historically, the bilateral OTC derivatives market has used margin almost entirely in the form of variation margin, and initial margin (independent amount) has been quite rare. Initial margin is a much more common concept on derivative exchanges and central counterparties (CCPs). However, regulation covering margin-posting in bilateral markets is making initial margin more common (Section 7.4).

7.2.4 Method of Transfer and Remuneration

OTC derivatives margin is typically exchanged by the parties directly and not held by a third-party custodian (although this is changing in part, as discussed in Section 7.4). In practice, there are two different methods of margin transfer:

• Security interest. In this case, the margin does not change hands, but the receiving party acquires an ownership interest in the margin assets and can use them only in the case of certain contractually-defined events (e.g. default). Other than this, the margin-giver generally continues to own the securities. This is typically the case under New York law.
• Title transfer. Here, legal possession of margin changes hands and the underlying margin assets (or cash) are transferred outright in terms of ownership, but with potential restrictions on their usage. Aside from any such restrictions, the margin-holder can generally use the assets freely, and the enforceability is therefore stronger. This ‘outright transfer’ is more common under English law.

ISDA (2015) reports that the New York law pledge agreement is most common, making up 46.8% of surveyed non-cleared margin agreements, and the English law title transfer makes up 30.1%.

The above choices can materially change the parties' rights and obligations and their position in the event of the default of the other party. In general, title transfer is more beneficial for the margin-receiver since they hold the physical assets and are less exposed to any issues such as legal risk. Security interest is preferable for the margin-giver since they still hold the margin assets and are less exposed to problems such as overcollateralisation in the event the margin receiver defaults (where, under title transfer, the additional margin may form part of the bankruptcy estate and not be returned).

The receiver of margin must pass on coupon payments, dividends, and any other cash flows. One exception to this rule is in the case where an immediate margin call would be triggered. In this case, the margin-holder may typically keep the minimum component of the cash flow (e.g. coupon on a bond) in order to remain appropriately collateralised.

The margin agreement will also stipulate the rate of interest to be paid on cash (irrespective of whether title transfer or security interest is used). Interest will typically be paid on cash margin at the overnight indexed spread (OIS) rate, for example, the Euro Overnight Index Average (EONIA) in Europe and Fed Funds in the US). Note that interest is not required on exchanges since the margin represents a settlement. Some counterparties, typically sovereigns or institutional investors, may subtract a spread on cash to discourage receiving cash margin (and encourage securities), since cash must be invested to earn interest or placed back in the banking system. Occasionally, a margin-receiver may agree to pay a rate higher than OIS to compensate for this funding mismatch or to incentivise the posting of cash.

The logic behind using the OIS rate is that since margin may only be held for short periods (due to potentially substantial daily valuation changes), only a short-term interest rate can be paid. However, OIS is not necessarily the most appropriate margin rate, especially for long-dated transactions where margin may need to be posted in substantial amounts for a long period. This may lead to a negative ‘carry’ problem due to an institution funding the margin posted at a rate significantly higher than the OIS rate they receive for posting the collateral. This can be considered to be the source of funding costs (FVA) (Chapter 18).

Where there is optionality in a margin agreement (e.g. the choice of different currencies of cash and/or different securities), this may be quantified via collateral funding adjustment (ColVA) (Chapter 16).

It may be that a party requires or wishes margin securities to be returned. This may be for a variety of reasons, such as:

• needing to deliver securities to a different party under either another margin agreement or repo transaction;⁶
• wanting to hold securities during the period when coupons or dividends are paid or where a party wants to exercise equity voting rights; or
• to maximise collateral optimisation (‘cheapest to deliver’ collateral).

In such a case, it is possible to make a substitution request to exchange an alternative amount of eligible margin (with the relevant haircut applied). The steps for doing this are:

1. the margin giver sends a notice requesting a substitution;
2. (the margin receiver consents to the substitution request);
3. the margin giver provides the new margin assets; and
4. upon receiving the new margin, the margin receiver returns the old margin within the trade settlement date.

The second step above (i.e. consent) may or may not be a requirement depending on the type of transfer being used and the election made by the parties. If consent is not required for the substitution, then such a request cannot be refused⁷ (unless the margin type is not admissible under the agreement), although the requested margin does not need to be released until the alternative margin has been received. Whether or not margin can be substituted freely is an important consideration in terms of the funding costs and benefits of margin and valuing the ‘cheapest to deliver’ optionality inherent in margin agreements. This will be discussed in Section 16.2.3.

Note that substitution of margin does create additional credit risk during the short period when the margin receiver holds both the old and new margin assets. Given that this exists for a short period only, it is similar to settlement risk (Section 3.1.2).

7.2.5 Rehypothecation and Segregation

Rehypothecation and segregation are broadly opposite terms that refer respectively to the reuse and non-reuse of securities or cash margin. Each can be seen to be relevant depending on the nature of the margin, specifically whether it is variation margin or initial margin.

Due to the nature of the OTC derivatives market, where intermediaries such as banks generally hedge transactions, margin reuse can be seen to be natural, as illustrated in Figure 7.7. Note that if the transactions were settled, then the cash amounts would naturally offset. Variation margin can, therefore, be seen as ‘quasi-settlement’.

Whilst cash margin and margin posted under title transfer (Section 7.2.4) are intrinsically reusable, in other situations the margin receiver must have the right of rehypothecation to allow it to be reused. Reuse means that it can be used by the margin receiver (e.g. in another margin agreement or a repo transaction). Rights of rehypothecation are generally used, as seen in Figure 7.8.

Rehypothecation is a natural concept for variation margin since this relates to a change in valuation (i.e. it is an amount that is owed) and is quasi-settlement. Rehypothecation is therefore important in OTC derivatives markets where many parties have multiple hedges and offsetting transactions, and there needs to be a flow of margin through the system. Note that these comments relate to minimising funding costs: from the point of view of funding, rehypothecation is important since it reduces the demand for high-quality collateral.

Figure 7.7 Illustration of the importance of reuse of margin. Party X transacts with counterparty A and hedges this transaction with party B, both under margin agreements. If party A posts margin, then it is natural that this is reused via posting to party B as the hedge will have equal and opposite value.

Figure 7.8 Illustration of rehypothecation of margin (large dealers only).

Source: ISDA (2015).

From the point of view of counterparty risk, rehypothecation is dangerous since it creates the possibility that rehypothecated margin will not be received in a default scenario. However, for variation margin, this impact is minimal because, in the event of a counterparty default, the margin provided can be set off (Section 6.3.6) against the liability⁸ (i.e. what is paid is owed based on the current valuation). There are two ways in which rehypothecation of variation margin does create counterparty risk:

• Counterparty risk is created by margin that needs to be returned against a positive change in value. However, under frequent exchange of margin (e.g. daily) this should be a relatively small problem.
• Margin assets requiring haircuts require overcollateralisation, which also creates counterparty risk due to the extra amount posted. Again, with small haircuts this is a relatively minimal effect.

Rehypothecation of variation margin is therefore not particularly problematic from a counterparty risk perspective since, by definition, it has a close relationship to the amount owed.

In contrast to variation margin, rehypothecation (or more broadly, reuse) is not a natural concept for initial margin. This is due to the fact that initial margin represents extra margin (overcollateralisation). Bankruptcies such as Lehman Brothers and MF Global illustrate the potential problems where rehypothecated assets were not returned. Singh and Aitken (2009) reported a significant drop in rehypothecation in the aftermath of the crisis, which is mainly related to initial margin (independent amount).

Even if margin is not rehypothecated or reused, there is a risk that it may not be retrieved in a default scenario. This risk can be mitigated by segregation, which aims to ensure that margin posted is legally protected in the event that the receiving party becomes insolvent. In practice, this can be achieved either through legal rules that ensure the return of any margin not required (in priority over any bankruptcy rules), or alternatively by a third-party custodian holding the initial margin. Segregation is therefore contrary and incompatible with the practice of rehypothecation. The basic concept of segregation is illustrated in Figure 7.9.

Figure 7.9 Illustration of the concept of segregation. Party A posts margin to party B, which is segregated either legally (priority rules in bankruptcy) and/or operationally (with a third party).

Since initial margin represents extra margin that is not owed, segregation can avoid there being additional counterparty risk in relation to the initial margin posted. Historically, rehypothecation of initial margin and co-mingling of variation and initial margins has created problems for surviving parties in default scenarios such as Lehman Brothers and MF Global. Such practices have, therefore, reduced. For example, hedge funds (as significant posters of initial margin) have become increasingly unwilling to allow rehypothecation. Whilst safer from the point of view of counterparty risk, segregation does create higher funding costs.

There are three potential ways in which segregated margin can be held:

• directly by the margin receiver;
• by a third party acting on behalf of one party; or
• in tri-party custody where a third party holds the margin and has a three-way contract with the two other parties concerned.

It is important to note that there are concepts of ‘legal segregation’ (achieved by all three methods above) and ‘operational segregation’ (achieved only by the latter two). In the MF Global default, parties had legal but not operational segregation, and due to fraudulent behaviour they lost margin that they would have expected to have had returned (see Section 10.3.1). Since cash is fungible by its nature, it is difficult to segregate on the balance sheet of the margin receiver. Hence, a tri-party arrangement where the margin is held in a designated account, and not rehypothecated or reinvested in any way, may be desirable. On the other hand, this limits investment options and makes it difficult for the margin giver to earn any return on their cash. Even if margin is held with a third- or tri-party agent, it is important to consider potential concentration risk with such third parties.

From the point of view of achieving segregation with minimal risks, the following conditions are important:

• the margin is pledged (not title transfer);⁹
• the margin is held with a third-party custodian that is not affiliated with the counterparty; and
• the margin cannot be reused or rehypothecated.

Note that segregation, whilst clearly the optimal method for reducing counterparty risk, causes potential funding issues for replacing margin that would otherwise simply be rehypothecated. This is at the heart of the cost/benefit balance of counterparty risk and funding that will be discussed in more detail in Section 7.6.

The above comments – that rehypothecation is natural for variation margin, whereas initial margin-posting requires segregation – is borne out in regulatory requirements over bilateral margin-posting (discussed later in Section 7.4), which generally require segregation of initial margin but not variation margin.

7.2.6 Settle to Market

As noted in Section 2.2.4, variation margin on exchanges is effectively a settlement, since it is generally required to be paid in cash and with no remuneration. On the other hand, OTC derivatives margin is not traditionally a settlement, which can be seen by the fact that an interest rate is paid to the margin giver on cash and the fact that – bilaterally – it may be possible to post margin in a range of different ‘bespoke’ securities or currencies of cash. This is outlined in Table 7.2.

Note that centrally cleared OTC and collateralised bilateral OTC derivatives are effectively ‘collateralised to market’. In the case of bilateral transactions, this is a result of the potentially bespoke nature of the margin terms (noting that these are becoming more standard, as discussed in Section 4.4.2). In centrally-cleared OTC derivatives, the margining is equivalent to that of an exchange, but an interest rate is paid, typically known as price alignment interest (PAI); this is an interest rate used by CCPs to replicate the economics of bilateral OTC contracts.

Given that CCPs require variation margin (VM) to be in the same currency as the derivative and in cash, this margining process can be considered to be analogous to settlement. For example, the European Banking Authority (EBA 2015c) comments that, ‘As cash for VM is considered the pure settlement of a claim, this should not be subject to any haircut.’ Hence, some centrally-cleared OTC derivatives under variation margining are being classed as settle-to-market (STM).¹⁰ The advantage of this is a potentially more secure legal framework in the event of default and reduced regulatory capital (including leverage ratio) costs since the regulatory maturity of the derivative is effectively shortened to one day or the lowest permitted maturity bucket (discussed later in Section 13.4.3). In an STM contract, interest is still paid (to maintain economic equivalence with bilateral OTC contracts).

Table 7.2 Comparison of the impact of variation margin in various derivatives markets.

	Type	Remuneration	Effect
Exchange	Cash (transaction currency)	-	Settlement
Centrally cleared OTC	Cash (transaction currency)	PAI	Collateralisation
Bilateral OTC	Bespoke	OIS	Collateralisation

7.2.7 Valuation Agent, Disputes, and Reconciliations

The valuation agent is normally the party calling for the delivery or return of margin and thus must handle all calculations. Bilaterally, large counterparties trading with smaller counterparties may be valuation agents for all purposes. Alternatively, both counterparties may be the valuation agent, and each will call for (the return of) margin when they have an exposure (less negative value). In these situations, the potential for margin disputes is significant.

The role of the valuation agent in a margin calculation is as follows:

• to calculate the current value, including the impact of netting;
• to calculate the market value of margin previously posted and adjust this by the relevant haircuts;
• to calculate the total uncollateralised exposure; and
• to calculate the credit support amount (the amount of margin to be posted by either counterparty).

A dispute over a margin call is common and can arise due to one or more of a number of factors:

• trade population;
• trade valuation methodology;
• application of credit support annex rules (e.g. thresholds and eligible collateral);
• market data and market close time; and
• valuation of previously-posted collateral.

Note that for centrally-cleared transactions, margin disputes are not relevant since the CCP is the valuation agent. However, CCPs will clearly aim to ensure that their valuation methodologies and resulting margin requirements are market standard, transparent, and robust.

In bilateral markets, reconciliations aim to minimise the chance of a dispute by agreeing on valuations and margin requirements across portfolios. The frequency of reconciliations is increasing; for example, ISDA (2015) notes that most portfolios – especially large ones – are reconciled on a daily basis. Both Dodd–Frank and the European Market Infrastructure Regulation (EMIR) require more rigorous and frequent portfolio reconciliations, and there may be increased capital requirements for banks as a result of disputes (Section 13.4.5). Margin management practices are being continually improved. One example of this is the increase in electronic messaging in order for margin management to move away from manual processes (ISDA 2015).

In order to reduce operational risks and improve the efficiency of margining, various third parties also offer services around valuation, exchange of margin, dispute management, and portfolio reconciliation.

7.3 MARGIN TERMS

7.3.1 The Credit Support Annex

Traditionally, there has been no obligation in an OTC derivatives contract for either party to post margin. However, within an ISDA Master Agreement (see Section 2.2.6), it is possible to append a credit support annex (CSA) where the parties agree to contractual margin-posting. The CSA has become the market-standard margin agreement in bilateral markets.¹¹ ISDA (2015) reports that 34% of market participants have more than 3,000 margin agreements.¹² As with netting, ISDA has legal opinions throughout a large number of jurisdictions regarding the enforceability of the provisions within a CSA. The CSA will typically cover the same range of transactions as included in the Master Agreement, and it will typically be the net value of these transactions that will form the basis of margin requirements. Exceptions to this include instances where two parties change terms by creating a new CSA only for future transactions, leaving legacy transactions under the old CSA(s). In other words, there can be two CSAs under a single ISDA Master Agreement. This has been the general approach for banks implementing the bilateral margin requirements (Section 7.4).

Within the CSA, two parties can choose a number of key parameters and terms that will define the margin-posting requirements (credit support amount) in detail. These cover many different aspects, such as:

• method and timings of the underlying valuations;
• the calculation of the amount of margin that will be posted;
• the mechanics and timing of margin transfers;
• eligible margin (currencies of cash and types of securities);
• margin substitutions;
• dispute resolution;
• remuneration of margin posted;
• any initial margin amounts;
• haircuts applied to margin securities;
• possible rehypothecation (reuse) of margin securities; and
• triggers that may change the margin conditions (e.g. rating downgrades that may lead to enhanced margin requirements).

One important point with bilateral CSA documentation is that the underlying terms are agreed upfront on signing the CSA and can only be changed by mutual agreement through an amendment to the documentation. In contrast, a CCP can unilaterally change terms in response to market changes (e.g. some CCPs have increased and/or restricted margin requirements in illiquid and volatile markets in recent years – see Section 9.3.6).

When parties have a given CSA in place, the underlying transactions must be valued regularly (typically daily) and the impact on the required margin amount recalculated. Depending on the result of this, one party may be required to post (or return) a specific amount of margin. There need to be procedures in place to deal with disputes over the calculations, and parties may make periodic reconciliations to reduce the risk of disputes.

7.3.2 Types of CSA

The parameters of a CSA are a matter of negotiation between the two parties concerned.¹³ Due to the very different nature of OTC derivatives counterparties, different margin arrangements exist. Some institutions (e.g. corporates) are unable to post margin; this is because, generally, they cannot commit to the resulting operational and liquidity requirements. Some institutions (e.g. sovereign, supranational entities, or multilateral development banks with high credit ratings) receive margin but are unwilling to post, a position partially supported by their exceptional (generally triple-A) credit quality. Non-margin-posting entities generally prefer (or have no choice) to pay xVA charges for the counterparty risk, funding, and capital requirements (KVA) they impose on a bank,¹⁴ rather than agreeing to post margin to mitigate these charges.

Broadly speaking, three possible margin agreements exist in practice:

• No CSA. In some OTC derivatives trading relationships, there is no CSA, since one party cannot or prefers not to commit to margin-posting. A typical example of this is the relationship between a bank and a corporate where the latter's inability to post margin means that a CSA is not usually in place (e.g. a corporate treasury department may find it very difficult to manage its liquidity needs under a CSA).¹⁵
• Two-way CSA. For two financial counterparties, a two-way CSA is more typical, where both parties agree to post margin. Two-way CSAs with zero thresholds are standard in the interbank market and aim to be beneficial (from a counterparty risk point of view, at least) to both parties.
• One-way CSA. In some situations, a one-way CSA is used, where only one party can receive collateral. This actually represents additional risk for the margin giver and puts them in a worse situation than if they were in a no-CSA relationship. A typical example is a high-quality entity such as a triple-A sovereign or supranational trading with a bank. Banks themselves have typically been able to demand one-way CSAs in their favour when transacting with some hedge funds. Note that one-way CSAs are often based on a rating schedule (see Table 7.4) and would be symmetric if the party's ratings were the same. The considerations of funding and capital costs have made such agreements particularly problematic in recent years.

Note that the above are general classifications and are not identified contractually. For example, a one-way CSA would specify a threshold (Section 7.3.4) of infinity, possibly contingent on a strong rating, for the non-posting party. Hence, an endless number of different CSA agreements actually exist based on the terms that will be defined in Sections 7.3.3 to 7.3.5.

Margin-posting across the market is quite mixed depending on the type of institution (Table 7.3). In general, banks will almost always trade under CSAs, financial institutions will often use CSAs, whilst non-financial counterparties will often not trade under a CSA. Note that high-credit-quality sovereigns and similar entities under one-way CSAs will be uncollateralised from a bank's point of view.

Table 7.3 Proportion of each type of counterparty transacting under a CSA.

Source: ISDA (2015).

Institution type	Proportion under a CSA
Dealers	90.4%
Banks and security firms	95.5%
Hedge funds	94.1%
Pension plans	75.3%
Mutual funds	68.8%
Other financial firms	70.4%
Non-financial institutions	28.6%
Government-sponsored entities/ Government agencies	42.4%
Sovereign national governments	69.0%
Local or regional government entities	75.6%

The main reasons for a party not agreeing to post margin are the liquidity needs and operational workload related to posting cash or high-quality securities under stringent margin agreements. Other aspects may include internal or external restrictions (for example, monoline insurers were only able to gain triple-A credit ratings by virtue of not posting margin, as discussed in Section 2.4.4) and the economic view that uncollateralised trading is cheaper than collateralised trading (when liquidity costs are factored in).

A margin agreement must explicitly define all the parameters of the collateralisation and account for all possible scenarios. The choice of parameters will often come down to a balance between the workload of calling and returning margin versus the risk-mitigation benefit of doing so. Funding implications, not historically deemed important, should also be considered. The following sections explain the parameters that define the amount of collateral that must be posted.

7.3.3 Margin Call Frequency

The margin call frequency refers to the contractual periodic timescale with which margin may be called and returned. A longer margin call frequency may be agreed upon, most probably to reduce operational workload and in order for the relevant valuations to be carried out. Some smaller institutions may struggle with the operational and funding requirements in relation to the daily margin calls required by larger counterparties. Whilst a margin call frequency longer than daily might be practical for asset classes and markets that are not so volatile, daily calls have become standard in most OTC derivatives markets. Furthermore, intraday margin calls are common for more vanilla and standard products such as repos and for derivatives cleared via CCPs.

Note that the margin call frequency is not the same as the MPoR (Section 7.2.3). The MPoR represents the effective delay in receiving margin that should be considered in the event of a counterparty default. The margin call frequency is one component of this, but there are other considerations also. It could, therefore, be that the margin call frequency is daily, but the MPoR is assumed to be 10 days (see Section 15.5.1 for more discussion).

7.3.4 Threshold, Initial Margin, and the Minimum Transfer Amount

The threshold is the amount below which margin is not required, leading to undercollateralisation. If the portfolio value is below the threshold, then no margin can be called, and the underlying portfolio is therefore uncollateralised. If the value is above the threshold, only the incremental amount of margin can be called for. Although this clearly limits the risk-reduction benefit of the collateralisation, it does reduce the operational burden and underlying liquidity costs of margin-posting. A threshold of zero means that margin would be posted under any circumstance,¹⁶ and an infinite threshold is used to specify that a counterparty will not post margin under any circumstance (as in a one-way CSA, for example).

Thresholds typically exist because one or both parties can gain in operational and liquidity costs and the associated weakening of the collateralisation is worth this. Some counterparties may be able to tolerate uncollateralised counterparty risk up to a certain level (e.g. banks will be comfortable with this up to the credit limit for the counterparty in question; see Section 3.1.5). Thresholds are increasingly zero and can only be non-zero for parties that are exempt from the bilateral margin requirements (Section 7.4).

A minimum transfer amount (MTA) is the smallest amount of margin that can be transferred. It is used to avoid the workload associated with a frequent transfer of insignificant amounts of (potentially non-cash) margin. The size of the minimum transfer amount again represents a balance between risk mitigation versus operational workload. The minimum transfer amount and threshold are additive in the sense that the exposure must exceed the sum of the two before any margin can be called. We note that this additively does not mean that the minimum transfer amount can be incorporated into the threshold – this would be correct in defining the point at which the margin call can be made, but not in terms of the margin due (more details are given in Section 15.5.3).

There may also be a specified rounding of margin amounts to avoid dealing with awkward quantities. This is typically a relatively small amount and will have a small effect on the impact of collateralisation. Note that minimum transfer amounts and rounding quantities are relevant for non-cash margin where the transfer of small amounts is problematic. In cases where cash-only margin is used (e.g. variation margin and CCP), these terms are generally zero.

Initial margin is the opposite of a threshold and defines an amount of extra margin that must be posted independently from the value of the underlying portfolio. The general aim of this is to provide the added safety of overcollateralisation as a cushion against potential risks such as delays in receiving margin and costs in the close-out process (Section 7.2.3). Initial margin relates to the variability of the value rather than the value itself.

As noted above, the initial margin in bilateral OTC derivatives (known as the independent amount) was historically quite rare and was a payment in only one direction (usually from a weaker-credit-quality party to a stronger-credit-quality party). For exchange-traded derivatives and centrally-cleared OTC derivatives, initial margin is always required. However, the bilateral margin requirements (Section 7.4) are requiring that initial margin be posted in bilateral OTC derivatives also, irrespective of credit quality. Initial margin is, therefore, becoming a common feature of derivatives transactions, with the exception of bilateral OTC derivatives where one party (or the product) is exempt from the bilateral margin requirements.

Unlike parameters such as thresholds which are fixed quantities, initial margin amounts typically change over time according to the perceived risk on the portfolio. Sometimes these are simple methodologies (e.g. based on notional, transaction type, and remaining maturity), but more commonly they are becoming quite complex, dynamic calculations. This is important to emphasise as the term initial margin may suggest a static quantity (this was the case when initial margin was first used, but it is not so any more). Dynamic initial margin methodologies are the subject of Chapter 9.

Note that since thresholds and initial margins essentially work in opposite directions, and an initial margin can be thought of (intuitively and mathematically) as a negative threshold, these terms are not seen together: either undercollateralisation is specified via a threshold (with zero initial margin), or an initial margin defines overcollateralisation (with a threshold of zero).

Historically, bilateral OTC derivative markets have sometimes also linked thresholds and initial margin requirements (and sometimes minimum transfer amounts) to credit quality (most commonly credit ratings).¹⁷ An example of this is given in Table 7.4. The motivation for doing this is to minimise the operational workload when a counterparty's rating is high (the threshold is infinity at a rating of AAA), but to tighten the collateralisation terms when their credit quality deteriorates (at A+ the threshold is zero, and below this an initial margin is also required).

Prior to the GFC, triple-A entities such as monoline insurers traded through one-way margin agreements (i.e. they did not post margin), but with triggers specifying that they must post if their ratings were to decline. Such agreements can lead to rather unpleasant effects, since a downgrade of a counterparty's credit rating can occur rather late with respect to the actual decline in credit quality, which in turn may cause further credit issues due to the requirement to post collateral. This is exactly what happened with AIG (see below) and monoline insurers (Section 2.4.4) in the GFC.

Table 7.4 Example of rating-linked margin parameters. This could correspond to one or both parties.

Rating	Initial margina	Threshold
AAA/Aaa	0%	∞
AA+/Aa1	0%	$100m
AA/Aa2	0%	$50m
AA-/Aa3	0%	$25m
A+/A1	0%	$0
A/A2	1%	$0
A-/A3	1%	$0
BBB+/Baa1	2%	$0

a Expressed in terms of the total notional of the portfolio.

---

The dangers of credit rating triggers

The case of American International Group (AIG) is probably the best example of the funding liquidity problems that can be induced by margin posting. In September 2008, AIG was essentially insolvent due to the margin requirements arising from credit default swap transactions executed by their financial products subsidiary AIGFP. In this example, one of the key aspects was that AIGFP posted margin as a function of their credit rating. The liquidity problems of AIG stemmed from the requirement to post an additional $20bn of margin as a result of its bonds being downgraded.¹⁸ Due to the systemic importance of AIG, the Federal Reserve Bank created a secured credit facility of up to $85bn to allow AIGFP to post the margin they owed and avoid the collapse of AIG.

---

These two problems of rating linkage are now quite characterised:

• from a risk-mitigating point of view, they may be ineffectual due to the rating reacting too slowly; and
• they may create liquidity problems and cliff-edge effects for a counterparty being downgraded.

The above asymmetry is recognised by regulation. For example, Basel capital rules do not allow a capital benefit from credit rating triggers (see BCBS 2011b) and so assume – conservatively – that a counterparty will default without first being downgraded. On the other hand, the liquidity coverage ratio rules (Section 4.3.3) require a bank to pre-fund outflows such as ratings-contingent initial margin-posting. For example, based on Table 7.4, a bank rated A+ would have to fund 2% of initial margin based on a three-notch downgrade to BBB+.

7.3.5 Margin Types and Haircuts

An agreement such as a CSA specifies the assets that are admissible to be posted as margin. Cash (mostly in US dollars and Euros) is the major form of margin taken against OTC derivatives exposures (Figure 7.10). The ability to post other forms of margin is often highly preferable for liquidity reasons. Cash margin has become increasingly common over recent years – a trend which is unlikely to reverse, especially due to cash variation margin requirements in situations such as central clearing. Government securities comprise a further 14.8% of total margin, with the remaining 10.3% comprising government agency securities, supranational bonds, US municipal bonds, covered bonds, corporate bonds, letters of credit, and equities.

To the extent that there is price variability of such assets, a ‘haircut’ acts as a discount to the value of the margin and provides a cushion in the event that the security is sold at a lower price. The haircut will depend on individual characteristics of the asset in question. Haircuts mean that extra margin must be posted, which acts as an overcollateralisation to mitigate against a drop in the value of the asset, so that lenders will have more chance of being fully collateralised.

Figure 7.10 Breakdown of the type of margin received against non-cleared OTC derivatives.

Source: ISDA (2014a).

Cash is the most common type of margin posted and may not attract haircuts. However, there may be haircuts applied to reflect the FX risk from receiving the margin in a currency other than the ‘termination’ currency of the portfolio in question.¹⁹ Bonds posted as margin will have price volatility arising from interest rate and credit spread moves, although these will be relatively moderate (typically a few percent). Margin based on equity or commodity (e.g. gold) underlyings will clearly require much higher haircuts due to high price variability.

Haircuts are typically defined as fixed percentage amounts, although they can potentially be based on more dynamic amounts derived from a model. In bilateral markets they are specified in the agreement and cannot, therefore, be adjusted in line with changes in the market (unless both parties agree). A CCP is typically able to change contractual haircuts and eligible securities and will do so if market conditions change (e.g. a particular asset is seen to have greater market volatility, liquidity risk, or credit risk).

The haircut is a reduction in the value of the asset to account for the fact that its value may fall between the last margin call and liquidation in the event of the counterparty's default. As such, the fixed haircut is theoretically driven by the volatility of the asset and its liquidity. A haircut of means that for every unit of that security posted as collateral, a ‘valuation percentage’ of only will be given, as illustrated in Figure 7.11. The margin giver must account for the haircut when posting.

Volatile assets such as equities or gold are not necessarily problematic, as relatively large haircuts can be taken as compensation for their price volatility and potential illiquidity. Assets with significant credit or liquidity risk are generally avoided as haircuts cannot practically be large enough to cover the default of the margin asset or having to liquidate it at a substantially reduced price. Some examples of haircuts, together with eligible margin types, are shown in Table 7.5.

Securities in various currencies may be specified as admissible margin but may also attract haircuts due to the additional FX risk. FX risk from posted margin can be hedged in the spot and forward FX markets, but it must be done dynamically as the value of the margin changes. If the credit rating of an underlying security held as margin declines below that specified in the margin agreement, then it will normally be necessary to replace this security immediately.

Figure 7.11 Illustration of a haircut applied to collateral.

Table 7.5 Example haircuts in a margin agreement.

	Party A	Party B	Valuation percentage	Haircut
Cash in eligible currency	X	X	100%	0%
Debt obligations issued by the governments of the US, UK or Germany with a maturity less than one year	X	X	98%	2%
Debt obligations issued by the governments of the US, UK or Germany with a maturity between 1 and 10 years	X	X	95%	5%
Debt obligations issued by the governments of the US, UK or Germany with a maturity greater than 10 years	X		90%	10%

The important points to consider in determining eligible margin and assigning haircuts are:

• time taken to liquidate the margin;
• the volatility of the underlying market variable(s) defining the value of the margin;
• default risk of the asset;
• maturity of the asset;
• liquidity of the asset; and
• any relationship between the value of the margin and either the default of the counterparty or the underlying exposure (wrong-way risk).

The last point above is often the hardest to implement in a prescriptive fashion. For example, high-quality (i.e. above some credit rating) sovereign bonds are likely to be deemed as eligible collateral. They will likely have good liquidity, low default risk, and reasonably low price volatility (depending on their maturity). However, the CSA may not prevent (for example) a bank posting bonds from its own sovereign.

An obvious way in which to derive a haircut would be to require that it covers a fairly severe potential worst-case drop in the value of the underlying asset during a given time period, as illustrated in Figure 7.12. The time period would depend on the liquidity of the underlying collateral, but would typically be in the region of a few days.

Figure 7.12 Illustration of the methodology for estimating a haircut.

Under normal distribution assumptions, such a formula is easy to derive:

(7.1)

where defines the number of standard deviations the haircut needs to cover involving the cumulative inverse normal distribution function and the confidence level (e.g. 99%). Also required is the volatility of the collateral, , and the liquidation time (similar to the MPoR concept, but typically shorter), denoted . For example, assuming a 99% confidence level, two-day time horizon, and an interest rate volatility of 1%, a 10-year, high-quality (i.e. minimal credit risk) government bond would have an approximate haircut of 2%,²⁰ which can be compared with the bilateral margin rules discussed in Section 7.4.4.²¹

Haircuts applied in this way should not only compensate for collateral volatility but also reduce exposure overall, since they should create an overcollateralisation in 99% of cases, and they will only be insufficient 1% of the time. Indeed, for a short MPoR the use of even quite volatile non-cash collateral does not increase the exposure materially, and a reasonably conservative haircut will at least neutralise the additional volatility. The major issue that can arise with respect to collateral type is where there is a significant relationship between the value of the collateral and the exposure or credit quality of the counterparty. A sovereign entity posting its own bonds provides an example of this.²² Note that adverse correlations can also be present with cash margin: an example would be receiving euros from European sovereigns or European banks. This will be described as ‘wrong-way collateral’ and discussed further in Section 7.5.3.

7.3.6 Credit Support Amount Calculations

ISDA CSA documentation defines the credit support amount as the amount of margin that may be requested at a given point in time. Due to parameters such as thresholds, minimum transfer amounts, and initial margins, this will not be the same as the value of the portfolio.

---

Spreadsheet 7.1 Margin calculation including thresholds and initial margins.

---

Consider initially that a party is only receiving margin. The relevant amount of margin considering only the threshold and initial margin will be:

(7.2)

where represents the current value of the relevant transactions²³ and and represent the threshold and initial margin for the counterparty respectively (note that the threshold is generally a fixed value, but the initial margin may be dynamic). The above equation shows that the threshold and initial margin act in opposite directions, and it does not make sense to include both. Indeed, when initial margins are present, the threshold will usually be zero.

More generally, including both parties, the equation defining the credit support amount for variation margin is:

(7.3)

where is the threshold for the party making the calculation and represents the amount of margin held already (known as the credit support balance). If the above calculation results in a positive value, then margin can be called (or requested to be returned), whilst a negative value indicates the requirement to post (or return) margin. This is the case as long as the amount is higher than the minimum transfer amount (rounding may also be applied).

Initial margins do not depend on any of the above terms and so can be considered to be computed separately. Traditionally, variation margin is captured in the CSA, as above, by the credit support amount, and the initial margin is referred to separately as independent amount. Initial margin is not netted against variation margin amounts and may also be segregated. In the case where both parties post initial margin (not common historically but required under future regulation), the initial margins themselves are not netted and are paid separately.

Note also that margin payments in OTC derivatives are not netted against cash flow payments, even if they are in the same denomination as the cash flows. This is due to the fact that margin is not a settlement. In exchange-traded markets, the settlement process means that such netting effectively does occur. In bilateral markets, this gives rise to so-called ‘collateral spikes’, where a large cash flow can cause a spike in the risk due to the delay in receiving margin against the change in valuation resulting from the payment (see Figure 3.2).

Table 7.6 illustrates the potential movement in the value of a portfolio and the associated margin requirements (without considering minimum transfer amounts and rounding). Important features are:

• Time . The credit support amount is defined by the value minus the threshold amount, which means that the portfolio is only partially collateralised.
• Time . Even though the portfolio is uncollateralised by 75, the credit support amount is -25, which corresponds to margin being returned (up to the threshold).

Note that minimum transfer amounts (MTAs) can cause path dependency in margin calculations, as illustrated in Table 7.7. Here, the values at times and in all scenarios are the same, but those at time are not. The result of this is that the credit support amount at time differs and can either be smaller (scenario 1) or larger (scenario 3). The amount of credit support that is required at depends on the credit support balance at , which in turn depends on the previous time . This shows that MTAs can cause path dependency, which can make the modelling of collateralised exposures more complicated (Section 15.5.3).

Table 7.8 illustrates the potential movement in the value of a portfolio with a zero threshold and initial margin. Some other important features are:

• Time . The portfolio is overcollateralised due to the initial margin.
• Time . The move in the value of the portfolio results in a margin deficit of 25 (375 - 350), but this is within the initial margin amount (in the event of default, there is the question of whether the close-out costs would be within the initial margin). The reduction of the initial margin will not be netted with the increase in variation margin.²⁴

Table 7.6 Example margin calculation. No initial margin.

Value	350	325
Threshold	100	100
Credit support balance	0	250
Credit support amount	250	−25

Table 7.7 Example margin calculation with minimum transfer amount (assumed to be 50) and a threshold of zero and no initial margin.

Scenario 1
Value	90	110	150
Credit support balance	0	90	90
Credit support amount	90	0	60
Scenario 2
Value	40	110	150
Credit support balance	0	0	110
Credit support amount	0	110	0
Scenario 3
Value	130	110	150
Credit support balance	0	130	130
Credit support amount	130	0	0

Table 7.8 Example margin calculation with (dynamic) initial margin.

Value	350	375
Threshold	0	0
Credit support balance	0	350
Credit support amount	350	25
Initial margin (independent amount)	50	45

7.3.7 Impact of Margin on Exposure

The impact of margin on a typical exposure profile is shown in Figure 7.13. There are essentially two reasons why margin cannot perfectly mitigate exposure. Firstly, the presence of a threshold means that a certain amount of exposure cannot be collateralised.²⁵ Secondly, the delay in receiving margin and parameters such as the minimum transfer amount create a discrete effect, as the movement of exposure cannot be tracked perfectly (this is illustrated by the grey blocks in Figure 7.13).²⁶

Initial margin can be thought of as making the threshold negative and can, therefore, potentially reduce the exposure to zero. This is illustrated in Figure 7.14.

Margin usage has increased significantly over the last decade, as illustrated in Figure 7.15, which shows the estimated amount of margin and gross credit exposure. The ratio of these quantities gives an estimate of the fraction of credit exposure that is collateralised. The overall figure is slightly misleading since it is a combination of portfolios that may either be uncollateralised (e.g. no CSA or one-way CSA), collateralised (e.g. traditional two-way CSA) or overcollateralised (i.e. including initial margin). The incoming regulatory rules on bilateral margin will increase the trend by requiring greater amounts of both variation and initial margin.

Figure 7.13 Illustration of the impact of margin on exposure without initial margin. The margin amount is depicted by grey areas.

Figure 7.14 Illustration of the impact of margin on exposure, including initial margin.

Figure 7.15 Illustration of the amount of margin for non-cleared OTC derivatives

(Source: ISDA 2014a) compared to the gross credit exposure (Source: BIS 2014) and the ratio giving the overall extent of collateralising of OTC derivatives. Note that the margin numbers are halved to account for double-counting, as discussed in ISDA (2014a).

7.3.8 Traditional Margin Practices in Bilateral and Centrally-cleared Markets

In bilateral OTC derivatives, margin agreements such as CSAs are not always used, and when they are relatively bespoke, even to the extent that only one party may post margin (one-way CSA). This means that agreements are non-standard and there is a significant amount of optionality (e.g. there are many possibilities about the type of margin that can be delivered across currency, asset class, and maturity). Furthermore, initial margin (independent amount) has traditionally been rare.

On the other hand, centrally-cleared OTC derivatives are subject to much more standard margining methods that have developed over many years of central clearing of exchange-traded derivatives. Such margining involves frequent paying (receiving) of variation margin against losses (gains) in cash in the currency of the transaction and added security for the CCP in the form of initial margin. This prevents the CCP from having any FX or liquidity risk (e.g. margin in different currencies) and minimises their counterparty risk exposure.

Table 7.9 contrasts traditional bilateral and centrally-cleared OTC derivative margining practices.

There have been initiatives to attempt to standardise bilateral margining practices. Notably, the ISDA standard credit support annex (SCSA) aimed to standardise and reduce embedded optionality in CSAs, whilst promoting the adoption of standard pricing (e.g. what is now typically known as OIS discounting, discussed in Section 16.1.3). An SCSA could also be seen as changing the mechanics of the SCSA to be more closely aligned to central clearing margin practices.

In a typical CSA, a single amount is calculated at each period for a portfolio, which may cover many currencies. Cash margin may, therefore, be posted in different currencies and also typically in other securities. In addition, thresholds and minimum transfer amounts are commonly not zero. The aim of the SCSA was to greatly simplify the process, requiring:

Table 7.9 Comparison of historical margining practices in bilateral and centrally-cleared OTC derivatives markets.

	Bilateral	Centrally cleared
Thresholds	Often non-zero	Zero
Initial margin	Rare (independent amount)	Must be posted by all parties (except CCP itself)
Type (variation margin)	Relatively flexible	Cash in transaction currency
Type (initial margin)		Cash and liquid securities
Frequency of posting	Daily or less frequently	Daily and potentially intradaily27
Negotiation	Bilateral	Defined by CCP

• cash margin only (with respect to variation margin, any initial margins will be allowed in other securities);
• only currencies with the most liquid OIS curves (USD, EUR, GBP, CHF, and JPY) to be eligible;
• zero thresholds and minimum transfer amounts; and
• one margin requirement per currency (cross-currency products are put into the USD bucket).

The SCSA did not gain in popularity due largely to the currency silo issue where each currency gives rise to a margin requirement in cash. It has been largely superseded by regulatory rules on bilateral margin-posting, which are discussed next.

7.4 BILATERAL MARGIN REQUIREMENTS

7.4.1 General Requirements

As discussed in Section 4.4.2, regulators have started to impose bilateral margin requirements, often known as the uncleared margin requirements (UMR), on most major participants when transacting non-centrally-cleared derivatives with one another. The final rules can be found in BCBS-IOSCO (2019). In general, these rules have the following aims:

• Reduction of systemic risk. Margin requirements are viewed as reducing ‘contagion and spillover effects’ by ensuring that margin is available to cover losses caused by the default of a derivatives counterparty.
• Promotion of central clearing. Whilst central clearing is often mandatory, the UMR will promote the adoption of central clearing by aligning costs (notably initial margin) in a bilateral market with those that will exist in centrally-cleared markets.

The concepts of variation and initial margin are intended to reflect current and potential future exposure, respectively.

The above rules apply to ‘covered entities’ in transactions with one another (i.e. if one party in a trading relationship is not a covered entity, then the other party does not need to comply with respect to this relationship). Parties that are exempt from the requirements include sovereigns, central banks, multilateral development banks, and the Bank for International Settlements. Other exemptions depend on the region in question (see Section 4.4.3 for more detail).

With respect to each component, standards state that covered entities for non-centrally-cleared derivatives must exchange:

• Variation margin:
  • Must be exchanged bilaterally on a regular basis (e.g. daily).
  • Full margin must be used (i.e. zero threshold).
  • The minimum transfer amount must not exceed €500,000.²⁸
  • Can be rehypothecated and netted.
  • Must be posted in full from the start of the rules.
• Initial margin:
  • To be exchanged by both parties with no netting of amounts.
  • Should be based on an extreme but plausible move in the underlying portfolio value at a 99% confidence level.
  • A 10-day time horizon should be assumed on top of the daily variation margin exchanged (this can be seen to align loosely with the MPoR, discussed in Section 7.2.3).
  • Can be calculated based on internal (validated) models or regulatory tables.
  • Must be exchanged on a gross basis (i.e. amounts posted between two parties cannot cancel), must be segregated and cannot be rehypothecated, repledged, or reused.²⁹
  • Follows a phased-in implementation (see Section 7.4.2).

Parties are required to meet strict delivery timing requirements for margin, in most cases requiring it to be provided within the same business day as the date of the calculation.³⁰

The collecting counterparty must segregate the initial margin either with a third-party holder or custodian or via other legally-binding arrangements so that it is protected from the default or insolvency of the party holding the initial margin. The initial margin must be available to the posting entity in a timely manner in case the collecting party defaults. In addition, the collecting counterparty should provide the posting counterparty with the option to segregate its collateral from the assets of the other counterparties. Counterparties will need to arrange for initial margin to be provided by way of security rather than title transfer (Section 7.2.4). The collecting counterparty must not rehypothecate, repledge, or reuse the collateral collected as initial margin, although variation margin may be reused by the collecting counterparty. This is all consistent with the aim that posting initial margin – being extra margin – should not increase counterparty risk (Section 7.2.5). Firms will, therefore, have to open accounts and build connectivity with custodians.

The use of a third party to achieve segregation of initial margin is viewed as the most robust protection, although this does raise the issue of whether such entities (the number of which is currently quite small) would become sources of systemic risk with the amount of margin they would need to hold. Arrangements for segregation will vary across jurisdictions depending on the local bankruptcy regime, and would need to be effective under the relevant laws and supported by periodically updated legal opinions.

Rigorous and robust dispute resolution procedures should be in place in case of disagreements over margin amounts. This is an important point since risk-sensitive initial margin methodologies will be, by their nature, quite complex and likely to lead to disputes.

7.4.2 Phase-in and Coverage

The UMR apply to all OTC derivatives with the exception of FX swaps and forwards, which are exempt.³¹ This, like a similar exemption over clearing, is controversial (Duffie 2011). On the one hand, such FX products are often short-dated and more prone to settlement risk than counterparty risk. On the other hand, FX rates can be quite volatile and occasionally linked to sovereign risk, and cross-currency swaps are typically long-dated.

Official implementation dates are shown in Table 7.10, although such timescales depend on the implementation of local regulations, which have been delayed in some cases.³² Note that there have been changes to Phase 5 and a new Phase 6 which were not in the original regulations (see BCBS-IOSCO 2019). See ISDA-SIFMA (2018) for details of thresholds in other currencies and details on implementation in other regions.

Whereas variation margins must be posted immediately, initial margin requirements are subject to a phase-in with declining thresholds until 1 September 2021 and need only apply to new transactions executed after the relevant date (applying the initial margin requirements to existing derivatives contracts is not required). Furthermore, parties can have a threshold of up to €50m (based on a consolidated group of entities)³³ in relation to their initial margin-posting (i.e. they would only post amounts above the threshold).³⁴ This threshold was introduced to reduce liquidity costs as a result of the requirements.³⁵

A particularly strange case for bilateral margining is a cross-currency swap. Such products have significant counterparty risk due to the large interest rate and FX volatilities and the fact that they are typically long dated. Due to the FX exemptions, the FX component of a cross-currency swap is exempt from initial margin-posting.³⁶ In practice, this means that there will still be material counterparty risk on such a transaction since the FX component drives the majority of the exposure. On the other hand, this component is not exempt from variation margin posting requirements.

Note that the requirements allow a party to remain exempt if they have a total notional of less than €8bn of OTC derivatives or can stay below a negotiated threshold of up to €50m with each of their counterparties.

Table 7.10 Timescales for the implementation of margin requirements for covered entities and transactions based on the aggregate group-wide, month-end average notional amount of non-centrally-cleared derivatives (including physically-settled FX forwards and swaps), newly executed during the immediately preceding June, July, and August.

Date	Requirement
Variation margin
1 September 2016	Exchange variation margin with respect to new non-centrally-cleared derivative transactions if average aggregate notionals exceed €3trn for both parties.
1 March 2017	As above but with no threshold.
Initial margin
1 September 2016 to 31 August 2017 (‘Phase 1’)	Exchange initial margin if average aggregate notionals exceed €3trn.
1 September 2017 to 31 August 2018 (‘Phase 2’)	Exchange initial margin if average aggregate notionals exceed €2.25trn.
1 September 2018 to 31 August 2019 (‘Phase 3’)	Exchange initial margin if average aggregate notionals exceed €1.5trn.
1 September 2019 to 31 August 2020 (‘Phase 4’)	Exchange initial margin if average aggregate notionals exceed €0.75trn.
1 September 2020 to 31 August 2021 (‘Phase 5’)	Exchange initial margin if average aggregate notionals exceed €50bn.
From 1 September 2021 (‘Phase 6’)	Exchange initial margin if average aggregate notionals exceed €8bn.

7.4.3 Initial Margin and Haircut Calculations

As noted above, initial margin is intended to cover the potential future exposure in an extreme but plausible scenario based on a 99% confidence interval over a 10-day horizon. Such a scenario must use historical data that incorporates a period of significant financial stress. The requirement for the stress period is to ensure that a sufficiently large amount of margin is available and also to reduce procyclicality (positively correlated to the general state of the economy; see Section 9.3.5) in the margin over time.

In general, the above implies that initial margin amounts will be portfolio-based and also dynamic (i.e. they will change over time). This is in contrast to the historical use of initial margin (independent amount), which is often transaction-based and either deterministic or driven by simple formulas (e.g. a percentage of notional based on the type and remaining maturity of a transaction).

There are two methods that can be used to calculate the required amount of initial margin:

• a quantitative portfolio margin model, either an entity's own or third-party (that must be validated by the relevant supervisory body); or
• a standardised margin schedule.

There can be no cherry-picking by mixing these approaches based on which gives the lowest requirement in a given situation (although it is presumably possible to choose different approaches for different asset classes, as long as there is no switching between these approaches).³⁷

The standardised initial margin schedule is shown in Table 7.11. The quantities shown should be used to calculate gross initial margin requirements by multiplying by the notional amount of each transaction and summing across all of them.

To account for portfolio effects across transactions, the following formula is used:

(7.4)

where NGR is defined as the level of the net replacement cost of the portfolio (i.e. including legally-enforceable netting agreements) over the level of gross replacement cost (ignoring netting). NGR gives a very simple representation of the future offset between positions, the logic being that 60% of the current impact of netting can be assumed for the potential future exposure. Whilst the above schedule-based approach is simplistic, there is still the question of how to treat more complex transactions which do not have easily-defined notionals (e.g. variance swaps) or maturity dates (e.g. callable or extendable structures). These will generally also have to be treated with the most realistic, conservative assumptions. The margin schedule will, therefore, likely give fairly conservative results and represent portfolio diversification badly. That said, even using a portfolio margin model, it is not permissible to benefit from potentially low historical correlations between risk factors between asset classes. The relevant asset classes – which must be treated on an additive basis – are defined as:

• currency/rates;
• equity;
• credit; and
• commodities.

Since initial margin calculations need to be agreed across many pairs of counterparties, there is clearly a need for a standard approach. This has been achieved by the ISDA SIMM (Standard Initial Margin Model), which will be described in more detail in Section 9.4.4.

Table 7.11 Standardised initial margin schedule as defined by BCBS-IOSCO (2015).

%	0–2 years	2–5 years	5+ years
Interest rate	1	2	4
Credit	2	5	10
Commodity	15
Equity	15
Foreign exchange	6
Other	15

7.4.4 Eligible Assets and Haircuts

Regarding the quality of initial margin, the margin should be ‘highly liquid’ and, in particular, should hold its value in a stressed market (accounting for the haircut). Risk-sensitive haircuts should be applied, and margin should not be exposed to excessive credit, market, or FX risk (including through differences between the currency of the collateral asset and the currency of settlement). Margin must not be ‘wrong-way’, meaning correlated to the default of the counterparty (e.g. a counterparty posting its own bonds or equity).

Examples of satisfactory margin are given as:

• cash;
• high-quality government and central bank securities;
• high-quality corporate/covered bonds;
• equity in major stock indices; and
• gold.

Note that US regulations limit eligible margin for variation margin to cash (in USD, another major currency, or the ‘settlement currency’), but for financial end users the same assets as permitted for initial margin are permissible.

As with initial margin, haircuts can either be based on a quantitative model or on a standard schedule (Table 7.12).

Regarding the FX component, no haircut is applied to cash variation margin, even where the payment is executed in a different currency than the currency of the contract. However, there is an 8% haircut for non-cash variation margin denominated in a currency other than the settlement currency or initial margin (cash and non-cash) under similar conditions.

As in the case of initial margin models, approved risk-sensitive quantitative models can be used for establishing haircuts, so long as the model for doing this meets regulatory approval. BCBS-IOSCO (2015) defines that haircut levels should be risk sensitive and reflect the underlying market, liquidity, and credit risks that affect the value of eligible margin in both normal and stressed market conditions. As with initial margins, haircuts should be set in order to mitigate procyclicality and avoid sharp and sudden increases in times of stress. The time horizon and confidence level for computing haircuts is not defined explicitly, but could be argued to be less (say 2–3 days) than the horizon for initial margin, since the margin may be liquidated independently and more quickly than the portfolio would be closed out.

Table 7.12 Standardised haircut schedule, as defined by BCBS-IOSCO (2015). Note that the FX add-on corresponds to cases where the currency of the derivative differs from that of the margin asset.

%	0–1 years	1–5 years	5+ years
High-quality government and central bank securities	0.5	2	4
High-quality corporate/covered bonds	1	4	8
Equity/gold	15
Cash (in the same currency)	0
FX add-on for different currencies	8

7.4.5 Implementation and Impact of the Requirements

Since the above requirements are phased in and – in the case of initial margin – impact only new transactions after both parties fall in scope, there is a phasing-in effect. Of course, it is possible for parties to agree to cover all transactions, not only those in scope. ISDA has published a Variation Margin Protocol to enable market participants to put in place documentation on a standardised basis with multiple counterparties, reducing the need for bilateral negotiations. The protocol allows parties either to amend their existing credit support documents or to enter into new credit support documents, in a way which is compliant with the regulatory margin requirements. The protocol allows the following possible methods to be used (which will only take effect between two adhering parties if they agree to the method):

• Amend method. Terms in existing CSAs are amended as necessary to comply with the regulatory requirements of the relevant jurisdictions. Both legacy trades and new trades are covered under the amended CSA.
• Replicate-and-amend. A new CSA is created with the existing CSA remaining in place for legacy transactions. The new CSA is then amended to comply with the regulatory requirements.
• New CSA. Parties enter into a new CSA with standard terms and certain optional terms that are agreed bilaterally. This method is useful for when parties do not have an existing CSA in place.

Many covered counterparties are agreeing on new CSAs that incorporate the rules to cover future transactions but that can keep old transactions under existing CSAs. New or modified CSAs will need to consider the following aspects:

• thresholds and minimum transfer amounts;
• margin eligibility;
• haircuts;
• calculations, timings, and deliveries;
• dispute resolution; and
• initial margin calculations and mechanisms for segregation.

ISDA (2018) has surveyed the impact of the margin requirements for Phase 1 firms (i.e. those that are impacted by the €3trn threshold in Table 7.10).³⁸ They found that a total of around $50bn of regulatory initial margin had been posted (and received)³⁹ against non-cleared derivatives. They also noted that these firms had delivered about $16bn and received about $61bn of discretionary initial margin as of 31 March 2017.⁴⁰ Most variation margin was posted in cash, whilst government securities were the most popular form of initial margin.

Clearly, the total amount of initial margin posted over time will increase as more firms become in scope. In addition to the operational and legal hurdles to comply with the regulation, there will be a growing need to understand:

• the methodologies (such as ISDA SIMM) for calculating initial margin (Section 9.4.4);
• the future costs of initial margin via margin value adjustment (Chapter 20); and
• the capital relief achievable when receiving initial margin (Section 13.4).

7.5 IMPACT OF MARGIN

Whilst margining is a useful mechanism for reducing counterparty risk, it has significant limitations that must be considered. It is also important to emphasise that margin, like netting, does not reduce risk overall but merely redistributes it. Margin also creates other residual risks, such as operational, legal, and liquidity risks. Other potential issues include wrong-way risk (where margin is adversely correlated to the underlying exposure), credit risk (where the margin securities may suffer from default or other adverse credit effects), and FX risk (due to margin being posted in a different currency). Some of these issues are outlined below. The broader issue of the link between margin and funding is discussed in Section 7.6.

7.5.1 Impact on Other Creditors

Risk mitigants such as margin are viewed narrowly only in terms of their impact on reducing exposure to a defaulted counterparty. However, more precisely, what actually happens is a redistribution of risk, where some creditors (e.g. derivatives creditors) are paid more in a default scenario at the expense of other creditors. Figure 7.16 shows the impact of the posting of margin against a derivative transaction. Assume that in default of party B, party A and the other creditors (OC) of B have the same seniority of the claim (pari passu). Party B owes derivatives creditors 50 and other creditors 100, and has assets of 100.

With respect to the amount of margin posted in Figure 7.16, it is useful to consider the following three cases:

• No margin. In the no-margin case, the other creditors will have a claim on two-thirds (100 divided by 150) of the assets of B, with the derivative claims of A receiving the remaining one-third. The derivatives and other creditors will both recover 67% of their claims.
• Variation margin. If party B posts 50 variation margin to A against their full derivative liability, then this will reduce the value received by the OCs in default. Now the remaining assets of B in default will be only 50, to be paid to the other creditors (recovery 50%). OTC derivatives creditors will receive 100% of their claim (ignoring close-out costs).
• Initial margin. Suppose that B pays 50 variation margin and 25 initial margin and the entire initial margin is used by A in the close-out and replacement costs of their transactions with B. In such a case, the OCs would receive only the remaining 25 (recovery 25%). It could be argued that some or all of the initial margin may need to be returned, potentially related to litigation (see Section 6.3.5), but a significant portion may be lost in close-out costs.

Figure 7.16 Example of the impact of derivatives margin on other creditors (OC). The margin posted (variation and possibly also initial margin) will reduce the claims of the other creditors.

Margin does not reduce risk, it merely redistributes it (although possibly in a beneficial way). Other creditors will be more exposed, leading to an increase in risk in other markets (e.g. loan market as opposed to derivatives market). Furthermore, the other creditors will react to their loss of seniority, for example, by charging more when lending money.

7.5.2 Market Risk and Margin Period of Risk

Margin can never completely eradicate counterparty risk, and it is important to consider the residual risk that remains under the margin agreement. Residual risk can exist due to contractual parameters such as thresholds and minimum transfer amounts that effectively delay the margin process. Thresholds and minimum transfer amounts can – and are increasingly being – set to zero (or very small) values. Frequent contractual margin calls obviously maximise the risk-reduction benefit at the expense of increasing operational and liquidity costs. Daily adjustment of margins is becoming increasingly standard in derivatives markets, although longer periods do sometimes exist in certain trading situations and products.

Even if margin is exchanged frequently and with zero thresholds, another important aspect is the inherent delay in receiving this collateral. This is market risk as it is defined by market movements after the counterparty last successfully posted margin. The MPoR is the term used to refer to the effective time between a counterparty ceasing to post margin and the time when all the underlying transactions have been successfully closed out and replaced (or otherwise hedged), as illustrated in Figure 7.17. Such a period is crucial since it defines the effective length of time without receiving margin where any increase in exposure (including close-out costs) will remain uncollateralised. Note that the MPoR is a counterparty-risk-specific concept (since it is related to default) and is not relevant when assessing funding costs (discussed in more detail in Chapter 18).

Figure 7.17 Illustration of the role of the MPoR.

In general, for bilateral relationships (the MPoR for centrally-cleared transactions will be discussed in Section 9.1.2), it is useful to define the MPoR as the combination of two periods:

• Pre-default. This represents the time prior to the counterparty being in default and includes the following components:
  • Valuation and margin call. This represents the time taken to compute the current MTM and market value of margin already held, working out if a valid call can be made, and making that call. This should include the time delay due to the contractual period between calls.
  • Receiving collateral. The delay between a counterparty receiving a margin request to the point at which they release cash or securities. Clearly, recent developments such as electronic messaging are preferable to older-style approaches such as fax and email. The possibility of a dispute (i.e. if the margin giver does not agree with the amount called for) should be incorporated here.
  • Settlement. Margin will not be immediately received as there is a settlement period depending on the type of asset. Cash margin may settle on an intraday basis, whereas other securities will take longer. For example, government and corporate bonds may be subject to one- and three-day settlement periods respectively in some markets.
  • Grace period. In the event a valid margin call is not followed by the receipt of the relevant assets, there may be a relevant grace period before the counterparty will be deemed to be in default. This is sometimes known as the ‘cure period’.
• Post-default. This represents the process after the counterparty is contractually in default and the surviving party can legally initiate a close-out process.
  • Macro-hedging. A party may use macro-hedges to rapidly neutralise any key sensitivities of a portfolio with a defaulting bilateral counterparty, although there may be an issue in claiming losses in the event that these hedges lose money.
  • Close-out of transactions. The contractual termination of transactions and representation of the future cash flow as a single MTM value.
  • Rehedging and replacement. The replacement or rehedging of defaulted transactions. This may be done in different ways, such as:
    • One-to-one. A direct replacement of each individual transaction with identical economic terms. Although potentially expensive, this would be the likely requirement for an end user of derivatives so as not to incur any additional market risk and to avoid accounting mismatches.
    • Portfolio. This may involve using a smaller number of transactions to neutralise the market risk of the portfolio in question. This ‘macro hedge’ would be more likely to be a strategy adopted by sophisticated market participants such as a bank and would be cheaper in terms of transaction costs.
  • Liquidation of assets. The liquidation (sale) of margin securities.⁴¹

Note that the MPoR will be much longer than the time taken to receive margin in normal cases and normal market conditions (which may well be small) because margin performs no function (in terms of mitigating counterparty risk, at least) in these non-default situations.⁴² Instead, a party must consider a scenario where their counterparty is in default and market conditions may be far from normal. Because of this, Basel II capital requirements specified that banks should use a minimum of 10 business days’ MPoR for OTC derivatives in their modelling.⁴³ The Basel III regime defines a more conservative 20-day minimum or more in certain cases. By contrast, CCPs make assumptions regarding the MPoR of around five business days (see Section 13.6.2).

The MPoR should also potentially be extended due to initiatives such as the ISDA Resolution Stay Protocol that would temporarily restrict certain default rights (by 24 or 48 hours) in the event of a counterparty default.⁴⁴ The 18 major global banks have agreed to sign this protocol, which is intended to give regulators time to facilitate an orderly resolution in the event that a large bank becomes financially distressed. Although it is intended to apply primarily to globally-systemically-important financial institutions (G-SIFIs), in time the protocol may apply to other market participants too.

Note that the MPoR is generally used as a model parameter to define the market risk arising from a collateralised exposure. It is therefore important not to take the definition too literally. For example, consider Figure 7.18, which illustrates a simplified version of the dynamics described above for bilateral markets, involving a pre- and post-default period. This shows that the risk of the portfolio will initially stay fixed and the surviving party may, therefore, be exposed to significant market risk. However, during the post-default period, the risk will decline as the close-out process is implemented.

Models are generally simpler than even the simple approach above and use only a single time step. They also implicitly assume that 100% of the risk will be present for the entire period assumed, whereas in reality the risk will reduce, as illustrated in Figure 7.18. One could, therefore, argue that the MPoR used in a model should be shorter due to the fact that risk will reduce during the MPoR. For example, Appendix 7A shows that under certain assumptions a 10-day exposure is equivalent to a 31-day exposure which decreases linearly through time.⁴⁵

Figure 7.18 Schematic illustration of the MPoR in bilateral markets.

Another point is that the volatility of a portfolio might be expected to increase after a counterparty default (depending, of course, on the type and size of the counterparty). Models generally do not incorporate such effects, and it could therefore be argued that they should be proxied by using a longer MPoR. For example, an increase in volatility of 20% would imply an MPoR almost 50% longer.⁴⁶ This is discussed in more detail by Pykhtin and Sokol (2013).

Yet another problem for MPoR assessment is the potential for cash flow and margin payments happening within the pre- or post-default periods and the fact that these are likely to be asymmetric (i.e. a defaulting party will likely not pay, but a surviving party will). Andersen et al. (2017a, 2017b) show that this can have a significant impact.

Given all the above complexities, it is important to view the MPoR as a fairly simple estimate of the effective – not actual – period for which a surviving party suffers market risk in the event of a counterparty default. The MPoR is a rather simple ‘catch-all’ parameter and should not be compared too literally with the actual time it may take to effect a close-out and the replacement of transactions. The modelling of MPoR for cases where margin is present will be discussed in Chapter 15. The choice of initial margin is closely tied to the assumed MPoR, as is clearly illustrated in Figure 7.17. Indeed, as discussed in Chapter 9, initial margin methodologies will typically use a time horizon that can be seen to be consistent with the MPoR.

The MPoR is the primary driver of the need for initial margin. Assuming only variation margin is present, the best-case reduction of counterparty risk would be expected to be in the region of the square root of the ratio of the maturity divided by the MPoR.⁴⁷ This would imply a reduction of times for a five-year portfolio.⁴⁸ In reality, the result is worse than this estimate, as explained in more detail in Section 15.5. Hence, the only way to reduce counterparty risk to negligible levels is to have additional security in the form of initial margin.

7.5.3 Liquidity, FX, and Wrong-way Risks

Holding margin creates liquidity risk in the event that margin has to be liquidated following the default of a counterparty. In such a case, the non-defaulting party faces transaction costs (e.g. bid-offer) and market volatility over the liquidation period when selling margin securities for cash needed to rehedge their derivatives transactions. This risk can be minimised by setting appropriate haircuts (Section 7.3.5) to provide a buffer against prices falling between the counterparty defaulting and the party being able to sell the securities in the market. There is also the risk that by liquidating an amount of a security that is large compared with the volume traded in that security, the price will be driven down and a larger loss (potentially well beyond the haircut) will be incurred. Finally, there is wrong-way risk in case there is an adverse linkage between the default of the counterparty and the value of the margin securities (e.g. an entity posts bonds of its own sovereign).

When agreeing to margin that may be posted, and when receiving securities as collateral, important considerations are:

• What is the total issue size or market capitalisation posted as collateral?
• Is there a link between the margin value and the credit quality of the counterparty? Such a link may not be obvious and may be predicted by looking at correlations between variables.⁴⁹
• How is the relative liquidity of the security in question likely to change if the counterparty concerned is in default?

Because of liquidity impacts, a concentration limit of 5–10% may be imposed to prevent severe liquidation risk in the event of a counterparty defaulting.

Margin posted in cash or securities denominated in a currency that does not match the currency or currencies of the portfolio in question will also give rise to FX risk. Even a European bank posting cash in euros will give rise to a potentially problematic linkage and wrong-way risk.

Given the above issues, it is not surprising that margin is increasingly denominated in cash in major currencies or in securities with minimal credit, market and liquidity risks. For example, ISDA (2017c) reports that, amongst the major financial institutions, 70.9% of the margin is in cash, 20.7% in government securities and 8.3% in other securities.⁵⁰ For smaller users of derivatives, a more bespoke margin arrangement still allows less liquid margin to be posted.

7.5.4 Legal and Operational Risks

The time-consuming and intensely dynamic nature of collateralisation means that operational risk is present, due to aspects such as missed margin calls, failed deliveries, or human error. Operational risk can be especially significant for the largest banks, which may have thousands of relatively non-standardised margin agreements with clients, requiring posting and receipt of billions of dollars of margin on a given day. As noted in Section 7.5.2, Basel III recognises operational risk in such situations by requiring a larger MPoR to be used in some cases (e.g. where the number of transactions exceeds 5,000, or the portfolio contains illiquid margin or exotic transactions that cannot easily be valued under stressed market conditions). A history of margin call disputes also requires a larger MPoR to be used. Larger MPoRs lead to higher capital requirements and therefore produce an incentive for reducing operational risk.

Receiving margin also involves some legal risk in case there is a challenge to the ability to net margin against the value of a portfolio or over the correct valuation of this portfolio. In major bankruptcies this can be an important consideration, such as the case between Citigroup and Lehman Brothers discussed in Section 6.3.5.

As already discussed above, rehypothecation and segregation are subject to possible legal risks. Holding margin gives rise to legal risk in that the non-defaulting party must be confident that the margin held is free from legal challenge by the administrator of their defaulted counterparty. In the case of MF Global (see below), segregation was not effective, and customers lost money as a result. This raises questions about the enforcement of segregation, especially in times of stress. Note that the extreme actions of senior members of MF Global in using segregated customer collaterals were caused by a desperation to avoid bankruptcy.

---

MF Global and segregation

The case of MF Global provides a good illustration of the potential risks of segregation. MF Global was a major derivatives broker that filed for bankruptcy in October 2011. The aim of segregation is to prevent rehypothecation and make margin safe in the event of default of the margin receiver (this applies mainly to overcollateralisation in the form of initial margin). Unfortunately, it became clear that, prior to the bankruptcy, MF Global had illegally transferred a total of $1.6bn of segregated customer margin to third parties to meet overdrafts and margin calls. As of June 2013, 89% of this had been located and returned. The violation of segregation laws resulted in private lawsuits and a civil lawsuit filed by the U.S. Commodity Futures Trading Commission against the former CEO of MF Global.⁵¹ It is perhaps not surprising that to avoid insolvency, extreme and even illegal actions may be taken. There is obviously the need to have very clear and enforceable rules on margin segregation.

---

7.6 MARGIN AND FUNDING

7.6.1 Overview

The traditional role of margin for bilateral OTC derivatives has been as a counterparty risk mitigant. However, there is another role which has become more important in recent years, which is a provision of funding. Without receiving margin, an institution could be owed money but would not be paid immediately for this asset. Since institutions are often engaged in hedging transactions, this can create funding problems (e.g. a bank not receiving margin on a transaction may have to post margin on the associated hedge trade, as shown in Figure 7.7). This also shows why aspects such as segregation and rehypothecation (which define the ability – or not – to reuse margin) are important.

As margin has relevance in funding as well as counterparty risk reduction, one point to bear in mind is that different types of margin may offer different counterparty risk and funding benefits. An important distinction is that margin as a counterparty risk mitigant is by definition required only in an actual default scenario. On the other hand, margin as a means of funding is relevant in all scenarios. The following two cases can be seen as extreme examples of this:

• Counterparty posts wrong-way margin. A counterparty posting margin such as their own bonds which have significant wrong-way risk will act as a poor counterparty risk mitigant, but will provide funding as long as the margin can be reused or rehypothecated.
• Counterparty posts non-reusable margin. In the event that it is not possible to reuse margin securities (e.g. corporate bonds that cannot be repoed or posted under other margin arrangements), counterparty risk will be mitigated, but a funding requirement will persist.

The Totem xVA consensus service (Section 5.3.5) makes use of the above when asking participants to price a transaction with a two-way margin agreement but where the ‘counterparty posts to a segregated account’. Although this is uncommon in practice (especially since this is related to variation margin and not initial margin), it gives a case where a bank will likely include only funding in the price of a transaction without any adjustment for counterparty risk.⁵²

In recent years, margin eligibility and reuse have gained significant interest as funding costs have been viewed as significant. Therefore, the consideration of the type of margin that will be posted and received is important since different forms of margin have different funding costs and remuneration rates. When posting and receiving margin, institutions are becoming increasingly aware of the need to optimise this process and maximise funding efficiencies. Margin management is no longer the work of a back-office operations centre, but can be an important asset-optimisation tool delivering (and substituting) the most cost-effective cash and securities. A party should consider the cheapest-to-deliver cash margin and account for the impact of haircuts and the ability to rehypothecate non-cash collateral. For example, different currencies of cash will pay different OIS rates, and non-cash collateral, if rehypothecated, will earn different rates on repo. Chapter 16 provides a more in-depth study of these aspects.

7.6.2 Margin and Funding Liquidity Risk

Another important link between counterparty risk and funding is that increasing collateralisation may reduce counterparty risk, but it increases funding liquidity risk. Funding liquidity risk, in this case, is defined as the potential risk arising from the difficulty to raise required funding in the future, especially when margin needs to be segregated and/or cannot be rehypothecated. Margin agreements clearly create such risk as they require contractual margin payments over short timescales, with the magnitude of these payments typically not being deterministic (since it is related to the future value of transactions).

At a high level, the move to more intensive margin regimes can be seen as reducing counterparty risk at the expense of increasing funding liquidity risk (Figure 7.19). This occurs when moving from uncollateralised to collateralised trading via a typical CSA (two-way variation margin posting). The BCBS-IOSCO bilateral margin rules increase margin further through requiring initial margin, and central clearing takes this even further by adding default funds and potentially intraday-posting requirements (discussed in more detail in Chapter 8).⁵³

The problem with margin is that it converts counterparty risk into funding liquidity risk. This conversion may be beneficial in normal, liquid markets where funding costs are low. However, in abnormal markets where liquidity is poor, funding costs can become significant and may put extreme pressure on a party.

It is easy to understand how an end user of OTC derivatives might have significant funding liquidity risk, since they use derivatives only for hedging purposes and the need to post margin is probably not offset with a similar benefit (see discussion around Figure 2.9 in Section 2.2.5). Furthermore, due to their hedging needs, end users have directional positions (e.g. paying the fixed rate in interest rate swaps to hedge floating-rate borrowing). This means that a significant move in market variables (interest rates, for example) can create substantial MTM moves in their OTC derivatives portfolio and large associated margin requirements. This is why many end users do not have CSAs with their bank counterparties. Additionally, most end users (e.g. corporates) do not have substantial cash reserves or liquid assets that can be posted as margin (and if they did, then they might rather be able to use them to fund potential projects). Some end users (e.g. pension funds) have liquid assets such as government and corporate bonds but hold limited amounts of cash.

Figure 7.19 Illustration of the increasing impact of margin on counterparty risk and funding liquidity risk.

In addition to the previously discussed case of AIG (Section 2.4.4), there are other examples of the problems caused by the need to post margin in derivatives (see the Ashanti case below).⁵⁴

---

The Ashanti case

Ashanti (now part of AngloGold Ashanti Limited) was a Ghanaian gold producer. When gold prices rose in September 1999, Ashanti experienced very large losses of $450m on OTC derivatives contracts (gold forward contracts and options) used to hedge (possibly excessively) their exposure to a falling gold price.⁵⁵ The negative value of Ashanti's hedge book meant that its OTC derivatives counterparties were due further variation margin payments totalling around $280m in cash. Ashanti had a funding liquidity problem: it had the physical gold to satisfy contracts but not the cash to make the margin payments. To solve its liquidity crisis, Ashanti then struck an agreement that made it exempt from posting margin for just over three years.⁵⁶

---

Some non-financial clients such as institutional investors, large corporates and sovereigns do trade under CSAs with banks. They may do this to increase the range of counterparties they can deal with and achieve lower transaction costs (due to reduced xVA charges). In volatile market conditions, such CSA terms can cause funding liquidity problems due to significant margin requirements. Consider a corporate entering into a collateralised five-year cross-currency swap to hedge a bond issue in another currency. The potential MTM move and therefore margin could be as much as 55% of the notional of the swap.⁵⁷ Clients entering into margin agreements need to include an assessment of the worst-case margin requirements in their cash management and funding plans and understand how they would source eligible collateral (Section 21.3.4).

End users face funding liquidity risks when posting margin since it may possibly cause them to default in a case where they are solvent but unable to meet the margin demand in eligible securities within the timescale required. In turn, their bank counterparties may carry some of the risk since they may waive the receipt of margin to avoid this, with the obvious problem that it will be converted back into uncollateralised counterparty risk and funding requirements for the banks in question. Funding liquidity risk may also mean that a rating agency may have a more negative view of a company's credit quality if they agree to post margin (note that monolines could only achieve triple-A ratings through not posting margin; Section 2.4.4).

For banks, funding liquidity issues arise due to the nature of trading with clients. Since most banks aim to run mainly flat (hedged) OTC derivatives books, funding costs arise from the nature of hedging: an uncollateralised transaction being hedged via a transaction within a CSA arrangement (Figure 7.5). The bank will need to fund the margin posted on the hedge when the uncollateralised (client) transaction moves in their favour, and will experience a benefit when the reverse happens. Many banks will have directional client portfolios, which can lead to large margin requirements. This funding problem is one way to explain the need for FVA, discussed in more detail in Chapter 18. It also explains why banks are keen for more clients to sign CSA agreements to balance the margin flows in the situation depicted in Figure 7.5.

Banks also have a problem with contingent funding requirements arising from the potential need to post more initial margin, and also as a result of any margin requirements linked to their own credit ratings. Such requirements create a need for funding under the LCR (Section 4.3.3). It is, therefore, important to consider LCR-related funding costs in the assessment of xVA funding-related components such as FVA (Chapter 18) and margin value adjustment (MVA; Chapter 20).

A key decision for market participants and regulators alike is the concentration of various trading on the spectrum represented in Figure 7.19 and the risks that this presents. Whilst pushing to the right minimises counterparty risk, it also increases opaque and complex funding liquidity risks. Indeed, the reduction of counterparty risk and CVA and KVA has been a driver for creating other funding-related xVA terms such as MVA, discussed later in Chapter 20.

NOTES

1. 1 Margin is the amount of financial resources that must be deposited, whereas collateral refers to the actual financial instrument (e.g. cash or securities). Terminology will be discussed in Section 7.2.1.
2. 2 Cameron, M. (2012). Dealers call for revised language on break clause close-outs. Risk (5 March). www.risk.net.
3. 3 It is likely that the ownership of an optional break would be with the client relationship manager and possibly also the risk management department.
4. 4 Cameron, M. (2012). Banks tout break clauses as capital mitigant. Risk (3 March). www.risk.net.
5. 5 Referencing a floating rate of interest in both currencies.
6. 6 Note that the margin returned need not be exactly the same but must be equivalent (e.g. the same bond issue).
7. 7 For example, on the grounds that the original margin has been repoed, posted to another counterparty, sold, or is otherwise inaccessible.
8. 8 In other words, the negative valuation, which is the reason the variation margin has been required. Note that, as with netting, it is important to check that this would be upheld from a legal perspective.
9. 9 Title transfer leaves the margin giver as an unsecured creditor in the event of default of the margin receiver, since ownership of the underlying asset passes to the margin receiver at the time of transfer (and title is passed on in the event of rehypothecation).
10. 10 LCH.Clearnet (2017). Changes to the LCH model for Settled-to-Market FCM Contracts. Press release (11 December). www.lch.com.
11. 11 Of margin agreements in use, 87% are ISDA agreements (ISDA 2013c).
12. 12 The respondents to this survey were 41 ISDA member firms, most of whom are banks or broker-dealers.
13. 13 Although future regulatory requirements will reduce the need for negotiation (Section 7.4).
14. 14 Note that some xVA terms can constitute benefits overall.
15. 15 Some large corporates do post margin.
16. 16 Assuming that the amount is above the minimum transfer amount discussed next.
17. 17 Other less common examples are net asset value, market value of equity or traded credit spreads.
18. 18 AIG Form 10-K (2008).
19. 19 This would refer to the currency in which the portfolio is closed out in the event of a default.
20. 20 , and then multiplying by an approximate duration of 8.5 years.
21. 21 This should not be taken to imply that these assumptions were used to derive the results in Table 7.5, which has higher (more conservative) values.
22. 22 Note that there are benefits in taking margin in this form. Firstly, more margin can be called for as the credit quality of the sovereign deteriorates. Secondly, even a sudden jump to default event provides the recovery value of the debt as collateral. After this, the bank would have access to a second recovery value as an unsecured creditor.
23. 23 As noted in Section 7.2.3, this is typically defined as a base value.
24. 24 Note that the initial margin amount will not depend on the value. Initial margin methodologies are discussed in more detail in Chapter 9.
25. 25 Note that thresholds are increasingly zero, in which case this is not an issue.
26. 26 The purpose of an initial margin is to mitigate this risk by providing a buffer.
27. 27 Intradaily margin is likely to be required in the event of significant changes in valuation.
28. 28 The amounts are defined in euros and there are conversions to other currencies. The US dollar equivalent amounts are the same.
29. 29 Rehypothecation of initial margin is allowed in very limited situations where the transaction is a hedge of a client position, and will be suitably protected once rehypothecated, with the client having a priority claim under the relevant insolvency regime. It must be ensured that the initial margin can only be rehypothecated once and the client must be informed and agree to the rehypothecation. This is of limited benefit due to the potentially large chain of hedges that is executed across the interbank market in response to a client transaction. Sawyer, N. (2013). Industry ‘won't bother’ with one-time rehypothecation. Risk (12 September). www.risk.net.
30. 30 Subject to certain conditions, variation margin may be provided within two business days of calculation, under limited circumstances.
31. 31 In the EU, physically-settled FX forwards are subject to variation margin.
32. 32 For example, the U.S. Commodity Futures Trading Commission (CFTC) stated that from 1 March 2017 to 1 September 2017, there would be no enforcement against a swap dealer failing to comply with the variation margin requirements subject to a 1 March 2017 compliance date. In the EU, the first date for compliance was 4 February 2017 rather than 1 September 2016.
33. 33 This means that if a firm engages in separate derivatives transactions with more than one counterparty, but belonging to the same larger consolidated group (such as a bank holding company), then the threshold must essentially be shared in some way between these counterparties.
34. 34 For example, for a threshold amount of 50 and a calculated initial margin of 35, no margin is required. However, if the calculated margin is 65, then an amount of 15 needs to be posted.
35. 35 BCBS-IOSCO (2013). Basel Committee and IOSCO issue near-final proposal on margin requirements for non-centrally-cleared derivatives. Press Release (15 February). www.bis.org.
36. 36 BCBS-IOSCO (2015) states ‘Initial margin requirements for cross-currency swaps do not apply to the fixed physically settled FX transactions associated with the exchange of principal of cross-currency swaps’.
37. 37 The precise wording from BCBS-IOSCO (2015) is ‘Accordingly, the choice between model- and schedule-based initial margin calculations should be made consistently over time for all transactions within the same well-defined asset class.’
38. 38 This survey included 18 out of the 20 Phase 1 firms, with an estimate made for the missing two participants.
39. 39 These numbers were roughly symmetric, as would be expected.
40. 40 The fact that a larger amount of discretionary initial was received is likely due to some Phase 1 firms requiring some smaller counterparties to post initial margin. Once such firms fall in scope, the larger firms will also need to post and the numbers will become more symmetric.
41. 41 Note that this aspect should be included in the haircuts assigned to the margin assets.
42. 42 For example, in such a situation, margin may provide funding benefit.
43. 43 Assuming daily margin calls. If this is not the case, then the additional number of contractual days must be added to the time interval used.
44. 44 ISDA (2014). ISDA Publishes 2014 Resolution Stay Protocol (12 November). www.isda.org.
45. 45 The formula used here, which is explained in Appendix 7A and Jorion (2007) is . The value of n is the number of days of linearly-decaying exposure, and the result of the formula gives the equivalent number of days of constant exposure.
46. 46 This is a result of the so-called ‘square root of time rule’, which is the result of independently-identically-distributed (iid) random variables. Hence, an increase of 20% increases the effective time by .
47. 47 This is also a consequence of the ‘square root of time rule’.
48. 48 Assuming 250 business days in a year.
49. 49 In the case of the Long-Term Capital Management (LTCM) default, a very large proprietary position on Russian government bonds made these securities far from ideal as margin.
50. 50 This survey covers most of the firms subject to the first phase of regulatory margin-posting (Section 4.4.2). The total is 99.9% due to rounding error.
51. 51 Congressional Research Service (2013). The MF Global Bankruptcy, Missing Customer Funds, and Proposals for Reform (1 August). www.crs.gov.
52. 52 Note that Totem does not include capital costs and so this example provides a means to isolate the impact of funding in the price.
53. 53 CCPs are generally more conservative in their initial margin methodologies. This is not surprising as they do not have to post initial margin themselves, although they can be more competitive via lowering such requirements.
54. 54 Unlike the AIG case, this example is not linked to any rating triggers.
55. 55 ‘Mr. Sam Jonah, Chief Executive of Ashanti, referring to their financial problems, stated “I am prepared to concede that we were reckless. We took a bet on the price of gold. We thought that it would go down and we took a position.”’ From O'Connor, G. (1999). Ashanti left exposed by ‘exotics’. Financial Times (8 November). www.ft.com.
56. 56 Hirschler, B. (1999). Ashanti wins three-year gold margin reprieve. GhanaWeb (2 November). www.ghanaweb.com.
57. 57 This assumes a five-year FX move at the 95% confidence level, with FX volatility at 15%.