6
Netting, Close-Out, and Related Aspects

6.1 OVERVIEW

This chapter describes the role of netting and close-out in over-the-counter (OTC) derivatives markets. Netting is a traditional way to mitigate counterparty risk where there may be a large number of transactions of both positive and negative value with either a single counterparty (bilateral netting) or multiple counterparties (multilateral netting). Close-out refers to the process of terminating and settling contracts with a defaulted counterparty. The contractual and legal basis for netting and close-out and their impact in terms of risk reduction and impact on valuation adjustments (xVA) will be described. Some other related forms of risk mitigation, such as trade compression and break clauses, will also be covered.

Financial markets, such as derivatives, can be fast moving, with some participants (e.g. banks and hedge funds) regularly changing their positions. Furthermore, portfolios may contain a large number of transactions, which may partially offset (hedge) one another. These transactions may themselves require contractual exchange of cash flows and/or assets through time. Where there are multiple redundant cash flows, these would ideally be simplified into a single payment where possible. This is the first role of netting, generally called ‘payment or settlement netting’. This relates primarily to settlement risk (Section 3.1.2).

Furthermore, in markets such as derivatives, the default of a counterparty – especially a major one such as a large bank – is a potentially very difficult event. A given surviving party may have hundreds or even thousands of separate bilateral transactions with that counterparty. They will need a mechanism to terminate their transactions rapidly and replace (rehedge) their overall position. The same is true for a central counterparty (CCP), which will have a large number of transactions to deal with in the event of the default of a clearing member. In such situations, it is clearly desirable for a party to be able to offset what it owes to the defaulted counterparty against what it itself is owed. This is the second role of netting, generally called ‘close-out netting’, and relates more directly to counterparty risk.

In order to understand netting and close-out in more detail, consider the situation illustrated in Figure 6.1. Suppose parties A and B trade bilaterally and have two transactions with one another, each with its own set of cash flows. This situation is potentially over-complex for two reasons:

Cash flows. Parties A and B are exchanging cash flows or assets on a periodic basis in relation to Transaction 1 and Transaction 2. However, where equivalent cash flows occur on the same day, this requires the exchange of gross amounts, giving rise to settlement risk. It would be preferable to amalgamate payments and exchange only a net amount (see discussion on settlement risk in Section 6.2.1).

Figure 6.1 Illustration of the need for netting in bilateral markets.
Close-out. In the event that either party A or B defaults, the surviving party may suffer from being responsible for one transaction that has moved against them, but may not be paid for the other transaction that may be in their favour. This can lead to uncertainty over cash flow payments or the ability to replace the transactions with another counterparty.

In general, netting can be seen as a method of aggregating obligations whilst keeping market risk constant (or close to constant), but reducing:

settlement risk;
counterparty risk;¹
operational risk;
liquidity risk; and
systemic risk.

We will define netting as being broadly based on either cash flows or valuations.

6.2 CASH FLOW NETTING

6.2.1 Payment Netting

Payment netting involves the netting of different payments between two counterparties that are (generally) denominated in the same currency. Such payments could have arisen from the same transaction (e.g. netting the fixed and floating interest rate swap payments due on a particular day) or different transactions (e.g. two interest rate swaps in the same currency). For example, suppose party B must pay party A 100 but also expects to receive 60 on the same day (Figure 6.2). It clearly makes sense to net these into a single payment of 40.

Schematic illustration of the impact of payment netting. — **Figure 6.2** Illustration of the impact of payment netting.

The above netting reduces the time and costs associated with making payments, which should also reduce operational risk. It also reduces the amount that has to be delivered by both parties, which should reduce counterparty and liquidity risk. Note that the above applies in both bilateral- and centrally-cleared markets.

Bilaterally, counterparties can also extend payment netting by proactively reducing the number of transactions between them by removing redundancies such as offsetting positions. This is often known as bilateral compression. Whilst this does not reduce net exposure, there are benefits arising from reducing gross exposure, notably a reduction in operational costs and legal risk. Furthermore, certain regulatory methodologies – due to their inherent simplicity – can unfairly penalise large gross (but not net) exposures. There is, therefore, an incentive for counterparties to reduce the total notional of their positions and gross exposure even without changing their net exposure to one another. The natural extension of this to multiple counterparties is achieved through multilateral compression (Section 6.2.4).

6.2.2 Currency Netting and CLS

Payment netting mitigates settlement risk primarily in a single currency. However, settlement risk is also a major consideration in foreign exchange (FX) markets, where the settlement of a contract involves payment of one currency against receiving the other. Due to the different currencies, the payments are made in different markets and potentially at different times. Such settlement risk or ‘Herstatt risk’ (Section 3.1.2) occurs because of the risk of delivering the entire notional in one currency images against receiving the notional equivalent in another currency images (Figure 6.3).

Schematic illustration of settlement risk in a physically-settled FX transaction. — **Figure 6.3** Illustration of settlement risk in a physically-settled FX transaction.

Schematic illustration of CLS. — **Figure 6.4** Illustration of CLS.

One way around the above problem is to use a non-deliverable forward (NDF) transaction, where the currency exchange is cash-settled based on the difference between the NDF rate and the prevailing spot FX rate applied to the notional. This prevents the exchange of the full notional amounts in the two currencies and affects payment netting within the contractual terms of the transaction.

It is often inconvenient or impossible to settle the currencies on a net basis and so FX transactions need to be settled physically via a gross exchange of notional, as in Figure 6.3. In 2002, banks developed a continuous linked settlement (CLS) service to reduce settlement risk in FX transactions across a range of eligible currencies.³ For example, Bank A delivers one currency payment to CLS Bank, and Bank B delivers the opposite currency to CLS Bank (Figure 6.4). Only after both currencies have arrived does CLS Bank make the outgoing payments to A and B. This is called payment versus payment (PVP). Parties still make the intended cash flows, but CLS ensures that one cannot occur without the other.

The use of CLS Bank can also provide operational efficiencies. For example, payments in the same currency may be netted multilaterally across multiple transactions settling on the same day.

The KfW Bankengruppe transaction that gave rise to the problem outlined above was a regular cross-currency swap, with euros being paid to Lehman and dollars being paid back to KfW. On the day Lehman Brothers declared bankruptcy, KfW made an automated transfer of €350m despite the fact that the stricken Lehman Brothers would almost certainly not be making the opposite dollar payment (at this time, this type of cross-currency swap could be safely settled via CLS). It should be noted that if they had withheld the payment, then this may have been challenged by the administrator of the Lehman Brothers estate.

6.2.3 Clearing Rings

Bilateral netting clearly works only with cooperation between two parties. However, with further cooperation, further benefits could be derived via multilateral netting. The first example of this arises through ‘clearing rings’, which were used on exchanges even before the development of central clearing. Clearing rings were relatively informal means of reducing exposure via a ring of three or more members who would agree to ‘ring out’ offsetting positions. To achieve benefits, participants in the ring had to be willing to accept substitutes for their original counterparties. Rings were voluntary, but upon joining a ring, exchange rules bound participants to the ensuing settlements. Some members would choose not to join a ring, whereas others might participate in multiple rings. In a clearing ring, groups of exchange members agree to accept each other's contracts and allow counterparties to be interchanged. This can be useful for further reducing bilateral exposure, as illustrated in Figure 6.5. Irrespective of the nature of the other positions, the positions between C and D, and D and B can allow a ‘ringing out’, where D is removed from the ring and two obligations are replaced with a single one from C to B. Clearing rings simplify the dependencies of a member's open positions and allow them to close out contracts more easily.

Schematic illustration of a clearing ring. The equivalent obligations between C and D and between D and B are replaced with a single obligation between C and B. — **Figure 6.5** Illustration of a clearing ring. The equivalent obligations between C and D and between D and B are replaced with a single obligation between C and B.

It is important to note that not all counterparties in the example shown in Figure 6.5 benefit from the clearing ring illustrated. Whilst D clearly benefits from being readily able to offset the transactions with C and B, A is indifferent to the formation of the ring since its positions are not changed. Furthermore, the positions of B and C have changed only in terms of the replacement counterparty they have been given. Clearly, if this counterparty is considered to have stronger (weaker) credit quality, then they will view the ring as a benefit (detriment). A ring, whilst offering a collective benefit, is unlikely to be seen as beneficial by all participants. A member at the ‘end of a ring’, with only a long or short position and therefore standing not to benefit, has no benefit to ring out. Historically, such aspects have played out with members refusing to participate in rings because, for example, they preferred larger exposures to certain counterparties rather than smaller exposures to other counterparties.

6.2.4 Portfolio Compression

Suppose parties A, B, and C need to make the same cash flows shown in Figure 6.6,⁴ which may already have been reduced by bilateral netting. There is an opportunity to reduce the magnitude of the payments further, as shown, because all three parties are both paying and receiving. This would require some form of more advanced trilateral, or in general multilateral, netting scheme. Such a scheme would potentially reduce exposure further: for example, party A would have no exposure since the payment originally to be received from party B is achieved via a reduction of the amount paid to party C.

A modern-day equivalent of clearing rings in bilateral OTC derivatives markets is portfolio compression, which achieves multilateral netting benefits via the cooperation of multiple counterparties. A well-known example is TriOptima's TriReduce service,⁵ which provides compression services covering major OTC derivatives products across interest rates, credit, FX, and commodities. This has been instrumental in reducing exposures in OTC derivatives markets.⁶

Schematic illustration of the potential exposure reduction offered by multilateral netting. The arrows represent fungible cash flows, differing only in their size. — **Figure 6.6** Illustration of the potential exposure reduction offered by multilateral netting. The arrows represent fungible cash flows, differing only in their size.

Portfolio compression is a risk-reduction exercise where multiple counterparties partially or entirely eliminate transactions, potentially replacing them with other transactions, with the aim of reducing the overall notional value of the transactions. Note that compression – strictly speaking – refers to the reduction of notional value and not the number of contracts, although in practice the two may well be aligned. Note also that the aim is to reduce the total notional of all parties involved and it does not necessarily follow that an individual party will experience a lower notional (although participants can specify constraints to prevent their exposures increasing, for example).

Compression has developed because OTC derivatives portfolios grow significantly through time but contain redundancies due to the nature of trading (e.g. with respect to unwinds). This suggests that the transactions can be reduced in terms of number and gross notional without changing the overall risk profile. In essence, compression can preserve the market risk position of participants whilst reducing non-market risks such as settlement and counterparty risk. It can also reduce operational costs and may also lower systemic risk by decreasing the number of contracts that need to be replaced in a counterparty default scenario.

Compression is subject to diminishing marginal returns over time as the maximum multilateral netting is achieved. It also relies to some degree on counterparties being readily interchangeable, which implies, for example, that they need to have comparable credit quality.

Broadly, a typical compression cycle works as follows:

Participants submit the relevant transactions.
The transaction details are matched according to each bilateral counterparty in the process and may also be cross-referenced against a trade-reporting warehouse.
An algorithm is run to determine changes to transactions that generate multilateral netting benefits, but keeping each participant's portfolio neutral in terms of value and risk. These changes are reviewed by participants.
Once the process is finished, all changes are binding and take effect by unwinding transactions, executing new transactions, and novating transactions to other counterparties.

An optimal overall solution to the compression cycle can involve positions between pairs of counterparties increasing or changing sign and may involve changes in mark-to-market (MTM) value and risk sensitivities. For these reasons, participants can specify constraints (such as the total exposure to a given counterparty, which may be related to the internal credit limits of a participant).

A simple example of the potential result of a credit default swap (CDS) compression exercise for one market participant is given in Table 6.1. Here, the net long position resulting from transactions with three counterparties is reduced to a single identical long position with one of the counterparties.

Portfolio compression is not without potential issues. The settlement of a CDS contract can differ depending on whether it is triggered by the protection buyer or seller. Hence, netting long and short protection positions will not definitely create the same economic value. New transactions may become subject to certain regulation (such as bilateral margining), whereas the legacy transactions that led to their creation may have been exempt from these requirements.

Basic portfolio compression by its nature requires standard contracts, which are therefore fungible, and so cash flows are essentially being netted multilaterally. A good example of producing standardisation of this type is the CDS market, where large banks together with the International Swaps and Derivatives Association (ISDA) standardised CDS contracts in terms of coupons and maturity dates to aid compression (and, indeed, facilitate central clearing). CDS contracts now trade with both fixed premiums and upfront payments, and scheduled termination dates of 20 March, 20 June, 20 September, or 20 December. This means that positions can be bucketed according to underlying reference entity (single name or index) and maturity but without any other differences (such as the previous standard where coupons and maturity dates would differ).

Standardisation of contracts to aid compression is not always possible. For example, interest rate swaps typically trade at par via a variable fixed rate. In such cases, compression is less easy because two interest rate swaps could have identical floating cash flows but different fixed cash flows (due to the different swap rates used at inception). In such cases, slightly more advanced algorithms have been developed, such as coupon blending. These will be discussed further in Section 6.2.5, which details compression at CCPs in more detail.

Recent regulation since the global financial crisis has sought to increase compression exercises. For example, European Market Infrastructure Regulation (EMIR) states that both financial and non-financial counterparties with more than 500 non-centrally-cleared derivatives contracts must have procedures in place to, at least twice a year, consider the potential counterparty-risk-mitigating benefit of a portfolio compression exercise.⁷ Where such an exercise is not undertaken, the reasons for this must be articulated to the relevant supervisory authority. Possible reasons for not undertaking compression could be the fact that no benefit is likely (which could be the case if a counterparty has very directional positions with all other counterparties). Alternatively, compression may be avoided due to accounting, tax, or legal disadvantages. For example, an end user using an interest rate swap to hedge the risk on a loan (Section 2.2.5) may have offsetting swaps hedging their own borrowing and compression may create problems, such as in the recognition of hedge accounting.

Table 6.1 Simple illustration of trade compression for single-name CDS contracts. A party has three contracts on the same reference credit and with identical maturities but transacted with different counterparties. It is beneficial to ‘compress’ the three into a net contract, which represents the total notional of the long and short positions. This may naturally be with counterparty A as a reduction of the initial transaction.

Reference	Notional	Long/short	Maturity	Coupon	Counterparty
ABC index	150	Long	20/12/2023	100	Counterparty A
ABC index	75	Short	20/12/2023	100	Counterparty B
ABC index	50	Short	20/12/2023	100	Counterparty C

ABC index	25	Long	20/12/2023	100	Counterparty A

6.2.5 Compression Algorithm

In order to understand the complexity of trade compression algorithms, consider the example ‘market’ represented in Figure 6.7. This shows position sizes⁸ between different counterparties in certain fungible (interchangeable) products. Note that the total gross notional between counterparties (counting each transaction twice, both points of view) is 1,250.

Suppose that the general aim of portfolio compression is to reduce the gross notional in Figure 6.7 without changing the net position of any counterparty. This is likely to be a subjective process for a number of reasons. Firstly, it is not clear what should be minimised. An obvious choice may be the total notional, although this would not penalise large positions or the total number of positions. Alternative choices could be to use the squared notional or the total number of positions, which would reduce large exposure and interconnectedness respectively (O'Kane 2017 discusses this point in more detail). Secondly, there may need to be constraints applied to the optimisation, such as the size of positions with single counterparties. In the above example, there is no transaction between counterparties 1 and 3. It may be that one or both of them would like to impose this as a constraint. Many different algorithms could be used to optimise the market above, and commercial applications have tended to follow relatively simple approaches (for example, see Brouwer 2012). The example below, albeit for a very small market, will provide some insight into how compression algorithms work in practice.

One obvious method to reduce the total notional is to look for opportunities for netting within rings in the market. A trilateral possibility occurs between counterparties 2, 3, and 4 (as illustrated in Figure 6.8) where notionals of 60, 70, and 85 occur in a ring and can, therefore, be reduced by the smallest amount (assuming positions cannot be reversed) of 60. This strategy corresponds to minimising the total notional (or indeed the total notional squared).

Schematic illustration of a simple market made up of positions in fungible contracts. — **Figure 6.7** Illustration of a simple ‘market’ made up of positions in fungible (interchangeable) contracts.

Schematic illustration of using trilateral netting between counterparties 2, 3, and 4 to reduce the overall notional of the system. — **Figure 6.8** Illustration of using trilateral netting between counterparties 2, 3, and 4 to reduce the overall notional of the system shown in Figure 6.7.

This leads to the total notional of the compressed system being reduced to 890 (from 1,250) on the right-hand side of Figure 6.8.

Continuing a process such as the one above could lead to a number of possible solutions, two of which are shown in Figure 6.9. Note that the solution on the left-hand side has reversed the exposure between counterparties 4 and 5, whilst on the right-hand side there is a transaction between counterparties 1 and 3 where none existed previously. The latter solution has a lower total notional of 110 (compared to 130 for the former). However, this also illustrates that constraints imposed by counterparties (e.g. 1 and 3 not wanting exposure to one another) will weaken the impact of compression.

Schematic illustration of two possible final results of compressing the original market in Figure 6.7 which leads to the total notionals of 130 on left-hand side and 110 on right-hand side. — **Figure 6.9** Illustration of two possible final results of compressing the original market in Figure 6.7, leading to total notionals of 130 (left-hand side) and 110 (right-hand side).

Figure 2.4 shows the impact of the much greater emphasis on compression in OTC derivatives in the last few years.

6.2.6 Benefits of Cashflow Netting

In general, the traditional netting approaches described above can be seen as reducing the gross value of transactions without changing the market risk profile. The immediate benefits of bilateral and multilateral netting of cash flows are relatively obvious via a reduction in settlement risk and, by extension, counterparty risk. However, there are some other benefits which may not be as obvious or intuitive:

Operational costs. By reducing the number of cash flows and transactions, operational costs can be reduced.
Legal risks. Portfolio compression may reduce legal risk by contractually simplifying portfolios and not relying on future risk mitigation (which may be subject to legal challenges).
Regulatory. The reduction of gross notional may not always seem beneficial since the net exposure may remain the same. However, some simple regulatory methodologies are partially based on gross notional and will, therefore, appear more beneficial if these values are reduced. This mainly applies to regulatory capital requirement, a good example being the leverage ratio (Section 4.2.7), which has provided a strong incentive for portfolio compression. Some regulatory thresholds are also based on gross notional, such as in relation to central clearing (for example, see discussion on NFC+ designation in Section 4.4.3) or bilateral margin requirements (Section 4.4.2).

6.3 VALUE NETTING

6.3.1 Overview

Cash flow netting has fairly obvious restrictions since it requires cash flows (and therefore transactions) to be interchangeable. This means that the type of cash flow must be the same, suggesting the same asset class, product, payment date, and reference rate. However, many derivatives transactions with a given counterparty can be similar but not have fungible cash flows and sometimes may relate to several underlying asset classes. A more general concept is, therefore, to apply netting across the value of underlying transactions.

A related point is that bankruptcy proceedings are, by their nature, long and unpredictable processes. During such processes, likely counterparty risk losses are compounded by the uncertainty regarding the termination of the proceedings. A creditor who holds an insolvent firm's debt has a known exposure, and whilst the eventual recovery is uncertain, it can be estimated. However, this is not the case for derivatives, where constant rebalancing is typically required to maintain hedged positions. Furthermore, once a counterparty is in default, cash flows will cease, and an institution will be likely to want or need to execute new replacement contracts.

For transactions such as derivatives that can be both assets and liabilities, there is also the question of whether these can offset one another in a default event. An even broader concept is whether such an offset can be applied to different business with a given counterparty. For example, a bank may have a lending relationship with a given client but also transact OTC derivatives with them. If the client defaults, then there is a question of whether the underlying transactions should be considered independently or not.

6.3.2 Close-out Netting

Close-out netting aims to minimise counterparty risk across a portfolio of transactions with a defaulted counterparty. Parties can reduce their risk to each other via contractual terms, such as an ISDA agreement. Close-out netting comes into force in the event that a counterparty defaults, and aims to allow a timely termination and settlement of the net value of all transactions with that counterparty. Essentially, this consists of two components:

Close-out. The right to terminate transactions with the defaulted counterparty and cease any contractual payments.
Netting. The right to offset the value across transactions and determine a net balance, which is the sum of positive and negative values, for the final close-out amount.⁹

Close-out netting permits the immediate termination of all contracts with a defaulted counterparty and the settlement of a net amount reflecting the total value of the portfolio (Figure 6.10). In essence, with close-out netting, all transactions (of any maturity, whether in- or out-of-the-money) collapse to a single net value. If the surviving party owes money, then it makes this payment; if it is owed money, then it makes a bankruptcy claim for that amount. Close-out netting allows the surviving institution to realise gains on transactions against losses on other transactions immediately and effectively jump the bankruptcy queue for all but its net exposure, as illustrated in Figure 6.10. Note that close-out netting is completely general since it only depends on MTM values at the time of default and not matching cash flows.

Schematic illustration of the impact of close-out netting. — **Figure 6.10** Illustration of the impact of close-out netting. In the event of the default of party A, without netting, party B would need to pay 200 and would not receive the full amount of 140 owed. With netting, party B would simply pay 60 to party A and suffer no loss.

Whilst payment netting reduces settlement risk, close-out netting is relevant to counterparty risk since it reduces pre-settlement risk.

Netting is not just important for reducing exposure but also for reducing the complexity involved in the close-out of transactions in the event that a counterparty defaults. In OTC derivatives markets, surviving parties will usually attempt to replace defaulted transactions. Without netting, the total number of transactions and their notional value that surviving parties would attempt to replace may be larger and hence may be more likely to cause market disturbances.

Netting legislation covering derivatives has been adopted in most countries with major financial markets. The ISDA has obtained legal opinions supporting the close-out and netting provisions in their Master Agreements in most relevant jurisdictions. (At the time of writing, they currently have such opinions covering 54 jurisdictions.) Thirty-seven countries have legislation that provides explicitly for the enforceability of netting. However, jurisdictions remain where netting is not clearly enforceable in a default scenario.¹⁰ Note that in these scenarios, a bank may judge that netting is enforceable for pricing (and possibly accounting) purposes (and accepting the legal risk), but may not be able to reflect this in other calculations such as regulatory capital.

Note that bilateral cash flow netting (Section 6.2.1), where counterparties actively reduce the number of transactions bilaterally, is preferable to value netting because there is no legal risk as transactions are removed completely. For example, if two parties have two completely opposite positions, terminating them is preferable to relying on the enforceability of close-out netting in the event that one of them defaults.

6.3.3 Payment Under Close-out

Standard assumptions when defining counterparty risk and computing credit value adjustment are that a surviving party must still pay liabilities to a defaulted counterparty. This means that their post-default position is the same as that pre-default in that they must still perform on their liability. On the other hand, if a surviving party has an asset with a defaulting counterparty, then they will make some associated loss in default, depending on the amount they recover. This is a simplistic definition of the economic impact of default used in modelling, but is not often borne out precisely in practice.

For example, although no longer common, some OTC derivatives were historically documented with ‘walkaway’ or ‘tear-up’ features. Such a clause effectively allows an institution to cancel transactions in the event that their counterparty defaults. They would clearly only choose to do this in case they were in debt to the counterparty. Whilst a feature such as this does not reduce credit exposure, it does allow an institution to benefit from ceasing payments and not being obliged to settle amounts owed to a counterparty. These types of agreements, which were common prior to the 1992 ISDA Master Agreement, have been less common since and are not now part of standardised ISDA documentation. However, they have sometimes been used in transactions since 1992. Whilst walkaway features do not mitigate counterparty risk per se, they do result in potential gains that offset the risk of potential losses.

Walkaway agreements were seen in the Drexel Burnham Lambert (DBL) bankruptcy of 1990. Interestingly, in this case, counterparties of DBL decided not to walk away and chose to settle net amounts owed. This was largely due to the relatively small gains compared with the potential legal cost of having to defend the validity of the walkaway agreements or the reputational cost of being seen as taking advantage of the DBL default.

Even without an explicit walkaway agreement, an institution can still attempt to gain in the event of a counterparty default by not closing out contracts that are out-of-the-money (OTM) to them, but ceasing underlying payments. Another interesting case is that between Enron Australia (Enron) and TXU Electricity (TXU) involving a number of electricity swaps which were against TXU when Enron went into liquidation in early 2002. Although the swaps were not transacted with a walkaway feature, ISDA documentation supported TXU avoiding paying the MTM owed to Enron (A$3.3m) by not terminating the transaction (close-out), but ceasing payments to their defaulted counterparty. The Enron liquidator went to court to try to force TXU to settle the swaps, but the New South Wales Supreme Court found in favour of TXU in that they would not have to pay the owed amount until the individual transactions expired (i.e. the obligation to pay was not cancelled, but it was postponed).

Some Lehman Brothers counterparties also chose (like TXU) not to close out swaps but to stop making contractual payments (as their ISDA Master Agreements seemed to support).¹¹ Since the swaps were very OTM from the counterparties' point of view, and therefore strongly in-the-money (ITM) for Lehman, there were potential gains to be made from doing this. Again, Lehman administrators challenged this in the courts. US and English courts came to different conclusions with respect to the enforceability of this, with the US court ruling that the action was improper,¹² whilst the English court ruled that the withholding of payments was upheld.¹³

Any type of walkaway feature is arguably rather unpleasant and should be avoided due to the additional costs for the counterparty in default and the creation of moral hazard (since an institution is potentially given the incentive to contribute to their counterparty's default due to the financial gain they can make).

6.3.4 Close-out and xVA

There is an important link between the close-out and xVA. The close-out process aims to allow a surviving party to determine a reasonable valuation for their derivatives contracts so as to establish the amount payable at this point where all the underlying transactions are likely to be closed out. If the surviving party is a creditor, then such a valuation will determine their claim in the bankruptcy process. If they are a debtor, then it will determine the payment they are required to make to their bankrupt counterparty's estate.

The valuation at close-out tends to reference ‘replacement costs’, since a surviving party will typically replace transactions either directly or indirectly (in the latter case, by executing replacement trades, macro-hedges, or unwinding hedges). Such costs are typically perceived as being added to the value so as to compensate the surviving party.

The first problem with the above is regarding the definition of value and, generally, whether it should be the base or actual value (Section 5.2.1). The latter is clearly more relevant as it likely reflects the surviving party's current view on the actual valuation. However, unlike the base value, this requires a definition of xVA terms, which is complex and may not be objectively defined from a legal standpoint.

Assuming the surviving party has a positive valuation (i.e. they are a creditor), then the situation will be as in Figure 6.11. Note that the xVA adjustment is likely to be negative since the overall adjustment will be dominated by costs over benefits.¹⁴ Note that, in this situation, the surviving party stands to gain by using a base valuation over the actual valuation since this will enable them to recover a higher amount as a creditor.

Graph depicts the determination of valuation in the event of counterparty default from the surviving party’s point of view and assuming they are a creditor. The xVA adjustment is assumed to be negative overall. — **Figure 6.11** Illustration of the determination of valuation in the event of counterparty default from the surviving party's point of view and assuming they are a creditor (i.e. their valuation is positive). The xVA adjustment is assumed to be negative overall.

If the surviving party is a debtor and owes the defaulted counterparty, then the situation might be reversed, with the actual value being higher than the base value (Figure 6.12).¹⁵ Note that, in this situation, the surviving party potentially gains by referencing the actual – and not the base – value.

Note that, together with xVA, there is the question of cost inherent in replacing transactions which may also be part of the defined value in default since the surviving counterparty may reasonably replace transactions (Figure 6.13). It may not be able to easily separate such costs (e.g. bid-offer costs) from xVA terms since they may both be seen as charges in executing replacement transactions.

Given the inherent problems with defining the value of a derivative in the event of a counterparty default, it may also be helpful to define a ‘legal value’, as shown in Figure 6.13. This is a value that might be claimed by a surviving party based on the underlying documentation, but which might be seen as being different (inflated) compared to the true value and associated costs, and might potentially lead to litigation. Clearly, documentation should aim to prevent such divergences since surviving parties may be able to gain at the expense of other creditors of the defaulted counterparty.

6.3.5 ISDA Definitions

The contractual definition regarding close-out is crucial in defining the economics of counterparty default and, as such, is a key element in defining counterparty risk and related xVA components. The default of Lehman Brothers in 2008 illustrated some of the issues with determining close-out valuations. In particular, Lehman Brothers had posted substantial amounts of collateral or security to counterparties as their credit quality deteriorated. Surviving counterparties were then incentivised to maximise their benefit under the relevant documentation in order to keep as much of this collateral as possible. The Lehman estate then had to proactively try to retrieve much of this collateral – often through the courts – in order to be fair to their creditors overall.¹⁶ However, there are few legal precedents due to the fact that many cases have settled out of court and contradictory decisions have been made by the English and US courts.

Schematic illustration of the determination of valuation in the event of counterparty default from the surviving party’s point of view, including costs and assuming they are a creditor. — **Figure 6.13** Illustration of the determination of valuation in the event of counterparty default from the surviving party's point of view, including costs (shown with respect to the actual value) and assuming they are a creditor.

The close-out amount represents the amount that is owed by one party to another in a default scenario. If this amount is positive from the point of view of the non-defaulting party, then they will have a claim on the estate of the defaulting party. If it is negative, then they will be obliged to pay this amount to the defaulting party. Although the defaulting party will be unable to pay the claim in full, establishing the size of the claim is important. The determination of the appropriate close-out amount is complex as parties will inevitably disagree. The non-defaulting party will likely consider their value of executing replacement transactions (‘replacement cost’) as the economically correct close-out amount. The defaulting party may not agree with this assessment since it will reflect charges such as bid-offer costs which it does not experience.

The ISDA Master Agreement (Section 2.2.6) is a market-standard contract used to document OTC derivative transactions, and it is important to understand the implication of the definitions with respect to the amount owing in the event of a counterparty default. There are two ISDA versions to consider – namely 1992 and 2002 – which differ in their definitions (Table 6.2).

Under the 1992 ISDA, there are two methods for defining the amount owed in default, namely ‘market quotation’ and ‘loss’, with the former often being elected as the primary method and the latter as a fallback (in case achieving a market quotation is not possible). These are characterised as follows:

Table 6.2 ISDA definitions regarding the determination of the net amount owing between two parties in the event of a default of one of them.

1992 ISDA		2002 ISDA
Market quotation	Obtain a minimum of three firm quotes for the portfolio in question and combine these quotes.	Close-out amount	Indicative quotations, public sources of price information, models.
Loss	Estimate total losses and gains reasonably and in good faith.	Close-out amount

Market quotation. The determining (surviving) party obtains a minimum of three quotes from market makers and uses these quotations (e.g. in the case of three quotes, the middle value should be used). In the event that three quotations cannot be achieved then market quotation is deemed to have failed.
Loss. The determining party is required to calculate its total losses or gains in good faith. Such an amount is intended to be representative of the amount required to put the determining party in the position that it would have been in had the contract been performed, and may include loss of bargain, funding costs, and trading-related costs (e.g. terminating or re-establishing hedges).

Market quotation is clearly designed to be a relatively objective measure, but it obviously requires a reasonable amount of liquidity in the market for the particular transactions in question. Such liquidity is not always present, especially in the aftermath of a major default (e.g. Lehman Brothers) and in more exotic or non-standard products or non-standard contractual terms (assuming the surviving counterparty aims to replicate such terms in replacement transactions). Therefore, it has sometimes been problematic to find market makers willing to price complex transactions realistically following a major default. Since 1992 there has been an increasing number of more complex and structured OTC derivative transactions, together with non-standard contractual terms (e.g. one-way margin agreements). This has led to a number of significant disputes in the determination of the market quotation amount (e.g. see Figure 2.11).

Lehman Brothers, Citigroup, and ‘phantom transaction costs’

In the aftermath of the Lehman Brothers' bankruptcy, Citigroup held around $2bn in collateral and argued that most of this amount was required to cover the replacement of these transactions. The administrators of the Lehman estate argued against this and claimed that the money, in fact, should go to its other creditors and accused Citigroup of using methods such as ‘phantom transaction costs’ to try and justify its claim. Eventually Citigroup agreed to give back $1.74bn to the estate of Lehman from a total of $2.1bn and stated:

Citigroup says it used its best professional judgment to determine close-out amounts on the trades. The contracts, the bank said, were governed by ISDA agreements which give the non-defaulting party – in this case Citigroup – the right to assess the close-out costs as long as it uses commercially reasonable procedures.¹⁷

On the other hand, loss is potentially too subjective and gives too much discretion to the determining party. This implies that there may be incentives for the determining party to deliberately cause market quotation to ‘fail’ so as to be able to use loss as a fallback. In the event that the determining party makes gains, these would be at the detriment of other creditors of the defaulting party. The Lehman Brothers bankruptcy also gave rise to cases of this type.¹⁸

Because of the above problems and market developments (such as the availability of more external pricing sources), the 2002 ISDA Master Agreement introduced a new, single definition known as ‘close-out amount’. On the one hand, close-out amount can be seen as a diluted form of market quotation, as it does not require actual tradable quotes but can instead rely on indicative quotations, public sources of prices, and market data and internal models to arrive at a commercially reasonable price. On the other hand, close-out amount allows for a similar calculation as loss, but with greater objectivity since the determining party must act in an objectively-reasonable manner.

The close-out amount is the only methodology provided in the 2002 ISDA contract and is intended to reflect the losses or costs/gains of the determining party in replacing or providing the economic equivalent of the material terms of the transactions under the prevailing circumstances, determined in good faith and in a commercially reasonable manner. In determining a close-out amount, the determining party may consider any relevant information, including:

firm or indicative quotations for replacement transactions supplied by one or more third parties that may take into account the creditworthiness of the determining party and the terms of any relevant documentation (such as collateral agreement) between the determining party and a third party;
relevant market data supplied by one or more third parties; and
internal information, as above, from internal sources that are used by the determining party in the regular course of its business for the valuation of similar transactions.

In summary, the market quotation method is an objective approach that uses actual firm quotes from external parties. The loss method is more flexible, with the determining party choosing any reasonable approach to determine its loss or gain. The close-out amount method is somewhere in between, giving the determining party flexibility to choose its approach, but aiming to ensure that such an approach is commercially reasonable.

Following the publication of the 2002 ISDA Master Agreement, some parties continued to use market quotation via the 1992 ISDA Master Agreement on the basis that it produced a more objective result. However, during the global financial crisis, the problems associated with this payment method (especially in relation to the Lehman Brothers bankruptcy) were again highlighted. As a result, there has been a growing trend towards using the 2002 close-out amount definition. In 2009, ISDA published a close-out amount protocol to provide parties with an efficient way to amend older Master Agreements to close-out amount with only one signed document, rather than changing bilateral documentation on a counterparty-by-counterparty basis. The ISDA close-out amount protocol was introduced to give market participants an efficient way to amend their 1992 ISDA Master Agreements to replace market quotation and loss with close-out amount.

Note that close-out valuations do seem implicitly or explicitly to allow xVA components to be included in the valuation to the extent that they are part of the costs associated with establishing new transactions. This is in contrast to the need for valuations for other reasons, such as collateral posting or terminating transactions which typically reference only base values (Section 5.3.6). Whilst allowing the actual value to be realised during the close-out process is more reasonable, it does create more complexity in the xVA calculation due to the recursive problem of needing to know xVA at the counterparty default time. This will be discussed in later chapters.

6.3.6 Set-off

As noted in Section 6.3.5, close-out netting under an ISDA contract is generally deemed enforceable in virtually all major jurisdictions and is therefore assumed to be an effective risk mitigant. As such, banks will usually recognise such netting benefits when calculating regulatory capital requirements and pricing new transactions.

However, some institutions trade many financial products (such as loans and repos, as well as interest rate, FX, commodity, equity, and credit products). The ability to apply netting to most or all of these products is desirable in order to reduce exposure. However, legal issues regarding the enforceability of netting arise due to transactions being booked with various different legal entities across different regions.

Bilateral netting is generally recognised for OTC derivatives, repo-style transactions, and on-balance-sheet loans and deposits. Cross-product netting is typically possible within one of these categories (e.g. between interest rate and FX transactions) since they are typically all covered by the same documentation, such as an ISDA Master Agreement (Section 2.2.6). However, netting across these product categories (e.g. OTC derivatives and loans) is not definitely possible as they are documented differently.

The case of OTC derivatives and loans is especially relevant as many banks will have lending relationships with derivative counterparties and may provide a floating-rate loan in conjunction with an interest rate swap (with terms linked to those of the loan) as a ‘packaged’ fixed-rate loan (Section 2.2.5). Since these products are treated completely separately within a bank, there is a question of whether their values would be netted in default (e.g. the bank is owed money on the loan but owes money on the swap). There is also the related question of access to loan collateral to cover a derivative exposure. As with netting enforceability, banks may reflect such offsets in pricing and accounting but will not achieve benefit in terms of reduced capital requirements.

Schematic illustration of the concept of set-off. — **Figure 6.14** Illustration of the concept of set-off.

‘Set-off’ is a broad term that allows a party to apply an amount owed to it by the other party against amounts owed in the other direction. For example, in Figure 6.14 a right of set-off would allow party B to reduce its debt to party A from $100 to $40 via set-off against an opposite debt of $60.

There is a subtle difference between payment netting (Section 6.2.1) and set-off. Set-off recognises the existence of cross-claims between parties and allows equivalent claims in opposite directions to be extinguished. Netting results in a single contractual claim at any point in time. The economic effect is typically the same in each situation.

Typically, set-off relates to actual obligations, whilst close-out netting refers only to a calculated amount. Set-off may therefore potentially be applied to offsetting amounts from other agreements against an ISDA close-out amount representing OTC derivatives. Under the 2002 ISDA Master Agreement, a standard set-off provision is included which would allow for the offset of any termination payment due against amounts owing to that party under other agreements. It is therefore potentially possible from a legal perspective to set-off derivatives against other products such as loans. However, this will depend on the precise wording of the different sets of documentation, the legal entities involved and legal interpretation in the relevant jurisdiction. Obtaining strong legal opinions is clearly critical, but since defaults are relatively rare events, there are often no practical examples to explore the enforceability of set-off.

6.4 THE IMPACT OF NETTING

6.4.1 Risk Reduction

Close-out netting is the single biggest risk mitigant for counterparty risk and has been critical for the growth of the OTC derivatives market. Without netting, the current size and liquidity of the OTC derivatives market would be unlikely to exist. Netting means that the overall credit exposure in the market grows at a lower rate than the notional growth of the market itself. Netting has also been recognised (at least partially) in regulatory capital rules (Chapter 13), which was an important aspect in allowing banks to grow their OTC derivative businesses. The expansion and greater concentration of derivatives markets have increased the extent of netting steadily over the last decade, such that netting currently reduces exposure by close to 90% (Figure 6.15).

6.4.2 The Impact of Netting

Netting has some subtle effects on the dynamics of derivatives markets. Firstly, although the size of exposures is smaller, netted positions are inherently more volatile than their underlying gross positions, which can create systemic risk. Another problem with netting occurs when an institution wants to trade out of a position. In such a situation, the relative illiquidity of OTC derivatives may be problematic. If the institution executes an offsetting position with another market participant, whilst removing the market risk as required, they will have counterparty risk with respect to the original and the new counterparty (unless this can later be reduced by compression). To offset the counterparty risk, it is necessary to trade with the original counterparty, who, knowing that the institution is heavily incentivised to trade out of the position with them, may offer unfavourable terms to extract the maximum financial gain. The institution can either accept these unfavourable terms or trade with another counterparty and accept the resulting counterparty risk. This point extends to establishing multiple positions with different risk exposures. Suppose an institution requires both interest rate and FX hedges. Since these transactions are imperfectly correlated, then by executing the hedges with the same counterparty, the overall counterparty risk is reduced and the institution may obtain more favourable terms. However, this creates an incentive to transact repeatedly with the same counterparty, leading to potential concentration risk.

Schematic illustration of the impact of netting on OTC derivatives exposure. The netting benefit is defined by dividing the gross credit exposure by the gross market value and subtracting this ratio from hundred percent. — **Figure 6.15** Illustration of the impact of netting on OTC derivatives exposure. The netting benefit (right-hand y-axis) is defined by dividing the gross credit exposure by the gross market value and subtracting this ratio from 100%.

Source: Bank for International Settlements. www.bis.org.

An additional implication of netting is that it can change the way market participants react to perceptions of increased risk of a particular counterparty. If credit exposures were driven by gross positions, then all those trading with the troubled counterparty would have strong incentives to attempt to terminate existing positions and stop any new trading. Such actions would likely result in even more financial distress for the troubled counterparty. With netting, an institution will be far less worried if there is no current exposure (MTM is negative). Whilst they will be concerned about potential future exposure and may require collateral, netting reduces the concern when a counterparty is in distress, which may, in turn, reduce systemic risk.

6.4.3 Multilateral Netting and Bifurcation

A complication to netting is created by regulatory mandates such as central clearing (Section 4.4.1) and bilateral margin requirements (Section 4.4.2).

In the case of central clearing, transactions that – from a netting perspective – would have been grouped against another bilateral counterparty are now grouped at the CCP level. This would seem to produce benefits through multilateral netting of positions held against a single CCP that would otherwise be facing multiple bilateral counterparties. Indeed, this is seen as a major advantage of central clearing.¹⁹

However, since not all transactions can be centrally cleared, there is a disadvantage due to the loss of bilateral netting benefits. This bifurcation between cleared and bilateral transactions can be particularly acute for market participants executing offsetting transactions (e.g. a swaption being hedged by a swap or index against single-name credit default swaps) since one product may be clearable and the other not. Clearly, there is a critical mass where enough OTC derivatives can be cleared through a reasonably small number of CCPs so as to create overall netting benefits. This has been illustrated by Duffie and Zhu (2011).

CCPs allow multilateral offset due to a clearing member facing the CCP directly on all cleared trades. As an example, consider the situation in Figure 6.16, where the arrows are probably best interpreted as cash flow or margin payments, as discussed in more detail below. This shows that although bilateral netting can reduce exposure significantly, central clearing can reduce it even more through multilateral netting.

As shown in Table 6.3, bilateral netting reduces the total exposure of the system in Figure 6.16 by a factor of three (360 to 120). This can be reduced further to 60 by central clearing, even if the exposure of the CCP is included (in practice this would be mitigated by margining).

Although the above example seems to be identical to compression (Section 6.2.4), there are important differences. For compression, trades need to be standardised, since this provides the fungibility so that contracts can be torn up to represent the result of the compression cycle. Contracts also need to be standardised for central clearing, but for different reasons relating to operational costs, margin calculations, and potential close-out in the event of clearing member default. Such differences mean that multilateral netting benefits can be seen for centrally-cleared trades that would not be achieved through trade compression. Put another way, portfolio compression can offset equivalent transactions, and possibly actual cash flows,²⁰ but central clearing can offset the actual value of these transactions against one another. This means that two different transactions with different counterparties that are not highly correlated (e.g. interest rate swaps in different currencies) will have a strong netting benefit under central clearing but are not appropriate for trade compression due to not being sufficiently fungible. Put another way, CCPs can compress risk, but in bilateral markets compression can only work on objectively-defined quantities such as notionals and cash flows.

Schematic illustration of the comparison of no netting, bilateral netting, and central clearing. — **Figure 6.16** Comparison of no netting, bilateral netting, and central clearing.

Table 6.3 Illustration of the reduction in exposure from bilateral netting and central clearing, as shown in Figure 6.16.

	No netting	Bilateral netting	Central clearing
Counterparty 1	170	50	30
Counterparty 2	90	20	0
Counterparty 3	100	50	0
CCP (C)	-	-	30
Total	360	120	60

When promoting central clearing, a key point made often by policymakers and regulators is that CCPs facilitate multilateral netting, which can alleviate systemic risk by reducing exposures more than in bilateral markets. Whilst multilateral netting is clearly more beneficial when all trades are covered, in reality fragmentation or bifurcation will be a problem. Two obvious sources of fragmentation are non-clearable trades (which remain bilateral) and multiple CCPs. Such a situation is illustrated in Figure 6.17, where some of the positions are assumed to be cleared outside the CCP shown.

Schematic illustration of the comparison of no netting, bilateral netting, and partial central clearing where only a subset of trades can be centrally cleared. — **Figure 6.17** Comparison of no netting, bilateral netting, and partial central clearing where only a subset of trades (black lines as opposed to grey ones) can be centrally cleared.

Table 6.4 Illustration of an increase in overall exposure caused by multilateral netting related to central clearing of only a subset of trades, as shown in Figure 6.17.

	No netting	Bilateral netting	Partial central clearing (excluding CCP positions)
Counterparty 1	170	50	100
Counterparty 2	90	20	90
Counterparty 3	100	50	20
Total	360	120	210

The quantitative impact of partial multilateral netting is shown in Table 6.4, which considers the total exposure under no netting, bilateral netting and partial central clearing. Even ignoring the exposure involving the CCP itself (i.e. assuming the CCP is risk free), the overall reduction in exposure is better with bilateral netting (total exposure 120) than with partial central clearing (total exposure 210). For example, with no netting, counterparty 1 has a total exposure of 170 (70 to counterparty 2 and 100 to counterparty 3), and under bilateral netting, this is reduced to 50 (to counterparty 3 only). However, under partial central clearing, counterparty 1 gains in some multilateral netting of their positions with counterparties 2 and 3, but loses the bilateral netting of the two sets of positions.

Note that the above example could correspond to a situation where certain trades cannot be centrally cleared, or alternatively where they are cleared via a separate CCP to the one shown. The above example illustrates that the loss of bilateral netting benefits may dominate the increase in multilateral netting ones and result in central clearing, increasing the overall exposure in the market. This splitting of netting sets is analysed with some simple examples by Duffie and Zhu (2009). Their results are based on considering the netting benefit for trading a single class of contracts through a CCP as opposed to bilateral clearing. They show, using a simple model,²¹ the required number of members trading through the CCP for a single asset class to achieve overall netting reduction. Overall, the Duffie and Zhu results illustrate that achieving overall netting benefits from central clearing (compared to bilateral trading) is not a foregone conclusion. Increased netting benefits can only be achieved by a relatively small number of CCPs clearing a relatively large volume of transactions.

In theory, the bilateral margin requirement does not share similar bifurcation problems as central clearing because transactions are still bilateral between the parties involved. However, since most counterparties have chosen to create new margining/collateral agreements in order to comply with such rules for new transactions, without affecting existing transactions, there is a bifurcation of collateral across these two agreements. There may even be a possibility of legal problems if such transactions would be deemed to be bifurcated across two different netting sets.

6.4.4 Netting Impact on Other Creditors

Close-out netting may seem very beneficial in OTC derivatives markets where it reduces exposure and potentially leads to easier close-outs. However, for financial markets generally, it merely redistributes value to OTC derivatives creditors from other creditors. Consider a generalisation of the example in Figure 6.10 to include other creditors. In Figure 6.18, party B has both derivative creditors (party A) and other creditors (OC).

Party B defaults with total assets of 180 (140 derivatives and 40 other) and total liabilities of 300 (200 derivatives and 100 other). Without netting, assuming other creditors and derivative creditors have the same seniority,²² a recovery of 60% (180/300) would apply, and the payments would be as on the left-hand side of Figure 6.19. However, if the derivatives contracts are subject to netting, as illustrated on the right-hand side of Figure 6.18, then the liabilities become 60 and 100 for derivatives and other creditors respectively against the assets of 40. This leads to a lower recovery of 25% (40/160) for the other creditors. The derivatives creditors receive a total of 155: 140 from being able to net their assets and liabilities and 15 from a recovery amount related to their netted claim. This leads to an overall recovery of 77.5% (155/200), whereas the other creditors receive 25% (25/100) (Figure 6.19, right-hand side).

Schematic illustration of an example of bilateral derivatives netting, including other creditors. — **Figure 6.18** Example of bilateral derivatives netting, including other creditors.

Schematic illustration of an example of bilateral derivatives netting, including other creditors, showing payments made in default of party B, assuming party A and other creditors are paid the same percentage recovery. — **Figure 6.19** Example of bilateral derivatives netting, including other creditors, showing payments made in default of party B, assuming party A and other creditors are paid the same percentage recovery.

The above example illustrates that bilateral netting of OTC derivatives increases the recovery for OTC derivatives counterparties (77.5% instead of 60%, in the above example) but reduces the recovery of other creditors (25% instead of 60%). This potentially highlights a much broader point, which is that certain benefits (netting, margining, central clearing) may be positive for OTC derivatives markets but not necessarily for financial markets in general since they merely redistribute risk (Pirrong 2014). Netting may reduce exposure to OTC derivatives counterparties but increase exposure to other creditors (e.g. bondholders). A bank may reduce its derivatives counterparty risk (and capital) through netting, but this may induce changes in other parts of the balance sheet of the bank. This could pose the question as to whether reducing systemic risk in derivatives markets at the expense of increasing systemic risk elsewhere is a worthwhile trade-off.