Part II Commercial Banking, 7 Credit Risk
- Banks as a lender — Credit risk — Default and credit — Repayment of credit — Debt
7.01 Of all the risks that banks are exposed to, credit risk is the most important and the most intuitively obvious. A one-sentence definition of a banker might be a man who judges credit— a judgement which can be characterized as a decision on how much to lend to whom on what terms. It is, however, important to remember that credit means more than simply loans. If I lend money, I have a credit exposure to the borrower. However, if I buy debt securities I have a credit claim on the issuer; if I enter into a derivative I have a credit claim on the counterparty; if I enter into an agreement to lend money at a future time (a commitment) I still have a contractual exposure to the credit of the borrower. Credit exposures are at the heart of financial transactions. For an economist, the function of a bank is maturity transformation and intertemporal transfers of resources, but in a world where debts were always repaid these functions would be as mechanical as the transmission of water or electricity. It is the unpredictability of credit that differentiates banking from sewerage as a business.
7.02 Finance is, however, about more than credit. The old banker’s maxim ‘lend on the credit, not the security’, is often broken in modern banking, and transactions are frequently encountered which are financial transactions but not credit transactions. To take a simple example: a loan to a special purpose vehicle whose only (p. 112) asset is an office block is not, in any meaningful sense, an exposure to the credit of the borrowing entity; it is an exposure to the value of the underlying asset (or, to be precise, to the cash flows which that asset may generate). Thus, although most bank assets are credit exposures, some are not, and there is more to bank capital adequacy than simply weighting credit. However, credit does remain the single most important component of the system.
7.03 Although we are accustomed to think of credit as arising from loan exposures, credit risk goes well beyond loans. In broad terms almost any financial transaction will involve a credit risk, and sometimes more than one credit risk may arise out of the same transaction. The regulatory system distinguishes between simple credit exposures and complex credit exposures—a simple credit exposure is an exposure which arises out of future payment obligations, where the amount due is certain and it is only the likelihood of repayment which varies, and a complex credit exposure is an exposure which arises out of a transaction (such as a derivative) where both the amount due and the likelihood of repayment may vary over the life of the transaction. This section deals (broadly) with simple credit exposures, where the amount of the exposure can be either known or easily estimated. Complex exposures, which typically arise under derivatives, securities lending, or long settlement transactions, are dealt with in the chapters on investment banking.
7.04 The basis of bank capital regulation is that banks should be required to maintain sufficient capital to make it unlikely that they will become insolvent in the event of a number of credit defaults. The issue for regulators is to determine the sizes and the numbers of the defaults which a bank should hold capital against. Clearly this is in some respects an arbitrary determination, since there is no mechanism that can determine with certainty how many credit exposures will default. However, although the chances of everything defaulting simultaneously can be disregarded, the chances of a number of different exposures defaulting at the same time cannot be—indeed, this is the most common explanation for bank failures. Consequently, the challenge for regulators is to decide the level of unlikelihood which they are seeking to protect against. The fundamental mechanism by which this is done is to determine the risk that any individual asset will default, multiply that risk by the size of the relevant exposure, and require that amount of capital to be held in respect of that asset.1 The starting point for the Basel approach is the arbitrarily determined figure that every asset has an 8 per cent risk. However, (p. 113) this may be increased or reduced by ‘weighting’ the exposure concerned—thus a mortgage which attracts a 50 per cent weighting is taken to have a 4 per cent risk of default, and so on.
7.05 Under the original Basel framework, there was a single calculation methodology applied to all exposures. This had the important consequence that all banks calculated their exposures in the same way; and the same asset would have the same regulatory capital weighting for all banks. Neither of these are true under the Basel II system.
7.06 Basel II gives banks three ways to calculate the credit weightings to be applied to assets.2 The first—the ‘standardized’ approach—employs a prescribed set of weightings based on asset type and external credit ratings. The second—the ‘internal ratings based approach’, known as IRB—permits banks to use their own statistical models to generate weightings. The IRB approach is then subdivided into two further approaches, ‘foundation’ and ‘advanced’, depending on the sophistication of the models operated by the relevant bank.
7.07 Each bank must decide whether it is a standardized bank, a foundation IRB bank, or an advanced IRB bank. Broadly, once a bank has classified itself it may elect to use a lower approach for some exposures, but may not elect a higher one—thus an advanced IRB bank can elect to use a standardized approach for certain exposures, but a standardized bank cannot elect to use the advanced IRB approach for any exposures. The general policy of regulators is to encourage all banks to progress towards advanced IRB status, but it is accepted that the advanced modelling capabilities required for advanced IRB status may not be cost-effective for some classes of institutions to develop.
7.08 In order to explain the basic distinction between the three approaches it is necessary to explain a little about the basics of weighting. In theory, it is possible for any given exposure to estimate probability of default (PD), loss given default (LGD) and the exposure at default (EAD). PD and LGD are reasonably self-explanatory—PD is the probability of a default occurring on a particular facility, LGD is the estimate of the loss which will be suffered if there is a default (ie the amount by which any recoveries eventually made will fall short of the amount of the exposure). EAD is a little more complex. For a facility such as a term loan, EAD is simply the amount of the loan—on a loan of £100, the expected exposure at default is £100. However, for a revolving credit facility of £100, the amount drawn at any given time may fluctuate, and for such exposures it is also necessary to make an estimate of the amount which is likely to have been drawn when a (p. 114) default occurs. This is the EAD. EAD also has to be estimated in the context of undrawn commitments, derivative exposures, and a variety of other exposures.
Bank capital requirements must clearly be sufficient to meet expected losses. However, in order to protect depositors, it is also necessary for banks to have sufficient capital to be able to withstand unexpected losses. It is clear that there is no limit to the scale of potential unexpected losses. However, by setting appropriate parameters it is possible to quantify what can perhaps best be described as ‘probable unexpected loss’. The ultimate aim of the bank regulatory capital system is to ensure that a bank has sufficient capital to cover both expected loss and probable unexpected loss. In the IRB world this ‘probable unexpected loss’ amount is derived by using the PD as one of the inputs into a formula, the full glory of which is set out in para 9.08. However, for our purposes we can represent it as f(PD).
f(PD) is always larger than PD, and unexpected loss (UL) is therefore always larger than EL.
7.10 The basis of risk weighting is to express UL as a percentage. In the Basle I world, the starting assumption was that the UL for all assets was 8 per cent, but this 8 per cent was varied by a factor (the ‘weighting’) which represented the level of riskiness of the relevant asset—thus government securities were weighted at 0 per cent, mortgage loans were weighted at 50 per cent (equivalent to a UL of 4 per cent) and so on. The weightings used in the Basle I approach (10 per cent, 20 per cent, 50 per cent, and 100 per cent) were therefore a composite estimate of PD and LGD in respect of each asset—EAD was dealt with through the credit conversion factor (CCF) regime.
7.11 The Basel II regimes are distinguished by the fact that they arrive at the weighting figure in different ways. In the standardized approach, PD and LGD are not visible, since the prescribed weightings embed them. EAD remains visible in the form of the prescribed CCF figures. Within the IRB approach, in contrast, PD, LGD, and (for advanced IRB banks) EAD must be calculated and used. The difference between the IRB approaches is that under the foundation approach the bank uses its own models to calculate PD, while under the advanced IRB approach it calculates all three. This can be expressed diagrammatically as shown in Table 7.1.
7.12 It may appear from Table 7.1 that advanced IRB banks will be free to set their own capital requirements in respect of unexpected loss, but this is not quite the case. Although advanced IRB banks are permitted to determine all of the inputs (p. 115) to the formula to be used for calculating capital requirements, they are still required to use the formula and the methodology prescribed by the regulator. The Accord also prescribes operational requirements in respect of the systems used by banks to make calculations for the purposes of the IRB approaches. There will therefore continue to be substantial differences between the capital requirement of an advanced IRB bank and its own assessment of its risk position.
7.13 The calculation of the capital requirement for any exposure is a matter of ascribing a weighting to that exposure—in other words, multiplying the value of the exposure by a percentage figure to arrive at a weighted valuation. The bulk of this chapter concerns the calculation of the weighting percentage, but before considering this we should first think about the number to be weighted—the value of the exposure.
7.14 Even the valuation of simple exposures is by no means always straightforward. It is a fundamental principle of the Accord that the value ascribed to an exposure for capital purposes must be the value given to it in its financial accounts. However, this is not as simple a rule as it may appear. UK firms, for example, may account using either International Accounting Standards or UK Financial Reporting Standards and Statements of Standard Accounting Practice, and some may account in accordance with insurance accounts rules, friendly societies accounts rules, building societies accounts rules, or various approved statements of recommended practice. However, the key issue is that the fundamental valuation for any asset is what the auditors say it is. This imports the applicable accounting rules to determine issues such as:
• netting of amounts due to and from the firm;3
• the effect of securitization of assets and liabilities;
• leasing of assets;
• assets transferred and received by way of initial or variation margin under a derivative or similar transaction.
7.15 However, it is important not to confuse the rule that an exposure has its accounting valuation with the proposition that the capital system follows accounting rules. Once exposures have been identified and valued using the applicable accounting principles, they must be dealt with according to the relevant regulatory capital rules. Thus, for example, the question of whether an asset is part of the trading book or not is a regulatory issue which is determined using regulatory rules, and this determination may well come to a different conclusion than the accounting analysis as to whether the relevant asset is held on an available-for-sale basis. It is therefore possible that there may be assets valued for accounting purposes as available for sale which are held in the banking book.
7.16 The issue of marking exposures to market is one which has recently attracted considerable controversy, and the discussion here is confined to the mechanics of the requirements of the current rules. A short summary of the position as it is in the UK is that for institutions reporting under International Accounting Standards, assets are divided into four classes.
Financial assets at fair value through profit or loss
7.17 Where an asset is held at fair value through profit and loss, it is accounted for at ‘fair value’, and fair value is defined for this purpose as ‘the amount for which an asset could be exchanged, or a liability settled, between knowledgeable, willing parties in an arm’s length transaction’.4
7.18 What this means in practice is that if the value of the asset has increased, the holder books an unrealized profit on the asset, and if it has gone down, the holder books an unrealized loss. This profit or loss is recognized in the profit and loss account and is ultimately reflected in an increase or decrease in the accounting capital of the entity.
7.19 An asset can fall into this category in one of two ways. One is if it is a derivative (except designated hedging instruments) or a financial asset acquired or held for the purpose of selling in the short term or for which there is a recent pattern of short-term profit taking held for trading.5 The second is by designation. The original principle enshrined in the accounting standard was that when a firm (p. 117) acquired any asset, it was required to make a once-for-all determination as to whether the asset was to be held in this category or not. Thereafter the asset could not be reclassified. This rule was intended to prevent firms from avoiding recognizing unrealized losses by reclassifying assets out of the fair value through profit and loss category. However, under the stress of the credit crunch this rule was changed to provide that firms were permitted to reclassify assets out of this category provided that the reclassification was the result of a genuine determination that the asset would now be held to maturity, and that such redesignation occurred only in ‘rare circumstances’.6
Available-for-sale financial assets
7.20 These are non-derivative financial assets designated on initial recognition as available for sale, and are measured at fair value in the same way as financial assets at fair value through profit and loss. The difference between the two classes is that changes in the value of available-for-sale assets are not taken to the profit and loss account, but are recognized directly in equity, through the statement of changes in equity. Thus if a bank makes a gain or loss on an asset which is held as available for sale, its total accounting capital is raised or lowered accordingly, but its profit or loss will remain unchanged.
Loans and receivables
7.21 Loans and receivables may not be marked-to-market, and must be accounted for ‘at amortized cost using the effective interest method’. What this means in broad terms is that if you lend £100 the asset value of the loan is likely to be £100 for as long as the loan remains fully performing, and this will be true regardless of fluctuations in interest rates. The full calculation rule is as follows: amortized cost of a financial asset or financial liability is the amount at which the asset or liability is measured at initial recognition (usually ‘cost’) minus any repayments of principal, minus any specific provision or write-off, and plus or minus the cumulative amortization of the difference between that initial amount and the maturity amount calculated using the effective interest method. This calculation is made by looking at the cash flows involved in the transaction and treating any variation as an amortization or overdrawing. Thus if a five-year facility of £100 with an effective interest rate of 10 per cent provided for no repayment of interest at all in the first year (but a double payment in the final year), at the end of the first year the facility would be treated as having increased in value to £110. Conversely if it provided for a double (p. 118) payment in the first year and no payment in the final year, it would be regarded as having dropped in value to £90.
Held-to-maturity investments other than loans and receivables
7.22 This class comprises all assets held by a bank other than those designated as financial assets at fair value through profit and loss, and loans and receivables. They are valued in the same way as loans and receivables.
7.23 Wherever possible, a firm must use a ‘mark-to-market’ approach in order to arrive at fair value. ‘Marking to market’ means, for this purpose, valuing positions at readily available close-out prices from independent sources. Where securities are traded on an exchange, then exchange closing prices should be used, but where this is not the case, screen prices or quotes from independent brokers satisfy the requirement. Marking to market should be done at the price at which a firm could deal—thus positions should normally be marked at the buy or sell price of the dealing spread (as appropriate) unless the firm is a significant market maker in a particular position type or is otherwise satisfied that it could close out at the mid-market.
7.24 Where marking to market is not possible, a firm must use mark-to-model. ‘Mark-to-model’ means any valuation which has to be benchmarked, extrapolated, or otherwise calculated from a market input. Where a bank wishes to mark its positions to model, it should be able to demonstrate to its regulator that the model concerned is suitably robust. There is no formal threshold within the regulatory system as to when mark-to-model valuations may be used—in practice, accounting conventions are followed.
7.25 Finally, in valuing any individual asset banks are required to deduct any asset recognized in respect of deferred acquisition costs and add back in any liability in respect of deferred income (but exclude from the deduction or addition any asset or liability which will give rise to future cash flows), together with any associated deferred tax.
Regulatory principles for asset valuation
7.26 It is a general principle of prudential supervision that assets are valued according to the appropriate accounting policies applied by the bank concerned unless the prudential regulatory system actively requires a different approach to be adopted. This is partly as a result of the fact that it is a regulatory obligation that banks provide timely, relevant, reliable, and decision-useful information to the market and to counterparties. This includes descriptions of valuation techniques used to determine fair value and the instruments to which they are applied.
7.27 Certain assets are recorded on a bank’s balance sheet at historic value, and do not subsequently change. However, for the majority of assets current accounting practice is to apply a fair value—ie a value which is derived from an independent market value or, where such a value is not available, from a valuation model.
7.29 The basic requirement for valuation is that it be subject to appropriate formal governance structure which is overseen by the bank board. In practice this means that the board must ensure that the valuation function within the bank is appropriately resourced and sufficiently independent, and that its output is monitored. In this context, resourcing needs to be sufficient to enable the valuation desk to produce very large numbers of valuations in a short space of time in the event of market disruption. Valuation desks also need to have the ability to access and process alternative valuation mechanisms in the event that primary markets become illiquid or otherwise unreliable. The use of third-party valuation service providers is permissible, but appropriate controls need to be put in place as regards the reliance on—and reliability of—such suppliers.
7.30 Valuation is generally done by reference to public market data. Where there is an active market for an asset with regular trades in significant size, this presents few difficulties. For less active markets, however, thoughtful review is needed of the available market data. When a market is completely inactive, fair value can only be established using a model. However, where there are infrequent transactions, a transaction price in the same or a similar instrument should not be automatically considered to reflect fair value. If such transaction prices are used in valuation, they might require significant adjustment based on unobservable data.
7.31 Valuation controls should be applied consistently across similar instruments (risks) and across business lines (books). A key feature of an adequate control process is that the final approval of valuations should not be the responsibility of the risk-taking units. This means that banks must maintain functional separation between the front office (the risk-taking units that typically provide the initial fair valuation estimates) and the measurement and control unit (the unit providing independent price verification or IPV) at all times. In addition, the unit responsible for IPV within the bank should not source prices from the relevant trading desk.
7.32 There is no absolute rule to the effect that regulatory and accounting valuation treatment must always coincide. Supervisors expect that a bank will initially categorize and report financial instruments in financial reports in accordance with applicable accounting and regulatory reporting requirements, but this may change. However, any significant differences in categorization for the measurement and management of risk and that are necessary for the applicable accounting framework should be properly documented and approved by senior management.
7.33 Reclassification of instruments may be a consequence of change in a bank’s strategy—thus an instrument which was initially acquired in order to be held to maturity may be reclassified as available for sale and vice versa. In such a case, reclassification should be strictly in accordance with accounting requirements.
(p. 120) 7.34 In some circumstances, adjustments may be necessary to result in a valuation estimate that meets the applicable valuation definition. Accordingly, the overall governance and control framework for valuations should include a policy to identify the types of valuation adjustments. Valuation adjustments should be initially authorized and subsequently monitored by an independent control group (for example an IPV or financial control unit, and/or independent model validation unit), and supported by appropriate documentation.
Prudential value adjustments
7.35 Paragraphs 698–701 of Basel II mandate banks to establish processes for considering valuation adjustments where marking assets to market or to model. The essence of adjustments of this kind is to take account of the possibility of life-cycle events which could result in the return to the bank being less than that which the valuation suggests. Particular issues which should be considered include unearned credit spreads, close-out costs, operational risks, early termination, investing and funding costs, future administrative costs, and, where appropriate, model risk.
7.36 These adjustments should also take into account the realities of closing or liquidating a position. The minimum ‘holding period’7 used in the value at risk (VaR) calculation is ten days, but in certain circumstances this may prove over-optimistic. Banks are therefore also required to consider whether further adjustments should be made to positions which, in stressed market conditions, might take longer than this to unwind. This requires consideration of the amount of time it would take to hedge out the position/risks within the position, the likely average volatility of bid/offer spreads or independent market quotes (ie the number and identity of market makers), the average and volatility of trading volumes, market concentrations, the ageing of positions, the extent to which valuation relies on marking to model, and the impact of other model risks.
7.38 Bank capital is held against unexpected losses—ie those losses which have not yet been reflected in the balance sheet of the bank concerned. That means that there is a necessary interaction between regulatory capital requirements and loss provisioning. As regards specific provisions—ie provisions reflecting losses which are regarded by banks as certain—there is general agreement that these should be treated as incurred (p. 121) losses. However the position as regards general provisions—ie provisions in respect of the probability of future losses—the position is more difficult.
7.39 Under the pre-crisis accounting orthodoxy, the accounting rules effectively said that provisions should only be taken where the loss concerned was highly likely to be incurred. This meant that the level of these provisions was reasonably low. By contrast, the regulatory capital system embedded an EL model under which losses were to be modelled statistically and the EL amount was to be deducted from regulatory capital. This meant that accounting provisions could reasonably be assumed to be lower than the EL figure derived from regulatory calculations.
7.40 This has all changed since the crisis of 2008. In particular, IFRS 9 and the Financial Standards Accounting Board (FASB) current expected credit loss (CECL) standard have the effect of moving bank accounting onto an expected loss standard. The issue for bank regulators is therefore as to how to combine accounting and EL calculations into a seamless whole.
7.42 ECL is a forward-looking calculation—the bank is required to calculate what the probable loss might be in respect of every asset and to take a charge in respect of that asset, even where the asset is currently completely unimpaired.
7.43 The FASB ECL calculation requires a measure of lifetime ECL for all exposures. The IFRS 9 measure is more formalized. Each asset is allocated to one of three buckets—roughly good standing, deteriorated, and impaired. The impaired bucket is similar to the old IAS 39 impaired classification, both in terms of trigger and in terms of the mechanisms applied to calculate loss. Thus the real difference between the old and the new system is that in the old system there were really only two states for an asset to be in—whole or impaired. In the new system there is no ‘completely unimpaired’ class—every asset is subject to some impairment charge, no matter how small. However, unimpaired assets are divided into those which have not experienced ‘significant credit deterioration’ since origination, and those which have. For those which have not, a charge will be taken to reflect the likelihood of the asset defaulting in the next twelve months8—this charge will be very low (although non-zero9). For those assets which have suffered deterioration, an ECL charge is taken which is intended to be equal to the expected loss over the entire life of the asset. This charge is likely to fluctuate significantly as credit risk improves or deteriorates over time. The effect of this in practice is likely to be that (p. 122) when an asset becomes impaired, the ECL will increase over time as the credit deteriorates, until it reaches formal default. At that point, the ECL is likely to be more or less equal to the impairment charge, so the impact on the bank’s accounts of declaring the asset formally impaired should be minimal.
7.44 Basel was relatively late to the IFRS9/CECL party. The implementation date for the IFRS9 changes is 1 January 2018,10 but by February 2015 Basel was still only consulting on possible policy responses. The reason for this is almost certainly that at first blush it was not obvious that the accounting changes would have any effect at all on the regulatory capital calculation. This is because the accounting ECLs were to be assessed on a neutral economic cycle whereas regulatory LGDs and EADs are assessed using downturn estimates. Thus the initial assessment would have been that regulatory EL would always be larger than accounting ECL. However, on more detailed consideration it became clear that the fact that accounting LGD/EAD was always lower than regulatory LGD/EAD did not necessarily mean that accounting ECL would always be lower than regulatory EL. This is because EL is LGD/EAD multiplied by PD, and the fact that the accounting assessment of PD—for all assets in the US approach, and for stage 2 and 3 assets in the IFRS approach—was based on lifetime PDs meant that the implied PD in the accounting calculation could well be significantly higher than regulatory PD, which is based on a twelve-month period.
7.45 Neither is the integration of the IFRS9/CECL approach into the regulatory system entirely straightforward. Under IFRS 9, a rise in impairment depletes the capital adequacy of banks that use the standardized approach to credit risk, as the 1:1 reduction in capital arising from increased impairments is not offset by reduced risk-weighted assets (RWAs). The result is less clear-cut for IRB banks, reflecting the more complex relationship between impairment and the outcomes of the IRB capital formula. However, since IRB banks are subject to a standardized approach (SA) floor, it is likely that they will end up with a ‘worst of’ treatment.
7.46 Basel has approached this by proposing an interim arrangement that in effect reverses out the impact of this accounting change for an interim period. The objective of this is to avoid a ‘capital shock’ and provide time for banks to rebuild their capital reserves—as a result, the maximum transition period is fixed at five years.
7.47 Reversing out the ECL figure is relatively easy—the problems come when questions are raised as to how it should be replaced during the interim period, since that replacement figure must itself be sensitive to changes in the provisioning of assets during that period. The problem here is that the existing Basel treatments (p. 123) rely on a distinction between specific provisions (SP) and general provisions (GP) which existed under the old accounting standards but does not exist in the IFRS9/CECL world. Consequently, as part of the interim arrangements regulatory authorities will have to give guidance to banks as to how GP and SP are to be calculated for regulatory purposes now that they are no longer calculated for accounting purposes. Also, the transitional provision is intended to apply only to ‘new’ provisions—ie provisions resulting from the implementation of ECL accounting, and should not have the effect of reducing provisions already put in place under the existing regime.
7.48 The transitional approach adjusts the Common Equity Tier 1 figure to reverse out the impact of ECL. This impact is then phased in over the transition period. Basel’s preference is for the interim relief to be a one-off figure amortized over a straight-line basis. However, an alternative approach would involve working out the exact difference between the IFRS provisions and the original provision amount pre-implementation and then scaling that by a figure—thus, in the first year of implementation, provisions would be taken to be original provisions plus 25 per cent of the increase between original provisions and IFRS 9 provisions, with the figure being 50 per cent the following year, 75 per cent the year after that, and the adjustment falling away in the last year. It is also notable that there are a number of other consequences of this disapplication—for example disapplied provisions must be reversed out of the exposure measure for the leverage ratio or SA, and should not be included in Tier 2 capital.
7.49 Step-in risk is created where there is an association between a bank and another entity which does not give rise to a consolidation relationship,11 and as a result the assets of the other entity do not appear on the balance sheet of the bank concerned. However, there are circumstances in which a bank may regard itself as committed to support such unconnected entities. There are documented examples of this having happened during the crisis with sponsored asset-backed commercial paper (ABCP) vehicles, money market funds, and some other types of financing vehicles. Consequently, the starting point for the step-in risk analysis is the idea that there may be assets which are currently not on the bank’s balance sheet, but in respect of which it might have credit risk by reason of the way in (p. 124) which they are held. The aim of the analysis is to identify such assets. It should be noted that the outcome of a positive step-in risk analysis is not automatically the inclusion of these assets on the regulatory balance sheet—the aim of the analysis is to assess the extent of the bank’s exposure to the risk, and the outcome should be an accurate reflection of that exposure.
7.50 The regulators begin with the (undoubtedly correct) position that there is no way in which a closed list of vehicle types which give rise to step-in risk can be identified. Consequently, the approach is to create a series of risk indicators that should be applied to any entity with which the bank is associated to identify whether step-in risk exists.
7.51 This means that the bank must begin by identifying the universe of all separate non-consolidated vehicles in respect of which a potential step-in risk might exist. It is then required to assess those entities against a set of regulatory indicators. Where these indicators indicate the existence of a risk, the firm must estimate the potential impact on its liquidity and capital position and determine a potential response. The bank must then self-report its assessment to its supervisor, which may take further supervisory measures to address the risk.
The step-in risk boundary
7.52 The starting point for step-in risk is that any entity which is included in the accounting but not the regulatory consolidation must be regarded as a potential step-in risk—thus any entity which satisfies an accounting ‘control’ test is within the framework. However, any entity in respect of which the relevant bank is the sponsor or a long-term debt or equity investor, or has exposure to equity-like risks in respect of the vehicle is potentially included. There is no closed list of possible entities, but the Basel-suggested list includes:
• covered bond issuing entities;
• collateralized debt obligation (CDO) and collateralized loan obligation (CLO) issuing entities;
• cash CDOs;
• synthetic CDOs;
• entities issuing tender option bonds;12
• ABCP issuers;
• securities arbitrage conduits;
• structured investment vehicles;
• repackaging vehicles;
• real estate investment trusts;
• mutual funds;
• exchange-traded funds;
• hedge funds;
• private equity funds;
• finance companies; and
7.53 After this ‘long list’ has been drawn up, a series of exclusions are applied. Insurance companies are excluded, but securitization vehicles are not, even if they are already fully covered by the existing securitization rules—the step-in risk in such cases must be considered de novo. Non-financial commercial entities may be excluded, although consideration may be needed in the case of vehicles providing critical operational services to the bank. Entities may also be excluded if they are so small that the impact of a step-in would be immaterial. Finally, and importantly, institutions can be excluded where national law explicitly prohibits banks generally from stepping in to support certain entities. The prohibition must be general and explicit (such prohibitions are referred to as ‘collective rebuttals’).
(1) Nature and degree of sponsorship. A bank may be exposed to a greater degree of step-in risk in certain cases, such as when it: (i) provides full sponsor support, via a guarantee or other credit enhancement; or (ii) provides partial credit enhancements and liquidity facilities while playing a role in decision-making.13
(3) Implicit support. This indicator takes into account whether investors’ rate-of-return expectations suggest that in reality a degree of credit support is being factored into the pricing on the relevant instruments. This is likely to be the case where the return to investors is significantly lower than would ordinarily be expected.
(4) Structured entities/variable interest entities and highly leveraged entities. This is intended to capture entities which are exceptionally highly leveraged, and where control may be exercised through methods other than traditional voting rights.
(5) Liquidity stress/first-mover incentive. Off-balance sheet entities that engage in maturity transformation by holding non-risk-free assets (ie those other than sovereign-backed bonds) increase the potential for step-in, as the sponsor may (p. 126) be required to provide liquidity to such a vehicle by purchasing assets from it. However, this potential is reduced if the assets that the vehicle holds are cash reserves or high-quality liquid asset (HQLA) equivalents because their increased liquidity needs during stress periods would already be covered, at least in part. This risk is heightened where the vehicle is at risk of runs—ie where it is advantageous for an investor to exit the entity before others do, and there are no potential barriers to redemptions (ie redemption gates).
(7) Accounting disclosures. Accounting disclosure requirements can provide meaningful information to evaluate the nature and risks of a bank’s involvement with unconsolidated entities. Thus, where exposures to unconsolidated entities are disclosed under IFRS 12 or US GAAP VIE disclosures, or comparable disclosures made under the relevant accounting regime, this may indicate an increased step-in risk.
(8) Investor risk alignment. This indicator refers to entities whose activities do not match the risk profiles of their clients/investors with those of the risk exposures of the entity. This risk may be created where banks provide investment products to investors who are loss-averse (confidence-sensitive products) or sell to retail customers products that have bundled, hard-to-price features (such as subordinated debt, preferred shares, or debt instruments with embedded optionality).
(9) Reputational risk from branding and cross-selling. Where a third party sells products using a bank’s brand, that fact could strengthen the presumption of step-in support, especially if the brand is associated with a deposit-taking institution in the same banking group. The evaluation of this indicator should consider the degree to which cross-selling is part of the bank’s overall strategy, as a greater degree of cross-selling increases reputational risk and, thus, the incentive to provide step-in support.
(11) Regulatory restrictions and mitigants. This is a counter-indicator, and is triggered where banking, securities, market, or other financial regulations restrict, without prohibiting, a bank’s ability and/or propensity to support an entity on terms that are unfavourable to the bank. An example would be the regulatory regime constituted by ss 23A and 23B of the US Federal Reserve Act14 and Regulation W.15
Inclusion in regulatory consolidation
7.56 This is a backstop measure, intended for cases where there is a clear investor expectation of step-in. However, consolidation itself is by no means the optimum solution in all cases—indeed, since it has the effect of presenting the assets of the consolidated entity as being available to the parent, it may be actively misleading.
7.58 Both liquidity coverage ratio (LCR) and net stable funding ratio (NSFR) contain provisions that permit the inclusion of informal or non-contractual obligations to buy back assets or advance liquidity. If a step-in risk is felt to exist, those provisions should be triggered in respect of the relevant exposures.
Punitive ex post charges
7.61 If a bank steps in in respect of a vehicle where no step-in risk was deemed to exist, the supervisor may seek to ‘punish’ the bank for this by imposing punitive capital charges after the event—for example by risk weighting the assets brought onto the bank’s balance sheet at a considerably higher level than would ordinarily be the case. The threat of this measure may itself deter step-ins.
7.62 This measure is similar to that adopted by the European Banking Authority (EBA).16 It requires a bank to apply an internal limit to the total of all of its contractual exposures and/or other estimations of exposures to shadow banking entities.
7.64 The topic of interest rate risk in the banking book has troubled bank regulators for a very long time. The fundamental issue is easily stated—when interest rates change, the present value of future cash flow changes, and this in turn changes the underlying value of the bank’s assets and liabilities. This risk would not be mitigated even by immediate marking-to-market of the relevant assets and liabilities, since the immediate impact is on cash flows, and such impacts may be seriously damaging in the short term regardless of balance sheet position. Changes in interest rates affect banks’ earnings by changing the relationship between interest received and interest paid—at its simplest, if a bank has long-term fixed rate assets and short-term or floating rate liabilities, increasing interest rates will reduce net interest income. Consequently the banking book business of a bank may be seriously harmed by changes in interest rates, and this risk is known as interest rate risk in the banking book (IRRBB).
7.65 It is debatable whether IRRBB is properly characterized as credit risk—it may be that it is better viewed as the risk cost of assuming credit risk, in the same way that credit valuation adjustment (CVA) can be viewed as the risk cost of assuming trading book positions. However, whereas regulators have found it possible to quantify CVA risk sufficiently to reduce it to a capital charge, the same has not proved to be true of IRRBB.
7.66 IRRBB is generally subdivided into three classes: gap risk, which describes the state where the interest rate on assets changes discontinuously from the interest rate payable on liabilities; basis risk, which describes the impact of relative changes in interest rates of similar tenors that are calculated by reference to different indices; (p. 129) and option risk, which arises from optional elements embedded in banks’ assets or liabilities—this includes early repayment, extension, and/or prepayment arising from both automatic options and behavioural options. There is also a further class of risk, known as credit spread risk in the banking book (CSRBB), which captures jump-to-default risk and any other asset/liability spread risk.
7.67 Although there is broad agreement that IRRBB exists, there is less agreement as to how it should be approached by bank supervisors. The Basel Committee initially proposed a standardized pillar 1 capital charge to capture IRRBB,17 but this proved to be immensely hard to calibrate, and what it has ended up recommending is a pillar 2 charge based on risk assessment. This charge is to be based on shock and stress scenarios carried out by banks as part of their stress-testing regime. Banks are also required to maintain key behavioural and modelling assumptions and an internal validation process, along with compliance with pillar 3 disclosure obligations. With regard to this last requirement, banks are required to disclose ∆EVE (impact of interest rate shocks on the change in economic value of equity) and ∆NII (impact of interest rate shocks on the change in net interest income). A standardized framework for IRRBB is also proposed for cases where supervisors do not believe that the bank’s internal risk management system adequately captures IRRBB. There is a presumption that this should be applied if a bank’s ∆EVE is greater than 15 per cent of Tier 1. This approach is set out in the relevant Basel paper,18 but is unlikely to be applied in practice and is not reproduced here.(p. 130)
1 As explained in greater detail elsewhere, the Basel calculation takes into account unexpected as well as expected loss—hence the capital requirement under Basel in respect of an asset is greater than the expected loss figure in respect of that asset.
2 Technically four, since there is a simplified standardized approach available for institutions of such simplicity that even the standardized approach is unreasonably demanding. However, this is a relative rarity in practice, and is not discussed further in this work.
3 There is of course a separate set of regulatory netting rules. The point here is that accounting rules trump regulatory rules in this regard: if the accountants apply a netting approach to an asset, there is no need to apply—or even consider—the regulatory rules.
6 IAS 39, new para 50B, inserted 1 July 2008. The standard also required that the transfer be at the then prevailing market valuation—in other words, accrued mark-to-market losses prior to the transfer would be made permanent by the transfer, regardless of any subsequent upward movement in the price of the instrument.
8 Note that a twelve-month ECL means the total potential loss on a default multiplied by the probability of a default in the next twelve months, not the expected loss incurred over the next twelve months in the event of a default.
9 The only case where a zero EL figure is envisaged for accounting purposes is fully collateralized loans. However, even here, potential variation in collateral value may have to be taken into account.
11 For this purpose, we include assets owned by vehicles which do appear on the accounting balance sheet, but have been derecognized for regulatory purposes because the significant risk transfer criteria have been met.
13 In most cases, the provision of full credit or liquidity support to a structured vehicle would result in accounting consolidation. Any exclusion of such entities from regulatory consolidation would thus be the result of regulatory guidance allowing or requiring such treatment (for example Art 248 of the EBA Capital Requirements Regulation on significant risk transfer securitizations).