Liquidity stress scenario

The assumptions are provided by the regulatory authorities based upon a combined market and idiosyncratic stress, which incorporates many of the shocks experienced during the financial crisis. They should be viewed as a minimum supervisory requirement for banks. Banks are expected to conduct their own stress tests to assess the level of liquidity they should hold beyond this minimum, and construct their own scenarios that could cause difficulties for their specific business activities. The results of these additional stress tests should be shared with supervisors (Individual Liquidity Adequacy Process and Liquidity Supervisory Review and Evaluation Process are covered in Section 19 of the handbook).

Liquid assets

Liquid assets eligible for inclusion in the calculation were initially split into two categories (Level 1 and Level 2), but this was subsequently increased to three (Level 1, Level 2A, and Level 2B) to capture a broader range of securities:

• Level 1: up to 100% of the total, comprising cash, deposits at the central bank government/government guaranteed bonds, and covered bonds rated ECAI 1 (External Credit Assessment Institutions – EU rating system) that meet certain conditions (subject to a 7% haircut and a 70% cap in the liquidity buffer);
• Level 2A: up to 40% of the total with a minimum haircut of 15%, comprising government bonds with 20% risk weight, EU covered bonds rated ECAI 2, and non‐EU covered/corporate bonds rated ECAI 2;
• Level 2B: up to 15% of the total with a haircut of 25–50%, comprising RMBS, auto, SME, and consumer loan securitisations, corporate bonds rated ECAI 3, shares that are part of a major stock index, and other high‐quality covered bonds.

Cash outflow/inflow assumptions

The rules for the EU, which were finalised in January 2015, contained a number of amendments to the assumptions on cash outflows/inflows:

• From 2019, a new lower outflow rate of 3% for retail deposits (subject to Member State and Commission approval);
• Two new risk buckets for other less stable retail deposits (10–15% and 15–20%);
• Lower outflow assumptions for maturing deposits from non‐financial customers, sovereigns, central banks, and PSEs (20–40%);
• More granularity for outflow assumptions on maturing secured funding transactions;
• Higher outflow assumptions (30–40%) for undrawn committed liquidity facilities to non‐financial corporates provided to replace funding that cannot be obtained in the financial markets;
• More favourable outflow assumptions on deposits from credit unions and undrawn credit or liquidity facilities for members of a group or institutional protection scheme.

The overall effect of the assumption changes has been a net relaxation of the rules, making it easier for banks to comply. As at 31 December 2012, Group 1 Banks (those with Tier 1 capital >€3 bn and internationally active) had an average LCR of 95% based on the 2010 BCBS rules and 119% based on the January 2013 version. As at the same date, Group 2 Banks had an average LCR of 99% based on the 2010 BCBS rules and 126% based on the January 2013 version.

LCR implementation in the US and Switzerland

The US has introduced an LCR that is consistent with that applying in Europe, although it differs in a number of respects. There is an additional feature called a “maturity mismatch add‐on” – it requires institutions to identify the largest single‐day maturity mismatch within the 30‐day period by calculating the daily difference in cumulative outflows and inflows that have set maturity dates. The maturity mismatch add‐on is the difference between the peak day amount and the net cumulative outflow amount on the last day of the 30‐day period. This is illustrated in Figure 10.2 – although the institution in question has a 100% LCR, there is an additional liquid asset holding requirement of US$230m due to the maturity mismatch add‐on.

The LCR rule only applies to the largest, internationally active US banks. The final rule has been effective since 1 January 2015 (min. 80% compliance), though there is a transitional phase with full application from 1 January 2017 (earlier than under the Basel Committee framework). The LCR rule will be complemented by the NSFR in 2019.

Due to the low level of public debt in Switzerland, “Alternative Liquidity Approaches” are available to enable compliance with minimum ratio requirements. Outflows in Swiss francs can be covered up to a defined portion in foreign currency assets. Also, a higher proportion of Category 2 assets may be permitted and short‐term repo transactions against high‐quality assets with a maturity of less than 30 days may be treated as non‐existent.

The changing liabilities structure of banks

Let's examine how bank balance sheets have evolved over time in order to better understand the context of a liabilities strategy. Figure 10.4 shows the changes in breakdown of funding types that took place with UK banks between 2007 and 2012.

The liability structure of many EU and MENA banks have changed since 2007–2008 and have focused on moving away from wholesale funds, both short term (supply) and long term (demand), and move towards more non‐interest bearing liabilities (NIBLs) and equity. This has largely been driven by the introduction of the ILAS regime and subsequent LCR stress testing regulatory metrics. Both of these stress tests penalise the use of wholesale funding and hence banks have diversified and focused on retail and NIBLs balances. For one high street UK bank, NIBLs rose from 28% of liabilities and equity in 2007 to 41% in 2012.

The liability structure: what is the optimum mix?

In compiling the optimum liability mix targets, there are some important considerations that need to be applied in the planning process. These revolve around:

• Regulation;
• Liquidity value (to what extent is a type of customer deposit or type beneficial towards the final LCR or NSFR metric?);
• Funding diversification/concentration;
• Impact to net interest income (NII) and net interest margin (NIM);
• NII sensitivity;
• How to build the customer franchise;
• Set up costs of new products.

Figure 10.5 highlights the detailed considerations that need to be applied during the planning process.

Banks fund their balance sheets using a variety of liability types. The liquidity value of each type is different. Ideally, a funding type has high liquidity value as well as high NII value, customer franchise value, etc. This is essentially the “strategic ALM” optimisation problem, attempting to structure the balance sheet mix, in this case with respect to liabilities, that best serves the competing requirements of the regulator, the customer stakeholder, and the shareholder.

In practice, there is usually a trade‐off among these factors as some liabilities have greater term liquidity value for a bank than others. At one end of the scale are retail current accounts; at the other end are short‐term unsecured wholesale liabilities, sourced in the inter‐bank market, which have much more volatile characteristics in stressed market conditions and therefore are less valuable for liquidity management purposes.

Figure 10.6 illustrates the hierarchy of value of different types of liabilities. From a liquidity value perspective, a bank should seek to maximise funding from customer relationship balances and minimise its reliance on wholesale inter‐bank funding.

The mix of these should be arrived at through design as much as possible and a bank's FTP process may be used to help drive liabilities raising of the “right” type and through incentivising businesses by rewarding them with a higher commission or sales credit for the higher value liquidity type.

Peer group analysis

In the UK, the Bank of England (BoE) publishes monthly data showing effective interest‐rate paid on (sterling) deposit balances of UK household and UK non‐FI corporate sectors, split by deposit type. It is a measure of back‐book deposit costs (in the UK), not front book (i.e. marginal cost of new funds).

Through this peer review, one can compare the current bank levels to peer levels to establish whether your bank is paying above the sector‐wide average and to establish a better understanding of what your bank is paying for its deposits and assess if this can be improved.

From Figure 10.7, we observe that for UK banks, 2‐year fixed‐rate bonds have reduced by ∼200bp since 2010, from an average of 3.25% to 1.46%. For instant access liability products, this has reduced by ∼90bp from 1.44% to 0.54%.

In addition to data published by central banks, a bank should survey the market itself to determine the current pricing associated with certain liability products. In Figure 10.8 we present the results of such a comparison. We note that “XYZ bank” is exactly where it most probably wishes to be, in the middle of the range among its peers and not an outlier for rates at either end. The highest deposit rate payers are sometimes perceived by customers to be struggling to raise deposits, and such a customer perception would not be welcome for a bank.

An analysis of front‐book deposit prices vs competitors would provide additional colour to the liabilities review. This information should also be presented regularly to either the Products Pricing Committee or Deposits Pricing Committee (subcommittee of asset–liability committee (ALCO)). As noted above, the general rule of thumb is not to be an outlier at either end.

Product volume sensitivity to interest‐rate movements

Some liability products are sensitive to movements in interest rates and are referred to as NII‐sensitive products. A high volume sensitivity may not be a good thing as this could lead to certain challenges, notably cannibalisation from other product types and “hot money”.

Cannibalisation from other deposit products offered by the bank can lead to a dramatically different impact on NII than originally anticipated, especially if large volumes of deposits switch to the new/repriced product from high margin accounts. Placing restrictions on who can apply for certain products can attract the attention of the regulators.

“Hot money” from interest‐rate seekers (stand‐alone customers): if a new/repriced account attracts deposits from customers who are only interested in the rate being paid, then you may finish up with expensive funding that has no liquidity value. For this reason, many banks have withdrawn from offering offshore instant access Internet‐based accounts.

The same can also be said of products that have been offered at below par rate. Deposit products often have tiered rates and/or customer relationship managers are given the discretion to pay a higher rate to protect/attract business. Average product margins can mask underlying trends. Figure 10.9 is an example of this type of review in more detail.

As part of the liability strategy, a bank should conduct a granular review to understand which deposits are being written at negative margins and consider the options to reduce the size of this segment of business as it is value destroying.

Output of the liability review

The result of the liabilities mix review should be a medium‐term (for example, 3‐ to 5‐year) strategy that specifies a precise target for:

• The liabilities mix: how much do we wish of each funding type, and why;
• What the drivers are;
• What, if any, funding types are of less value and should be reduced or withdrawn from;
• Plan for implementation.

This liabilities strategy review would be presented for ALCO approval and subsequent implementation. This plan could include a series of initiatives that would be tracked at ALCO on a regular basis to ensure that the balance sheet is tracking in the right direction and is being measured by some key performance indicators.

Initiatives could include any or all of those listed in Figure 10.10.

Tracking progress through ALCO

The liabilities strategic review output should articulate the high‐level deposits strategy. For example, this could be a simple statement such as “The 2016 strategy is to retain current balances and enhance the value of those retained balances for NIM” or “The medium‐term strategy is to grow the deposit book by £2 billion”.

The ALCO review should have a level of product emphasis aligned to the specific initiatives that the plan has identified such as focus on instant access savings or term deposits.

There should be a regular competitive analysis, as previously described, to ensure that pricing is still relevant in the market and that process against key performance indicators is tracked monthly and any variations clearly understood.

Summary

A balance sheet funding review is an essential start to the bank's drafting of its optimum funding mix and liabilities strategy. The larger and more complex the bank (IB, corporate, retail), the more complex the review. It is important to understand the share of the balance sheet, and costs currently, and review these in the context of what the balance sheet should look like in 1 to 3 years' time.

The overall optimum liabilities strategy will consist of a series of tactical/strategic initiatives that will help the bank re‐engineer the funding profile to achieve an optimal mix.

This is the start of a more logical and coherent approach to balance sheet funding.

Summary

The LAB is an integral part of the balance sheet and is predominantly derived from the quality of the liability base through the lens of stress tests. It needs to be representative of the risk appetite of the Board and commensurate with the underlying breakdown of the balance sheet. The LAB needs to be easily and quickly liquefiable and this needs to be tested regularly via sale or repo of a test sample of the portfolio.

The daily maximum liquidity requirement

Banks need to calculate their net intra‐day liquidity position (the difference between the total value of payments received and payments made at any point in the day) during the course of the day. The bank's largest net cumulative position during the day will determine its maximum intra‐day requirement on that day. If a bank runs a negative net cumulative position at some point during the day, it will need to access intra‐day liquidity to fund this balance. In the example in Figure 10.16, this is just over 10 currency units.

Time‐specific and other critical obligations

Banks must identify the volume and value of time‐specific obligations and any missed payments. This enables supervisors to assess whether these obligations are being effectively managed. Examples include payments required to settle positions in payment and settlement systems and those related to market activities, such as the delivery or return of money market transactions or margin payments. A bank's failure to settle such obligations on time could result in financial penalty, reputational damage, or loss of future business.

Intra‐day throughput

This metric shows the proportion, by value, of a bank's outgoing payments that settle by specific times during the day (for example, by 9.00 a.m., by 10.00 a.m., etc.) and enables supervisors to identify specific times during the day when a bank may be more vulnerable to liquidity or operational stresses and any changes in its payment and settlement behaviour.

Intra‐day liquidity stress testing

The monitoring tools discussed provide supervisors with information on a bank's intra‐day liquidity profile in normal conditions. However, the availability and usage of intra‐day liquidity can change markedly in times of stress. As guidance, the Basel Committee has developed four stress scenarios relating to own financial stress: customer stress, counterparty stress, and market‐wide credit or liquidity stress. Banks should use the scenarios to assess how their intra‐day liquidity profile in normal conditions would change in a stressful situation and discuss with their supervisors how any adverse impact would be addressed either through contingency planning and/or the wider intra‐day liquidity risk management framework.

How are the intra‐day liquidity rules influencing banks?

The intra‐day liquidity rules have encouraged banks to focus on IT development to generate the required metrics and real‐time information flows. Senior management understanding of sources and uses of intra‐day liquidity has improved rapidly and banks are employing their available liquidity resources more efficiently, for example, through payment scheduling to reduce maximum intra‐day liquidity usage and ensuring that liquid assets are held in the right place to support payment systems as necessary. Also, banks have considered what might happen in a stress situation and put plans in place to mitigate the outcomes.

Exposures and CVA

A vanilla interest‐rate swap (IRS) exhibits the following characteristics:

• No exchange of principal at inception;
• Parties exchange fixed for floating (i.e. a variable rate such as Libor) throughout the life;
• No re‐exchange or principal at termination.

A cross‐currency swap (XCY) has the following characteristics:

• Full exchange of principal at inception, one currency vs another;
• Exchange of floating payments, one currency for the other throughout the life;
• Full re‐exchange or principal at termination.

The counterparty risks on the IRS are as follows:

• Traded “at market” at inception, hence no net counterparty risk on day 1;
• Unless yield curves are (exactly) flat, IRS has implied payments throughout the life: as time passes payments are made;
• Net payable/receivable remain on structure;
• Interest rates fluctuate throughout the life:
  • Expected payments on floating leg vary;
  • Discounting of implied cash flows varies.

The counterparty risks on the XCY are:

• Traded “at market” at inception, hence no net counterparty risk on day 1;
• Unless yield curves are (exactly) flat, XCY has implied payments throughout its life:
  • As time passes, payments are made;
  • Net payable/receivable remains on structure.
• Interest rates fluctuate throughout the life:
  • Expected payments on floating leg vary;
  • Discounting of implied cash flows varies.
• FX rates fluctuate throughout the life:
  • Value of the re‐exchange of principal varies.

The fact that market rates vary from day 1 onwards throughout the life of the contract presents the problem with estimating counterparty risk exposure. “At market” derivatives start with a zero net present value; however, over time and the movement of markets, the structure will develop a present value (PV); but we don't know how much, because we don't know how market prices will evolve. Therefore, the amount of our counterparty exposure is uncertain. The questions to answer are (1) How do we know how much to price for counterparty risk at inception? And (2) How can we hedge this counterparty risk?

We don't know exactly how (i.e. in which direction) market prices will evolve but we do know (can hypothesise) their dynamics and distribution. For example, under the Black‐Scholes model assumption we assume a log‐normal interest rate and foreign exchange rate movements profile. Hence, the most common approach is to simulate the variation in PV. For instance, how will interest rates develop? Figure 10.19 shows a set of interest‐rate simulations, where we assume a yield curve flat at 4%, and a set of possible profiles on how the remaining maturity swap rate may develop over the next 5 years. Solid lines are one standard deviation move.

The future distribution of possible market rates and FX rates expands at a decreasing rate (the √T law), converging to a log‐normal distribution. But what about the remaining mark‐to‐market (MtM) value on the swap, and hence the counterparty risk? Over time, fewer cash flows remain on the IRS as duration shortens, and on the XCY the exposure is dominated by the principal occurring in one bullet payment at maturity. So again we run the simulation; expected exposure on an IRS peaks at between 18 and 30 months, then falls back towards zero at maturity. Figure 10.20 shows an example of remaining maturity IRS MtM simulations.

However, with an XCY the expected exposure grows throughout the life, peaking at maturity (due to the bullet exchange of principal on maturity). We show this at Figure 10.21.

To reiterate, XCY swaps involve principal exchange, whereas interest rate swaps involve only coupon exchanges; therefore, expected exposures are much larger on the XCY than on the IRS, as we compare in Figure 10.22.

So how do we use these PV MtM distributions to calculate counterparty risk, and hence the counterparty credit valuation adjustment (CVA)? The most common approach involves:

1. Price the cost of insuring expected exposure on day 1;
2. Only increase the hedge if realised exposure exceeds expected exposure.

Using this approach, CVA is therefore the cost of insuring the area under the curve shown at Figure 10.23, given by:

where

• LGD_C is the counterparty loss given default;
• Q_C(t_i–1,t_i) is the counterparty default probability.

LGD and Q can be estimated from market credit curves (CDS) or estimated from fundamental credit analysis. So we see that CVA is the cost of hedging expected counterparty credit exposure.

We should be aware that there are some obvious problems with the standard CVA approach:

• Where one party has a positive mark, the other has a negative mark:
  • Both should be charging each other CVA;
  • No agreed, open market price, therefore no trade;
  • In reality, there is usually a significant disparity between (bank) dealer credit and customer credit, which partly explains the need for collateralisation between inter‐dealer/trading counterparties;
  • Dealers are typically unable to monetise a negative market (their own payable).
• Exposure is not capped at the expected exposure:
  • This is only an arbitrarily chosen percentile exposure (normally one standard deviation);
  • Exposures (and hence potential losses) can be much higher.
• Credit curves may be unobservable and/or untradable.

As one can appreciate, there is more than way to approach this issue!

While CVA relates to the credit exposures from derivatives, FCA relates to funding costs. An uncollateralised derivative hedged with a collateralised one can give rise to a funding requirement: if markets move such that the collateralised trade has a negative MtM, then collateral must be posted. There would be no offsetting flows on the uncollateralised derivative, thus giving rise to a net funding requirement. At the same time, the CVA will increase due to the net counterparty exposure. So we see that CVA and FCA operate in the same direction. We consider FVA next.

First principles

Inter‐bank derivatives trading takes place under the CSA arrangement in the standard ISDA agreement. This means that the mark‐to‐market value of each derivative contract is passed over as collateral, usually in the form of cash but sometimes as risk‐free sovereign securities. In general, the collateral requirements under a two‐way CSA agreement should result in a netted zero cash flow position, because what a bank needs to pass over as collateral on a derivative that is offside, it will receive from the counterparty to the hedge on this derivative. However, a number of counterparties, such as corporates, sovereign authorities, sovereign debt management offices, and central banks, do not sign CSA agreements. This one‐way CSA arrangement will create a funding requirement for a bank, as it will have to transfer cash if it is MtM negative, while it will not receive any cash if it is MtM positive.

To incorporate the correct discipline with regard to the liquidity effects generated by uncollateralised derivatives business, the funding policy needs to incorporate an appropriate liquidity premium charge. This will apply to the net MtM value of all uncollateralised derivatives on the balance sheet. By charging the right rate, the business lines are incentivised to work towards reducing uncollateralised business wherever possible.

A bank will fund its balance sheet at its specific COF, which splits into four categories:

• Secured short‐term funding costs: the rate at which the bank borrows against collateral. This is generally the OIS curve, and thus the lowest funding rate available (OIS lies below Libor). It is not relevant in an uncollateralised derivative context, because such instruments cannot be used as collateral (even if they are positive MtM);
• Secured long‐term funding costs: the rate at which the bank can borrow by issuing term secured liabilities such as covered bonds and mortgage‐backed securities;
• Short‐term unsecured funding costs: the bank's COF for short‐dated (0–12‐month) tenors. At its lowest this will be around Libor, although many banks' short‐term unsecured borrowing rate is at a spread above Libor. That said, for certain banks the short‐term unsecured funding will be at zero, for example, retail deposits such as current (“checking”) account balances;
• Long‐term unsecured funding costs: the bank's COF for long‐dated (2–10‐year) tenors, also referred to as the term liquidity premium (TLP).

A general position of OIS, Libor, and TLP curves is shown in Figure 10.24.

The above of course is in the “wholesale” or investment bank space. In reality, derivatives dealers are often part of “universal” banking groups that also include retail and corporate business lines. A large part of the balance sheet will be funded therefore by low‐cost liabilities of contractual short‐dated but behaviourally long‐dated maturity. These also need to be factored into the pricing curve in a way that is appropriate. One approach might be to derive the derivatives COF off a “weighted average cost of funds” (WACF) curve that is an average of all balance sheet liabilities. Either way, care needs to be taken that business lines in the wholesale bank, which includes the derivatives desk, are charged an appropriate price and not necessarily the retail or corporate COF rate, particularly since customer deposits have short contractual tenors but long behavioural tenors, and so do not inflict a term liquidity premium (TLP) on the bank.

Derivative liabilities and assets

Derivative liabilities correspond to what is termed an overall expected negative exposure (ENE), the most basic example of which is a deposit. A derivative asset corresponds to an overall expected positive exposure or EPE and at its simplest would be a loan.

An appreciation of the terms of a derivatives funding policy requires an understanding of credit valuation adjustment (CVA), debt value adjustment (DVA), and funding value adjustment (FVA). Following existing literature (for example, see Picault (2005) and Gregory (2009)), under a set of simplifying assumptions we have:

where EPE is expected positive earnings, and PD and LGD are standard credit analyst expressions for default probability and loss given default. More formally we write:

where R is the recovery rate, q is the probability density function of counterparty default, and v is the value of the derivative payoff. In discrete time we write:

where q_i is the probability of default between times t_i‐1 and t_i:

and

In other words, the discounting to be applied for valuation is at the appropriate tenor bank funding cost.

The funding cost to apply to the derivatives portfolio cash flows may sometimes be selected depending on what assumption we make about the ease of unwinding the portfolio:

• Assume no easy unwind: if we cannot unwind the portfolio without punitive costs, we must assume we will have to fund the transaction for the full term. The funding cost of this commitment is given by the bank's long‐term COF. If we fund (value) at short‐term COF we run the risk that sudden spikes in the short‐term COF will create funding losses, or that a liquidity squeeze in general will impact our ability to roll over funding for the position. To avoid this risk, we would fund with long‐term borrowing, and discount unsecured derivatives off the long‐term COF (TLP) curve;
• Assume easy unwind: if we can unwind the position with no extraneous cost, we can apply the short‐term COF, say the 1‐year TLP. The assumption of easy unwind means that we are not committed to rolling over funding; in the event of liquidity stress we would simply unwind the portfolio and eliminate the funding commitment. This is a strong assumption to make, particularly at a time of stress, and would be a high‐risk policy.

Therefore, in theory we recommend that the derivative asset be discounted at TLP and the funding for collateral postings be substantially term funded. That said, in some cases the funding generated from a derivative book (assuming no counterparty default) is contractually for a long maturity, and so the case may be made that this should be charged for or receive the secured funding rate as opposed to the unsecured COF rate. That being the case, the derivatives funding curve then sits below the bank's COF curve and closer to the secured funding curve. This is shown at Figure 10.24.

In general, when applying derivatives funding policy we assume no netting arrangements are in place, but in practice these are quite common and will have an impact on the bank's collateral funding position in the event of default.

Derivative portfolio maturity

The tenor period to apply when applying the correct funding cost to derivative book cash flows can be the contractual maturity of the derivative in question, but not necessarily so. An alternative approach is to split the portfolio into tenor buckets commensurate with the tenors at which we wish to fund the cash flows, with each bucket funded at the appropriate tenor COF.

Placing the derivative portfolio cash flows into appropriate term tenor buckets is a logical position on which to base how we choose to fund these cash flows. In practice, the derivative valuation model itself can be used to produce this tenor bucket breakdown, in the form of a “funding risk per basis point” (FR01) delta ladder. Using this model output removes the need for a subjective analysis of the maturity profile of the portfolio. In other words, the maturity profile of the portfolio is given by the model output. The appropriate tenor TLP is charged on the amount in each bucket.

This is shown at Figure 10.25.

In practice of course the profile is unlikely to look like the one in Figure 10.25 (although it might do). Rather, it is more likely to be all one way – either net long or short across most if not all tenor buckets.

In summary, cash flows arising out of the derivatives business, both contractual and collateral, must be funded at the appropriate TLP COF for their tenor. This means term funding a large part of the portfolio cash flows.