Introduction

• Until the global financial crisis (GFC), many banks treated liquidity as a free good. In the years preceding the GFC, liquidity was plentiful and cheap which was unsustainable in the long run. This encouraged leverage and maturity transformation, leading to record profits of banks. At the same time these conditions led many to believe that funding would always be available, and at permanently cheap rates. One consequence of this belief was that it provided little incentive for banks to devote attention to liquidity risk management. As a result, many banks failed to recognize the true nature of the liquidity risk embedded in their business activities.

• LTP is a process that attributes the costs, benefits and risks of liquidity to respective business units within a bank. The purpose of LTP is to transfer liquidity costs and benefits from business units to a centrally managed pool.
  • To achieve this, LTP charges
    • users of funds (assets/loans) for the cost of liquidity, and
    • credits providers of funds (liabilities/deposits) for the benefit of liquidity.
  • LTP also regains the cost of carrying a liquidity cushion by charging contingent commitments, such as lines of credit, based on their predicted (expected) use of liquidity.

This is depicted in the figure in the next page.

A Graphical Representation of the LTP process

LTP Process – Shortcomings

• Banks with poor LTP practices typically underprice or (even worse) fail to price liquidity. Such banks are more likely to accrue illiquid assets and contingent exposures, and under-value stable sources of funding. This outcome applied to many banks and other financial institutions prior to the GFC. LTP has gained considerable attention since the GFC with some reports linking poor LTP practices to the funding and liquidity issues witnessed at several banks. In 2009, a group of prudential regulators conducted an international survey to assess the progress banks are making to enhance LTP. The survey covered 38 banks from nine countries. The survey responses revealed that many of the LTP practices employed by banks were short of good practice.
  • Most banks included in the survey lacked an LTP policy. As such, LTP was not defined nor were there any rules or principles in regard to how LTP should operate.
  • Most banks which were operating with decentralized funding centers had inconsistent LTP regimes. These banks relied on manual off-line processes to intervene.
  • For many banks, internal treasuries lacked visibility over individual business balance sheets, limiting their understanding of individual funding requirements and liquidity exposures.
  • Oversight of the LTP process at nearly all banks that participated in the survey was poor to nonexistent, especially by risk and financial control functions.
  • Liquidity Management Information Systems (LMIS) of most banks were simplistic and inflexible. Many of the systems were unable to attribute the costs, benefits, and risks of liquidity appropriately to respective businesses, and at a sufficiently granular level. This resulted in product mispricing, which distorted profit and performance assessments. Profit pools were derived from a simple percentage of accrued revenues without any regard for the liquidity risk taken to generate such profits. This encouraged revenue and risk maximization rather than risk-adjusted earnings.
  • Some banks treated liquidity as a “free” good, completely ignoring the costs, benefits and risks of liquidity. These banks neglected to charge or credit respective businesses, products and/or transactions accordingly.
  • Most banks employed a pooled average cost of funds approach to derive the costs and benefits of liquidity. This resulted in short- and long-term assets receiving` the same charge for the cost of liquidity and, conversely, short and long-term liabilities receiving the same credit for the benefit of liquidity.
  • Liquidity cushions were derived from stress assumptions revolving around a single bank’s sudden inability to raise funds. Without consideration of systemic funding scenarios, most cushions were too small to withstand prolonged market disruptions. Also, cushions comprised liquid assets that were themselves funded short-term.

LTP Process – Best Practices

• The best practices for governing the LTP process include the following –
  • Banks should have a well defined LTP policy with proper guidelines to ensure business units understand the reasoning behind charges relating to the use of liquidity.
  • All wholesale funding should be centrally managed by a group treasury, which should have complete knowledge of all liquidity exposures. This restricts arbitrage between business units and treasury, and between business units themselves.
  • The policies of the trading book should be well defined and understood. Banks should develop risk controls and limits for trading activities to properly measure, monitor and assess the liquidity risk embedded in products and business units.
  • The LTP process should be supervised by the senior management as well as independent risk Management at all levels, treasury functions, as well as independent risk and financial control personnel should become more engaged in the LTP process. In addition, monthly meetings should be held to monitor costs and benefits.
  • LMIS should be able to attribute the costs, benefits, and risks of liquidity appropriately to respective businesses, and at a sufficiently granular level.
  • Business units should consider the cost of liquidity as part of their decision to book certain assets. Haircuts on traded assets should be widened to account for more severe and prolonged market disruptions, and to ensure that assumptions surrounding the amount of liquidity that can be generated during a crisis are appropriately conservative.
  • Higher funding charges should be applied to trading positions that are more likely to become “stale” (i.e. positions that have a higher probability of rolling from the trading book to the banking book).
  • For relatively small trading book exposures, over-trading behavior should be curbed by imposing higher funding charges on net funding requirements when certain funding limits are breached. For larger trading book exposures, more attention should be given to understand the funding requirements of individual trading desks, and charges should be applied on a more granular basis.

Governing LTP – Challenges

• Broadly speaking, all policies, processes and practices require governing. This is normally achieved through a combination of external control factors, such as regulation and competition, and internal control factors, such as board oversight and risk management. Because external control factors affect institutions in much the same way, governance is differentiated largely by the internal control factors that are employed.

• Institutions with weak internal controls are more prone to the problems of moral hazard and adverse selection, leading to poor performance. Many of the poor LTP practices across banks that participated in the survey were the direct result of weak internal controls.

• Some of the challenges in the implementation of LTP are as follows –
  • 1) LTP Policies
  • 2) Internal Funding Structure – Centralized vs Decentralized
  • 3) Trading Book Funding Policies and Identifying Funding Requirements
  • 4) Oversight
  • 5) LMIS (Liquidity Management Information Systems)
  • 6) Remuneration

Challenge 1 – LTP Policies

• For most banks in the survey, the internal pricing of liquidity risk is a relatively new concept, brought to light by the recent breakdown in wholesale funding markets, and the consequent increase in funding costs. It will take time for banks to establish adequate LTP policies and procedures, but this is a necessary first step towards better LTP practice.

• Banks have traditionally relied on internal transfer pricing to manage interest rate risk in the banking book, and to assess and monitor the performance of products and business units, but with no or only minimal adjustments for liquidity costs, benefits and risks.

Challenge 2 – Internal Funding Structures

• There is substantial debate surrounding the optimal internal structure of banks – is it better to have a centralized funding center, whereby wholesale funding is restricted to a group or subsidiary treasury or, alternatively, decentralized funding centers, whereby certain business units are able to raise funding themselves from their own sources to cover their own liquidity needs?

• There are reasonable academic and economic arguments that provide support for both approaches. However, the survey identified that banks with decentralized funding centers, particularly those with large prime brokerage business activities, were more susceptible to poor LTP practices. For example, some business units that were able to raise wholesale funds from external sources then sold the funds to treasury and in some cases to other business units, at a higher rate. This resulted in a “risk-free” profit to the business unit at the cost of more and possibly badly managed risk for the bank as a whole.

• In addition to this, most of the banks with decentralized funding structures employed inconsistent LTP regimes and relied on manual off-line processes to update funding costs.

Challenge 3 – Trading Book Funding Policies

• Probably the worst LTP practices identified in the survey were in relation to trading and investment banking activities. A combination of poorly designed trading book policies, inadequate risk controls and limits, as well as a lack of oversight were to blame. For example, some banks that took part in the survey lacked trading book funding policies and procedures, which allowed for over-aggressive trading behavior and the accumulation of illiquid assets in search of revenues, not risk-adjusted profit.

• Most of the banks included in the survey did have trading book funding policies, but nearly all of these policies assumed that assets were only held short-term (i.e. for 180 days or less). One problem with this approach is that, irrespective of whether assets are likely to be held for more than the 180-day threshold, long-term funding charges only apply when assets roll from the trading book to the banking book. This provided little incentive for banks to develop risk controls and limits to adequately measure, monitor and assess the liquidity risk in traded assets, and was evident through the build-up of positions that were highly illiquid.

• Many of the larger banks included in the survey, particularly those with substantial trading businesses, lacked a line of sight to individual business balance sheets, and thus could not identify the funding requirements of individual trading desks. As a result, trading and investment banking activities were funded based on the total net funding requirement across all related business units. This method essentially provides a line of credit to the trading book, and gives no regard to the liquidity risk embedded in business activities. This approach is therefore considered to be poor practice.
• On a separate but related issue, banks with large trading businesses that participated in the survey also applied insufficient haircuts to many of the traded assets they held. These banks clearly underestimated the likelihood of a market disruption, and the extent to which market liquidity could evaporate. The severe drop in market prices led to calls on margin positions and placed severe pressure on banks’ abilities to meet funding requirements. Part of the reason this occurred was because no one had previously thought of the need to price the liquidity costs of potential margin calls.

Challenge 4 – Oversight

• Ineffective oversight of the LTP process contributed to many of the problems that were identified at banks that took part in the survey. A major challenge is to decide the level of optimal oversight.

• The accrual of long-term illiquid assets and short-term volatile liabilities created a large and poorly understood mismatch between the maturities of assets and liabilities, and therefore exposed banks to greater structural liquidity risk.

• Probably the most striking example highlighting the implications of poor oversight was how some of the banks’ LTP processes enabled them to accumulate significant amounts of highly rated, yet highly illiquid, tranches of collateralized debt obligations (CDOs) in their respective trading accounts. These portfolios were assumed to be safely funded with much shorter-term liabilities, typically in the order of overnight to 90-day funds.

Challenge 5 – LMIS

• LMIS are widely used by management as a primary source of measuring and monitoring the performance of businesses. LMIS play a pivotal role in helping management achieve group-wide goals. Weak LMIS could easily distort the information for decision-making and prevent the bank from achieving its objectives.

• If LMIS employed by banks are too basic, it will limit the effectiveness and efficiency of the LTP process. In some cases, for example, the basic and rigid nature of LMIS means that certain business activities can fail to receive a charge for the cost of liquidity or, conversely, a credit for the benefit of liquidity.

• Another weakness in many of the LMIS can be that they prevent the costs, benefits and risks of liquidity from being attributed at a sufficiently granular level.

• One implication of the weakness in LMIS and the poor LTP practices that is that businesses might report performance (and employees might claim bonuses) on a basis that may not reflect their actual performance. Essentially, this would limit management’s ability to monitor performance.
• Upgrading LMIS in a large bank is a costly and long-term process. But the benefits of appropriately charging business activities for the cost, benefits and risk of liquidity, and at a sufficiently granular level, will far outweigh the costs and limitations of the basic LMIS that were previously employed. LMIS that are sufficiently advanced to achieve these outcomes will promote better LTP practice.

Challenge 6 – Remuneration

• If designed well, incentive pay can have enormous benefits. It encourages behavior that is consistent with the culture of an institution, and assists management in achieving group-wide objectives. On the other hand, poorly designed remuneration can promote perverse behaviors such as excessive risk-taking, which could severely impact the performance of an institution.

• In 2009, the Financial Stability Board (FSB) reported that poor remuneration practices were one of the factors that contributed to the GFC. High short-term profits led to generous bonus payments to employees without adequate regard to the longer term risks they imposed on their firms. These perverse incentives amplified the excessive risk-taking that severely threatened the global financial system and left firms with fewer resources to absorb losses. Remuneration was largely insensitive to the risks taken to generate income, and to costs associated with long-term funding commitments that were required to hold illiquid assets Profit measures used as a basis for determining remuneration were often distorted. Profit pools, for example, which are generally used to determine short-term incentives, or bonuses for employees, were derived from a simple percentage of accrued revenues, without any regard to the cost of liquidity. This placed more emphasis on maximizing revenues rather than risk-adjusted earnings. Remuneration was not structured properly, particularly for those employees responsible for oversight. For many staff in these areas, remuneration was designed such that it largely depended on the performance of front-line businesses they were responsible for overseeing. Thus, including the actual costs for liquidity would have impacted negatively upon business unit performance, which inevitably would have reduced personal remuneration and benefits for employees. Clearly, this would have also impacted the independence of their role.
• Banks should develop their respective LTP processes to ensure that profit and performance measures include the relevant costs for liquidity (and capital, although this is a separate issue).

Zero Cost of Funds Approach – Liquidity as “Free” Good

• Probably the most striking example of poor practice identified in the survey was that some banks failed to account for the costs, benefits and risks of liquidity in all or some aspects of their business activities. These banks came to view funding liquidity as essentially free, and funding liquidity risk as essentially zero. As a result, there was simply no charge attributed to some assets for the cost of using funding liquidity, and conversely no credit attributed to some liabilities for the benefit of providing funding liquidity. This was   undoubtedly   the   worst   practice identified in the survey.
• This figure provides a graphical representation of what this would look like in practice. The rate charged to users of funds in this instance would have been derived from the swap curve only. If it is assumed that interest rate risk is properly accounted for using the swap curve, then a zero spread above the swap curve implies a zero charge for the cost of funding liquidity.

• A zero charge for the cost of liquidity and, conversely, a zero credit for the benefit of liquidity exacerbated maturity transformation to the largest degree possible. This approach resulted in the hoarding of long-term highly illiquid assets, and very few long-term stable liabilities to meet funding demands as they became due.

• Ideal funding conditions in the years preceding the crisis could provide one explanation of why some banks viewed liquidity as a free good, and funding liquidity risk as essentially zero. This figure shows how the spread between one-year LIBOR and the one-year swap rate changed during the period June 2005 to October 2010. In June 2005, at the peak of robust share market growth, the spread was only 0.5 basis points (bps). With funding conditions so easy, it is likely that banks viewed spreads as pure credit risk adjustments and neglected (ignored) funding liquidity risk altogether. If banks believed funding would always be available and at permanently cheap rates, this simply could have masked the need to charge assets for the cost of liquidity, and conversely, credit liabilities for the benefit of liquidity.

Pooled Average Cost of Funds Approach

• Some banks recognized the need to charge users and credit providers of funding liquidity and employed a pooled approach to LTP, where an average rate was calculated based on the interest expense (cost of funds) across all existing funding sources. For example, if deposits were a bank’s only source of funding the average rate would be based on the total interest expense for all deposits divided by average total deposits, adjusted for floats and reserve requirements. There is only one “average” rate calculated, and all assets irrespective of their maturity are charged the same rate for their use of funds (cost of liquidity), as depicted in this figure.

• To illustrate how charges and credits for the use and benefit of funds would be allocated under an average cost of funds approach, consider the following example. If the average rate across all funding sources was 10 bps, all loans would receive a charge of $1,000 on a principal amount of $ 1 million, irrespective of their maturity. Assuming this rate was also used to reward fund
• providers, then all deposits would receive a credit of $1,000 on a principal amount of $ 1 million, irrespective of their maturity. This can be seen in this table.

• Problems with this approach –
  • 1) It ignores the heightened liquidity risk embedded in longer-term assets. Charging one “average” rate for the use of funds inherently assumes that all assets, irrespective of their maturity, pose the same liquidity risk.
  • 2) If this “average” rate is also used to credit fund providers, then an incentive to write loans will be met with a direct disincentive to gather deposits. For example, decreasing the rate charged to fund users from 10 bps to five bps will encourage loan generation, but at the same time, this will provide less incentive for business units to raise deposits. Having separate “average” rates for the costs and benefits of funds is a better approach, as depicted in this figure. To illustrate the effect of having separate average rates

for the cost and benefit of funds, consider the following example. If the    average cost of funds is 10 bps, as in the example presented above, all loans would be charged $1,000 on a principal of $1 million, irrespective of their maturity.

Further, if the average benefit of funds is four bps, all deposits would be credited $400 on a principal of $1 million, irrespective of their maturity. Under this approach, lowering the average cost of funds from 10 bps to five bps will encourage loan generation. However, because of the separate rate

for the average benefit of funds, this change will not directly discourage business units from raising deposits. This information is presented in this table.

• 3) Having one “average” rate for fund providers ignores the increased benefits of liquidity in longer-term liabilities.

• 4) Using an average cost of funds reflects historical rates and prices, but does not appropriately reflect the actual market cost of funds. If five-year funding was to increase by 20 bps, for example, the respective change in the average (cost of funds) rate would be much less. Changes in the actual market cost of funds would need to be sustained for a period of time for the effect to be fully integrated into the average cost of funds. Because the average cost of funds lags changes in the actual market cost of funds, it does not appropriately reflect market perceptions of risk for new business entering a bank’s books.
• Implications of this approach –
  • 1) Promotes Maturity Transformation – One implication of employing a pooled “average” cost of funds approach to LTP is that it promotes unhealthy as well as healthy maturity transformation.
    • Business units will be unduly encouraged to write long term assets because they do not receive higher charges for their use of funds over a longer period.
    • Conversely, business units will be discouraged from raising long-term liabilities because there is no premium credited to liabilities that provide funding for longer periods of time.

The net effect of this is a larger mismatch between the maturities of assets and liabilities on banks’ balance sheets, which inherently exposes them to greater structural liquidity risk.

This point is supported by the SSG (2009), which claims that “borrowers had taken advantage of the opportunity the market afforded to obtain short-term (often overnight) financing for assets that should more appropriately have been funded with long-term, stable funding”.

Moreover, some institutions ignored did not appropriately match funding originated transactions in their funds transfer pricing (FTP) systems on a cash-flow basis. This led to the cross-subsidization of longer-dated liquidity risk at the expense of shorter-dated risk. Such subsidization skewed business incentives and behaviors to the detriment of bank soundness.

Other factors, such as remuneration and information asymmetries, naturally encourage long-term asset generation, but under an average cost of funds approach the incentive is exacerbated.

For example, if remuneration is based on performance, which is measured via net interest income, businesses will ordinarily be encouraged to write long-term loans as they generate more interest income, with less effort, over several years. Where an average cost of funds approach is employed, this incentive becomes even more attractive for business units because long-term assets are not charged more for the cost of liquidity.

In regard to information asymmetries, business unit managers are likely to know more about their businesses’ activities than treasury. Hence, if business unit managers believe treasury is under-charging for the use of long-term funds, it will naturally encourage them to write long-term assets. But since all funds are charged the same rate for the use of funds under an average cost of funds approach, where information symmetries exist, this incentive will be magnified. A similar but opposite effect will exist for liabilities.

• 2) Distorts Profit Assessment – Another implication of the pooled average cost of funds approach to LTP is that it distorts profit assessment. The average cost of funds lags changes in banks’ actual market cost of funds, especially in volatile markets. Banks employing this approach found that their pricing methodologies resulted in the mispricing of and accumulation of assets on significantly distorted risk-adjusted terms. This made it difficult to identify poor performing products and business units on a risk- adjusted basis.

• Even though there are issues in this approach, but still, there are several reasons why some of the banks included in the survey might have chosen to adapt a pooled average cost approach to LTP.
  • First, averaging funding costs across all assets is much simpler than having to charge individual assets, products or transactions based on their contractual or behavioral (expected) maturities.
  • Second, the simplicity of the average cost of funds approach makes it easier for business units to understand the LTP process and therefore provides more incentive for them to comply.
  • Third, under this approach, the LTP process could be managed efficiently using basic LMIS.
  • Fourth, the average cost of funds is less susceptible to intermediate changes in banks’ actual market cost of funding, thereby reducing net interest income volatility across businesses. This is advantageous because it provides central management with more control over group-wide objectives.

Matched Maturity Marginal Cost of Funds Approach

• A matched-maturity marginal (MMM) cost of funds approach to LTP is the current best practice for assets and liabilities on the balance sheet. This approach calculates the portion of the cost that is attributable to liquidity. It seeks to achieve this by converting fixed-rate borrowing costs to floating-rate borrowing costs through an internal swap transaction and observing the spread over the reference rate, which is depicted from the swap curve. This spread is usually referred to as a term liquidity premium and is the rate that charges assets for the use of funds, and credits liabilities for the benefit of funds. This is presented graphically in this figure.The marginal cost of funds curve is lower than the average cost of funding curve for shorter-term assets, which indicates a smaller term liquidity premium based on current market rates. The marginal cost of funds curve is greater than the average cost of funds curve for long-term assets, indicating the greater liquidity premium for longer term assets.

• Banks incur fixed-rate costs when issuing unsecured wholesale term debt. Using these costs alone, it is difficult to strip out the portion attributable to liquidity. But swapping fixed rate costs to floating rates provides a solution. This generally involves stripping structured debt issuances into embedded derivatives and floating rate cash instruments, which are pegged to a reference rate. The spread above the reference rate is the rate that values the internal swap transaction at par. This is the term liquidity premium. It reflects both idiosyncratic credit risks and market access premiums and it better measures the cost of liquidity than an average cost of funds.

• Just like pooled average cost approach, reference rates in the marginal cost of funds approach are generally depicted from a swap curve, which is constructed from a combination of LIBOR or Euribor rates for funding up to one year. Swap curve reflects a term structure of interbank lending rates. Using the swap curve may not capture the credit risk completely since principal amounts are not exchanged between respective parties in a swap agreement. Still, swap curves are considered to provide better estimates of “base” reference rates to determine liquidity premiums than, say, government curves. This is because swap curves more closely reflect the risks to which banks are exposed when borrowing and lending money in the interbank market Swap curves also capture idiosyncratic risk and changes in general market conditions.

Examples of LTP in Practice

Non-Amortizing Bullet Loans

• Non-amortizing bullet loans provide no repayments (cash flows) throughout the life of the loan. Since all principal and interest is repaid at maturity, a funding commitment is required for the entire life (term) of the loan.

Assume the term liquidity premiums based on MMM and average cost of funds as given in this table, recorded by a bank at a point in time prior to the crisis (pre-GFC in Panel A), and more recently (current in Panel B).

• Non-amortizing bullet loans with a term of one- year would have received a higher charge for the use of funding if banks applied an average cost of funds approach rather than MMM cost of funds. However, for all other maturities, this is not true. Using an average cost of funds approach in the pre-crisis period, a five-year loan would have been undercharged eight bps ($800 on a loan of $1 million). If the same loan was originated more recently it would have been undercharged 32 bps ($3,200 per $1 million). This example highlights one of the major weaknesses of the average cost of funds method, viz., its inability to immediately reflect changes in the actual market cost of funds. For banks in the survey employing this approach, it would have encouraged business units to write long-term loans at the expense of short-term deposits.

Amortizing Bullet Loans

• Amortizing loans do provide repayments (cash flows) throughout the life of the loan. A funding commitment is not normally required for the entire life (term) of the loan and, at some point, between origination and maturity, the loan becomes self-funding.

• Consider the simplest loan type in this category, a five-year linearly amortising bullet loan with a principal amount of $1 million. This can be thought of as five separate annual loans, each of $200,000 , using MMM cost of funds approach, this loan (assuming it was originated pre-crisis) should have received a charge of:

\(\frac{1(1) + 2(2) + 3(3) + 4(6) + 5(10)}{1 + 2 + 3 + 4 + 5} = 5.9 \, bps\)

This is a tenor-weighted (blended) term liquidity premium, derived from what is commonly referred to as the tranching approach. Following this same approach, if the loan was originated recently it should have received a charge of:

\(\frac{1(1) + 2(2) + 3(3) + 4(6) + 5(10)}{1 + 2 + 3 + 4 + 5} = 5.9 \, bps\)

In both cases, the charge for the use of funds indicates that a funding commitment is required for somewhere between three and four years and not the entire term of the loan, which was five years.

• If an average cost of funds approach had been employed, the loan originated pre-crisis would have received a charge of two bps (Panel A). This would have resulted in an undercharge of 3.9 bps (5.9 — 2). If the loan had been originated more recently, it would have received a charge of eight bps, which would have resulted in an undercharge of 18.1 bps (26.1 — 8). Although the differences in the funding charges are not as severe as in the non-amortizing bullet loan example above, it still highlights the weakness of the average cost of funds approach in reflecting changes in the actual market cost of funding. Once again, this would have encouraged long-term loan (asset) generation.

• Not all amortizing loans provide known cash flows for the entire life of the loan. Take standard variable- (adjustable-) rate mortgages, for example. Often their contractual maturity will be 25 or 30 years at origination, but their actual maturity will vary depending on factors such as repayment frequency and repayment amount. In this case, a better approach is to bundle mortgages into vintages, based on their origination date, and model the repayment history (decay) over time. If mortgages tend to behave similarly, irrespective of the vintage to which they belong, then a single charge for funding liquidity can be attributed to the entire portfolio, instead of to each individual transaction. This charge is based on the behavioural maturity of the portfolio, which is often calculated by banks using the weighted average life (𝑊𝐴𝐿) method.

\(WAL = \sum_{i=1}^{n} \frac{p_i}{P} t_i\)

where

𝑃_i = principal amount in distribution 𝑖,

𝑃 = amount of loan, and

𝑡_i = time (in years) of payment 𝑖

WAL can be interpreted as the weighted-average time it takes to recoup $1 of principal (i.e. the time it takes for the loan to start paying for itself) .

• As an example, suppose a large bank writes around $2 billion of mortgage loans on average, per month, and upon examining the decay of its loans finds the behavioral maturity (WAL) of the mortgage portfolio to be approximately four years. If MMM cost of funds approach is employed, then all mortgage loans should receive a charge, at point of origination, based on the four-year term liquidity premium. Based on the figures from the previous table, this would be six bps (Panel A) or 28 bps (Panel B) depending on when the loan was originated.

Across the entire portfolio, this would translate into dollar charges of $1.2 million or $5.6 million, respectively. In contrast, if an average cost of funds approach is employed, mortgage loans should receive a charge of two bps (Panel A) if originated pre-crisis, and eight bps (Panel B) if originated more recently. Collectively, for all mortgage loans, this would translate into dollar charges of $400,000 or $1.6 million, respectively.

• This example further demonstrates how the average cost of funds lags changes in banks actual market cost of funds and, at the same time, highlights how costly this could be, especially when products are priced at the portfolio level and comprise a large portion of bank assets.

Deposits

• Deposits should be categorized as “sticky” or “hot/volatile” and credited based on their likelihood of withdrawal. Generally, sticky money, such as term deposits, are less likely to be withdrawn and should therefore receive larger credits than hot/volatile money, such as demand deposits, savings and transaction accounts, which are more likely to be withdrawn at any time.

• For example, using the same figures as presented in the previous table, a one-year term deposit should have been credited one bp if originated pre-crisis and five bps if originated more recently. Similarly, a five-year term deposit should have received a credit of 10 bps if originated pre-crisis and 40 bps if originated more recently. Had an average cost of funds approach been employed, all term deposits would have received a credit of two bps if originated pre-crisis and eight bps if originated more recently, irrespective of their maturities. This example highlights the limitations of the average cost of funds approach. For banks employing this approach, it would have encouraged business units to raise short-term deposits rather than long-term, more stable sources of funding. Collectively, with the finding from above, this would have led to more structural liquidity risk on the balance sheet.
• Hot/volatile sources of deposits are often referred to as indeterminate maturity products, given the uncertainty surrounding their cash flows.
  • However, these types of deposits sometimes provide stable sources of funding. Demand deposits, for example, can be withdrawn at any time without notice. But, if all similar accounts were to be pooled and the behavior of the cash flows modelled over time, there would be a proportion that is rarely withdrawn (stable or core part) and a proportion that is more often withdrawn (hot or volatile part).
  • Making this distinction is important, because if a bank simply applied a MMM cost of funding approach, all demand deposits would only receive a credit based on the overnight term liquidity premium. Since this would be close to zero, business units would be discouraged from raising demand deposits.
  • A better approach would be to assign larger credits to core parts of funding, based on the modelled behavioral maturity, and smaller credits to hot/volatile parts of funding.

• Banks employing an average cost of funds approach would have no incentive to make the distinction between core and volatile parts of funding since, under this approach, the same credit for the benefit of funding is applied to all deposits, irrespective of their maturity.

Key Points Regarding The Approaches

• The average cost approach to LTP is simple, but has two major defects.
  • First, it neglects the varying maturity of assets and liabilities by applying a single charge for the use and benefit of funds.
  • Second, it lags changes in banks’ actual market cost of funding.

These defects essentially promote maturity transformation, which inherently exposes banks to more structural (mismatch/funding) liquidity risk.

• Overall, a matched-maturity marginal cost of funds approach promotes better LTP practice. It is more complex than the pooled average cost of funds approach, but it has some significant advantages.
  • First, it recognizes that the costs and benefits of liquidity are related to the maturities of assets and liabilities, and therefore allows higher rates to be assigned to products that use or provide liquidity for longer periods of time.
  • Second, it recognizes the importance of having changes in market conditions incorporated quickly and efficiently into the rate used to charge and credit users and providers of funds, and therefore relies on the actual market cost of funds. Banks should be encouraged to move towards this approach, if they are not already doing so.

Managing Contingent Liquidity Risk

• For many on-balance sheet items, calculating the charge for using, or the credit for providing, funding liquidity is quite straightforward. However, the same cannot be said about contingent commitments such as lines of credit, collateral postings for derivatives and other financial contracts, and liquidity facilities to name a few. In these cases, the best approach is to impose a scenario model, determine a reasonable low probability worst-case outcome and charge at the most granular level the transaction, product, or business unit for the costs of covering this outcome.
• Some contingent commitments that can potentially result in large, unexpected demands for liquidity are –
  • Retail deposit run-offs
  • Wholesale funding run-offs
  • Draw downs on lines of credit
  • Collateral Calls on Derivatives
  • Secured Funding run-offs

Banks carry a liquidity cushion, a “buffer” of highly liquid assets or, alternatively, stand-by liquidity to help them survive periods of unexpected funding outflows. A graphical illustration of this is depicted in this figure.

• Liquidity cushions are considered a fundamental principle for the management of liquidity risk. This is clearly outlined in Principle 1 of the BCBS Principles for Sound Liquidity Risk Management and Supervision (September 2008), and also reinforced by Principle 12, which states that “a bank should maintain a cushion of unencumbered, high quality liquid assets to be held as insurance against a range of liquidity stress scenarios, including those that involve the loss or impairment of unsecured and typically available secured funding sources. There should be no legal, regulatory or operational impediment to using these assets to obtain funding” (p 4).

• Extant guidance provided in association with liquidity cushions focuses mainly on
  • size,
  • composition, and
  • marketability

of the assets contained within the cushion.

• To ensure banks’ liquidity cushions are adequately sized, the BCBS recommends they be aligned with stress-testing outcomes that consider both idiosyncratic and systemic scenarios, plus a combination of the two. The level of stress assumed in the tests should reflect a bank’s overall risk tolerance. To assess their risk tolerance, banks should consider factors such as structural liquidity risk and the complexities of both on- and off-balance sheet business activities, which affect the cash flows.
• Composition – According to the BCBS, liquidity cushions should comprise “a core of the most reliably liquid assets, such as cash and high quality government bonds or similar instruments, to guard against the most severe stress scenarios” (p 30).

• Banks should also consider the marketability of these assets –
  • Assets that are more transparent are generally also easier to value, and the certainty surrounding this will inherently improve marketability.
  • In addition, assets that are central bank-eligible and/or have good market depth will generally be more marketable.
  • A bank’s reputation, credit rating and active participation in certain markets will also impact asset marketability

Problems in Liquidity Cushions

• The global financial crisis exposed some fundamental problems with banks’ liquidity cushions –
  • Very few used the results of stress-testing to determine the size of their liquidity cushion. Even those few banks determined the size of their liquidity cushion was based on outcomes stemming from idiosyncratic funding scenarios only, and not considering prolonged market- wide disruptions. This meant that cushions were inadequately sized to protect the banks from larger-scale unexpected (contingent) outflows.
  • The parameters used were too narrow, and were based purely on historical data. This meant that events that had not previously occurred were neglected.
  • Banks had liquidity cushions comprised of assets that were thought to be highly liquid, but were found to be highly illiquid and highly correlated. Some assets held as standby liquidity were not unencumbered, i.e. bank did not have legal claim over the asset or the asset was not entirely free from debt.
  • Nearly all of the banks included in the survey funded their liquidity cushions short-term (e.g. overnight), consistent with the perception that funding could be easily accessed and any market disruption would only be short-lived. While this minimized negative carry costs, it also provided banks little incentive to attribute the relevant costs back to the businesses that created the need to carry additional liquidity.

Better Management of Contingent Liquidity Risk

• In December 2010, the BCBS published two global standards for liquidity risk.
  • Liquidity Coverage Ratio (LCR) to ensure banks have sufficient high quality liquid assets to meet their daily net cumulative cash outflows during an idiosyncratic shock, for a period of one calendar month.
  • Net Stable Funding Ratio (NSFR) aimed at reducing banks structural liquidity risk by encouraging the use of longer-term funding of assets and other business activities.

• These two standards are aimed at making banks more self-sufficient and stable over a longer period is in part to reduce the burden of central banks having to act as the lender-of-last-resort, and the potential implications of moral hazard as a result of these actions.

• These measures are designed also to increase and/or improve the liquidity cushion.
• In one form or another, all banks that participated in the survey are enhancing the way they manage contingent liquidity risk –
  • Many are incorporating a wider variety of scenarios as part of their stress-testing processes to account for different types of market disruptions that might occur including both idiosyncratic and systemic funding shocks and a combination of the two.
  • Senior management are becoming more engaged with stress-testing results and using them as a basis for deriving the size of the liquidity cushion. The composition of assets in liquidity cushions is broadly improving, and many of the larger banks are now holding a larger proportion of cash and government securities than previously. This is most likely due to the development of the LCR.
  • Banks are applying higher funding charges to assets held as part of the liquidity cushion on the premise that it could take longer than expected to generate liquidity when needed. The charges applied depend on banks’ assumptions surrounding the length and severity of potential market disruptions. If, for example, a bank assumes funding markets for a particular asset remain stressed for two years, then the cost of holding that asset as additional liquidity should be based on the two-year term liquidity premium.
• The move to apply higher funding costs to liquid assets is considered significant for banks because carrying a more costly liquidity cushion creates more profit drag. The contingent liquidity risk embedded in various business activities should be examined and charges should be attributed based on their predicted, or expected, use of funding liquidity. Higher contingent liquidity charges should be applied to business activities that pose more threat to large and unexpected funding outflows. This process is depicted in this figure.

• The uncertainty surrounding future cash-flow demands stemming from contingent commitments makes it particularly difficult for banks to assess and price contingent liquidity risk. This is one of the reasons why it was neglected prior to the GFC. Some of the products that received little attention but then warranted significant funding included:
  • credit card loans and investments,
  • trading positions and derivatives,
  • revolving lines of credit, and
  • liquidity lines.

It is important for banks to understand that all contingent commitments need to be charged.

• At the most basic level of what is considered to be better practice, all banks should be charging contingent commitments based on their likelihood of drawdown. For example, suppose a line of credit with a limit of $10 million has $4 million already drawn. The rate charged for contingent liquidity risk should be derived as:

The limit and the drawn amount are already known. The likelihood of drawdown (sometimes referred to as a drawdown factor) should be assessed using behavioural modelling and should depend on factors such as customer drawdown history, credit rating of the customer, and other important factors. The cost of term funding liquidity cushion can be estimated from the corresponding year term liquidity premium.

For example – Assume there is a 60 per cent chance the customer will draw on the remaining credit and that the cost of term funding assets in the liquidity cushion is 18 bps (depicted from the three-year term liquidity premium in one of the previous tables).

The rate charged for the cost of contingent liquidity risk should be equal to:

Multiplying this by the limit of $10 million on the line of credit yields a dollar charge of $6,480.