Iain Ritchie outlines his work on funding liquidity risk at the Actuarial Research Centre
Banks are in the business of maturity transformation - borrowing short, that is, taking on customer deposits, and lending long - issuing mortgages. This exposes the bank to funding liquidity risk. It is an inherent part of the business model of fractional reserve banking, and while it can be controlled, it cannot be entirely eliminated.
Funding liquidity risk is different from market liquidity risk. Funding liquidity risk is the risk that a bank will be unable to pay its debts when they fall due. In simple terms, it is the risk that the bank cannot meet the demand of customers wishing to withdraw their deposits. Market liquidity risk, on the other hand, is the risk of not being able to sell assets in a timely fashion without having to offer a heavy discount. Research has shown that funding liquidity issues can often lead to market liquidity risk and vice versa.
Poor management of funding liquidity was one of the reasons that ultimately caused banks to fail, or need assistance from their respective governments, during the financial crisis in 2007-2008. A major component of this is how liquidity risk was priced internally by the bank and how that flowed through to the pricing of customer products.
This article will provide an overview of my research on how banks manage their funding liquidity risk. The significant interest in this area inspired me to study towards a PhD on this topic, carried out at the Actuarial Research Centre (ARC).
For banks to pay their liabilities as they fall due, they need to be able to meet expected levels of withdrawals, plus have additional liquidity resources to meet unexpected withdrawals. Liquidity resources include:
- Deposits at the central bank;
- Short-dated liquid assets such as treasury bills that provide liquidity on maturity;
- Longer-dated liquid assets such as gilts that can be pledged as collateral in sale and repurchase (repo) transactions;
- Contingency funding lines, where the bank has pre-agreed access to a line of credit with another institution (usually interbank lending);
- Assets available for sale. This can include books of mortgages that can be securitised.
The first four do not require an asset to be sold, and hence are not subject to market liquidity risk. The most useful liquid assets have minimal credit risk and can be easily realised in stressed market conditions, either through sale or repo. The downside is that they have low yield. Some higher yielding assets that do have credit risk (for example, investment grade corporate bonds or retail mortgage backed securities) or are volatile (equities) may be included in the bank's definition of liquid assets but appropriate haircuts (reduction) to the value of the assets need to be applied.
Banks must find a balance between holding enough liquid assets to meet unexpected funding needs versus the higher yield available from less liquid assets. The liquidity coverage ratio (LCR), published by the Basel Committee on Banking Supervision, specifies the type of assets that the bank can classify as high quality liquid assets (HQLA) and sets the minimum amount of liquid assets that must be held - enough to cover net cash outflows for 30 days under stressed market conditions.
Funds transfer pricing
Banks should assess their liquidity costs upfront and include this cost in pricing of products. They do this via their fund transfer pricing (FTP) framework - a method for assigning revenue and costs associated with interest rates and funding liquidity. It provides the various business units within the bank with an understanding of funding liquidity risk, motivating those units to proactively reduce liquidity risk.
Figure 1 shows a simplified example of how the FTP framework works, using the example of a two-year fixed-term deposit funding a five-year fixed-term loan. The deposit unit brings in funds, paying them to the treasury unit in exchange for a revenue credit. This credit is the FTP rate, and depends on the duration of the funds. When the loan unit wants to lend funds to customers, it borrows funds from the treasury unit, with treasury charging an FTP rate that again depends on the duration of the funds. The difference in FTP rates for providing and using funds relates to the duration mismatch between borrowing and lending, and compensates the treasury for managing interest rate and liquidity risk.
If the FTP framework is set appropriately, the cost of liquidity risk across the bank can be assigned to each business unit, incentivising all to manage liquidity risk for the overall benefit of the bank. If the framework is not set appropriately, business units can work in their own interests, increasing liquidity risk for the bank as a whole.
Formalising the FTP
For FTP rates to be set appropriately, they need to include interest rate and funding liquidity risk costs. To do this, we will consider a simple one period model to maximise profits:
P is Profit for the bank;
L is Loan amount; iL is rate charged on loans;
ML is amount loaned in wholesale money markets; WL is rate available for lending in wholesale money markets;
A is amount of liquid assets; iA is expected return on liquid asset holding;
D is deposit amount; iD is rate payable on deposits;
MB is amount borrowed in wholesale money markets; WB is rate for borrowing in wholesale money markets.
If we assume the bank is deposit rich, i.e. loan to deposit ratio less than 100%, we can rearrange the formula and derive the appropriate FTP rates:
Where a is the percentage of deposits that must be held as liquid assets.
Similarly, if the bank is deposit poor, that is, loan to deposit ratio greater than 100%, the appropriate FTP rates can be derived. Although this is only a simple one period model, and not representative of reality, it does give us insights into how to set the FTP rates and demonstrates the benefits of using the FTP framework. In particular, FTP makes it possible for the business units to work independently of each other while still maximising the return for the bank.
Setting FTP rates across multiple time periods is more complex.
If banks knew exactly when customers were going to withdraw their money, the one time period model could easily be extended to allow for withdrawal at each time period. The difficulty lies in that we have to understand and model customer behaviour and allow for the associated uncertainty. Granting customers the option to withdraw their money immediately (current accounts), or allowing prepayment of their mortgage, has a cost to the bank. As a result, we need to allow for this cost within the FTP framework. We can calculate the cost of these customer options by simulating all possible outcomes.
Overall, the FTP framework allows the business units to better understand the cost of products and embedded liquidity options they are offering to their customers. It also provides a properly quantified method of incorporating liquidity costs into customer pricing, lower liquidity risk for the bank by reducing maturity mismatch and nudging consumers towards products which have lower liquidity risk. Ultimately, it reduces the systemic underpricing of liquidity that prevailed prior to the financial crisis.
More details on this and other ARC research projects can be found at: bit.ly/ARCresearch. Thank you to professors Andrew Cairns and Alex McNeil at Heriot-Watt University, and also Karen Brolly and Garry Smith at Hymans Robertson for their industry support.