In the preceding chapters, we examined the ALM interest-rate and liquidity risk to the bank. However, we did not specify which departments within the bank carry that risk. For example, we showed that if the bank has made long-term loans by using money from short-term savings accounts, the bank will suffer a loss if rates rise. However, we did not specify whether that loss should be counted as a loss on the loans, a loss on the savings accounts, or a loss to the fixed-income trading desk. The assignment of this profit and loss is important because it determines the risk and profitability of each business unit and product and therefore the price that the bank should charge its customers.
The way we determine where the risk is located is by using transfer pricing. Transfer pricing is a framework of internal transactions and payments between business units. For ALM purposes, the major units are the lending units, the deposit-taking units, the trading unit, and the ALM desk. The payments between the business units are fictitious in that they are recorded in the accounts for each unit, but no money actually leaves the bank as a whole.
The transfer payments significantly affect the measured accounting profitability of each unit. By affecting the measured profitability, we affect the prices that each unit must charge to its customers, and we affect the bonuses of the staff. If one unit is forced to pay a higher transfer price to another unit, the first unit’s measured profitability will fall, their bonuses will be reduced, and senior management may decide to scale back the activities of the less-profitable unit. For these reasons, transfer pricing can be highly political. Transfer prices, along with capital charges and limits, set the rules within which bankers, traders, and business units must play.
Transfer payments between units are made for one of two reasons: services provided or funds provided. Transfer payments for services is not part of ALM. In this chapter, we are concerned only with funds-transfer pricing, which is also often called the cost of funds (COF). Funds-transfer pricing can be viewed as the interest payments charged when one unit lends funds to another. It is the structure of funds-transfer pricing which moves interest-rate and liquidity risks between units.
In this chapter, we first describe the traditional way of calculating funds-transfer prices and the associated problems of this traditional approach. The bulk of the chapter then discusses matched-funds-transfer pricing, which eliminates most of the problems. In discussing matched-funds-transfer pricing, we first show how it can be used to eliminate or move interest-rate risks, and we give the general rules for transfer pricing. We then discuss the difficulties of transfer pricing for indeterminate maturity products such as checking accounts. Most of the discussion deals with transfer pricing for debt costs, but towards the end we add the complication of transfer pricing for economic capital. Transfer pricing deals with the conceptual movement of funds between accounting entities to measure profit and risk. At the end of the chapter, we discuss how the real business groups within the bank access funds and manage risk.
A typical situation in a universal bank is that the retail banking group takes in deposits and lends them out to retail customers. The amount of deposits generally exceeds retail loans, so the excess is given to the bank’s ALM desk. The ALM desk is closely associated with the trading group. It is sometimes referred to as the “funding desk” or “treasury,” although the term “treasury” may also be used to refer to the whole trading group. The ALM desk takes the excess funds from the retail unit and lends them out. It may lend them to another bank in the interbank market or to another group within the bank, such as commercial lending or the trading group. In return for providing these funds, the retail group receives some interest payments from the ALM desk. In the traditional framework, the transfer rate for these interest payments is typically either the overnight interbank rate or the retail deposit rate, plus a small spread to cover operating expenses. This traditional transfer-pricing framework has several negative consequences.
First, within the retail banking group, there is no clear line between the profitability of retail loans and deposits. If the retail group as a whole is profitable, it is not clear whether the profit is driven by raising cheap funds from deposits or giving well-priced loans. Consequently, it is not clear whether to expand either the deposit program, the loan program, or both.
Second, the interest rate given to the retail group for their excess funds tends to be lower than the rate they would have received if they had been able to lend the funds directly into the interbank capital market. This falsely reduces the measured profitability of the retail unit. The lower rate also falsely boosts the measured profit of the trading unit or commercial loan unit who would have paid more for their funds if they had been forced to borrow them from the interbank market.
Third, traditional transfer pricing makes it difficult to monitor and control interest-rate risk because a change in market rates may affect the profitability of all business units. Even if the commercial lending department had done a good job in making well-priced, well-structured loans to creditworthy customers, the department may still suffer a loss if their funding costs suddenly rise.
To avoid these problems, a transfer-pricing framework is needed that recognizes the true value of the funds and concentrates the interest-rate risk into one unit: the ALM desk. This can be achieved by matched-funds-transfer pricing.
To introduce matched-funds-transfer pricing, let us start with an example using traditional transfer pricing, and then show how matched-funds-transfer pricing can be introduced to bring clarity to the risk and profitability.
Consider a traditional bank raising funds in the form of 3-month deposits (FDs) and lending 5-year, fixed-rate loans. If the bank pays 4% for the FDs and receives 11% for the loans, the nominal net interest margin (NIM) is 7%. This is illustrated in Figure 15-1.
The 7% spread between the loans and deposits should cover the administrative costs, the credit loss on the loan, and the interest-rate risks due to the mismatch between the 3-month funding and 5-year lending. There are two problems with this situation. One, it is not possible to attribute profitability separately to the loans and FDs. Two, there is interest-rate risk because if rates rise to 8% in 3 months, the bank will find itself with a net interest margin reduced to only 3%, which would not be enough to cover expenses.
As an alternative, consider taking in 3-month FDs, and lending the resulting funds for 3 months in the interbank market at the current available rate, e.g., 6%. Then to fund the loans, borrow money for 5 years in the capital markets fixed at the current available 5-year rate, e.g., 7%. This is illustrated in Figure 15-2.
FIGURE 15-1 Example of Traditional Transfer Pricing
FIGURE 15-2 Transfer Pricing via the Interbank Market
In this arrangement, the bank has no interest-rate risk because the 3-month liabilities are matched with a 3-month asset, and the 5-year loans are matched with a 5-year liability. Furthermore, we can clearly see the profitability of each product. The net interest margin for the FDs is clearly 2% (6%–4%), and the NIM for the loans is 4% (11%–7%). The profitability for the FDs is therefore 2%, minus the FD administrative costs. The profitability for the loans is 4%, minus the loan administrative costs and the costs of credit risk.
Figure 15-3 illustrates how the rates are set relative to the current yield curve for interbank lending. This figure very clearly shows the spread that each unit receives. By funding the bank’s assets and liabilities with matched interbank assets and liabilities, we have removed the interest-rate risk from both the business units and the bank as a whole, and we have brought clarity to the profitability of each product.
Notice that the bank has lost 1% of income to the capital markets, but in exchange, it has no interest-rate risk and therefore does not need to hold capital against ALM risks. You may also notice something strange about this bank. Its net value is zero; i.e., it has no capital to cover other risks such as loan defaults. We return to the complications of capital later, but for now let us assume that the loans are risk free and concentrate on the transfer pricing required for the cost of raising debt.
FIGURE 15-3 Pricing of Internal Transactions Relative to the Yield Curve
In practice, it would be difficult and costly to implement a structure in which every individual deposit is lent separately into the market and every loan requires funds to be individually borrowed from the market. The practical alternative is to set up an internal market that aggregates all the individual transactions, and only to use the external market to borrow or lend the net amount. This is illustrated in Figure 15-4, in which the business units interact with the ALM desk as if it were the interbank market.
This concept of having an ALM desk that acts as internal reflection of the external market is the foundation of matched-funds-transfer pricing. In this framework, every business unit interacts with the ALM desk as if it were an external counterparty. The units give fictitious loans to each other, and the interest payments that they receive on the loans are the same as the rate they would have received by lending the funds in the external capital market.
The fictitious loans and trades are entered in the accounting system as if they were real trades with an external counterparty, except that both counterparties are different business units within the bank.
Consider the example of the commercial lending department making a five-year, fixed-rate loan to a customer. The department would borrow at a fixed rate for five years from the ALM desk. The rate quoted to the department for that internal loan would be the rate that it would cost the ALM desk to borrow the money in the interbank market for five years at a fixed rate. When the commercial lending department borrows from the ALM desk, a fictitious liability is created for the lending department. This is mirrored by a fictitious asset created for the ALM desk.
FIGURE 15-4 Transfer Pricing Using an Internal Reflection of the Market
The department then lends the funds to the client, charging them the five-year interbank rate, plus a spread to cover operating expenses and the credit risk. This creates a real asset for the department. The resulting situation is that the lending unit has a real asset and a fictitious, matching liability. The ALM desk has a fictitious asset that has the interest-rate characteristics of the loan, but without credit risk.
On the day of making the loan, the commercial lending department knows that for the next five years it will receive fixed-interest payments from its client and make fixed-interest payments to the ALM desk. At the end of five years, the client will repay the loan, and the lending unit will use the proceeds to repay the ALM desk. The lending department therefore has locked-in its profit for this transaction and knows that the profit will not change if general market rates change.
By matching the real assets and liabilities with fictitious liabilities and assets of the same maturity, the business units are hedged, and changes in interest-rates will only affect the profitability of the ALM desk. The managers of the ALM desk chose how they wish to deal with this net interest-rate risk. They may choose to hedge it by doing real trades with the external capital market, or they may choose to keep it, thereby gaining a net expected profit but requiring the bank to hold additional economic capital to protect it from interest-rate risk.
Now that we have seen how matched-transfer pricing works for a specific example, let us lay out some general rules for setting up a matched-funds-transfer pricing framework.
1. For the purpose of transfer pricing, the funding requirements for all transactions are considered to go through the ALM desk. This process is illustrated in Figure 15-5. Notice that the business units do not just go to the ALM desk for their net requirements. Each business unit gives all of its deposits to the ALM desk to be invested at market rates, and goes to the ALM desk for all its funding requirements if it wishes to make loans.
2. For every transaction, there is an agreement between the business unit and the ALM desk about the terms of the fictitious asset or liability. These terms are the same as would be agreed between the bank and an external counterparty. The terms specify the amount, the repricing frequency, the time for final repayment of the principle, any amortization, any prepayment options, and the rate, which is the current market rate. These terms should mirror the interest-rate characteristics of the business unit’s transaction with the customer.
3. For each transaction, the business unit receives a fictitious asset (liability) and the ALM desk receives the opposite fictitious liability (asset).
FIGURE 15-5 Conceptual Transfer of Funds between Units for Transfer Pricing
4. The trading unit is a special case. Like the other units, any transaction between the trading unit and the ALM desk has the price fixed according to the effective maturity of the loan. However, unlike the other business units, the trading unit has access directly to the interbank market and is not required go to the ALM desk to match every transaction. For risk-measurement purposes, any fictitious liability that the trading unit has goes into the trading VaR calculator, and the corresponding fictitious asset goes into the ALM simulation.
After all the internal transactions have been conducted, each business unit has a balance sheet showing its actual assets and liabilities and interest-rate matched fictitious liabilities and assets. The accounting unit charges and credits each business unit with the internal interest payments established by the matched-funds-transfer-pricing agreements and the consequent assets and liabilities on each unit’s balance sheet. Figure 15-6 shows an example of a balance sheet. The real assets and liabilities are shown in bold. The fictitious liabilities and assets are denoted by an asterisk (*).
Let us examine this balance sheet in detail. Starting with the trading unit, we find that it has $30 billion in assets created by long positions, e.g., owning equities. It has real liabilities of $20 billion in short positions and $5 billion in money that it has borrowed from other banks in the interbank market. It also has a fictitious liability for $5 billion that it has borrowed from the ALM desk for 1 year. This appears as an asset to the ALM desk.
FIGURE 15-6 Balance Sheet for Transfer Pricing
The commercial lending group has made a series of loans to its customers. These are matched by a series of fictitious loans of similar maturities and almost the same value. The difference in value is because the true loans are prepayable, whereas the fictitious loans are not normally prepayable. However, the prepayment risk is captured by an option which allows the business units to put the fictitious notes to the ALM desk in case the customer prepays. The $10 billion of deposits from commercial customers are paid rates that are close to the interbank rate, and are therefore hedged with an overnight loan.
The retail unit has $40 billion of assets in the form of car loans. However, the interest-rate on these loans floats and is tied to prime. Because prime is an administered rate, it is not quite the same as any single market rate, and is therefore best hedged with a series of loans of different maturities, as we discuss later. The retail unit also has $100 billion of liabilities in the form of checking accounts. Although these accounts are demand deposits, it is unlikely that the whole amount will be withdrawn for at least 2 years, so the bank can be confident in lending out the deposit funds as a mixture of loans from overnight to 2 years.
Finally, let’s look at the ALM desk. It has the reflection of all the fictitious assets and liabilities of the other business units, plus it also has some real assets and liabilities of its own. In this case is has lent $8 billion in the form of various interbank loans and has borrowed $5 billion at other times. It also has a swap that it is using to modify its interest-rate risk, and it has issued $10 billion of 5-year bonds in the name of the bank.
In this concept of transfer pricing, the net value of each business unit is zero, and all the available capital to support the business units is concentrated on the ALM desk. We discuss this and its alternatives in more depth later. But now, let us further investigate the strange mixture of loans that was used to hedge the checking accounts and prime-based loans.
For products such as fixed-rate loans, it is relatively easy to find traded instruments in the interbank and capital markets that can be used to hedge the interest-rate risk. However, in the balance-sheet example, we introduced the problem of dealing with products such as checking accounts and prime-based loans that have indeterminate maturities and no fixed relationship to traded instruments. Products such as these are best hedged using a mixture of instruments.
The hedge can include not only bonds, but also options. For example, if customers are able to prepay when rates fall, the lending unit and the ALM desk should enter into an agreement in which the ALM desk agrees to take any prepaid funds. The ALM unit thereby commits to reinvesting the funds at the new market rates, without passing on any losses to the lending unit. Effectively in this arrangement, the lending unit has the right to prepay its loan from the ALM unit. In exchange, the ALM unit will charge an option premium to the lending unit when it first funds the prepayable loan. The ALM unit is left to manage or hedge the option as it sees fit.
Determining the best hedge for an indeterminate product is difficult. The best way is to model the payments from the product as described in the chapter on interest-rate risk. The product can then be put into the simulation model along with a series of traded instruments, such as loans and options. The best hedge is then the combination of traded instruments that best matches the product. In matching the product we have several choices. If the performance of all the business units is being measured according the value created each year, then the best hedge is the one whose change in value over one year is most similar to the change in value of the product. An alternative way of saying this is that the best hedge is the set of instruments that, when combined with the product, results in the minimum possible variation in value.
In the simulation, it is relatively easy to calculate the change in value of the traded instruments because we have pricing equations (e.g., the bond equation) that give their value as a function of the rates prevailing at the end of one year. However, for the indeterminate-maturity product, the value at the end of the year must be determined by a new set of nested simulations, as explained in the chapter on interest-rate risk.
After applying matched-funds-transfer pricing, the lending units should be left with only credit and operating risk, and the deposit-taking units are left with only operating risk. The ALM desk has the structural interest-rate risk, and the trading group has general market risk, counterparty-credit risk, and operating risk. This is illustrated in Figure 15-7, where the extent of shading reflects the amount of risk.
In our discussion so far, we have not specified how transfer pricing treats capital for credit and operating risk. For example, we implicitly assumed that the funding for a loan would be in the form of debt raised in the interbank market. However, if all loans were supported by 100% debt, the bank would not be able to repay the debt holders if any of the loans defaulted. Banks must therefore fund their customers’ loans from a mixture of debt and equity.
The total funding requirement is the amount of the asset (A). The amount of equity (E) equals the economic capital required to support the credit risk of the loan. The remaining funding requirement is made up of debt (D):
A = E + D
FIGURE 15-7 Distribution of Risk after Transfer Pricing
The extent of the shading reflects the amount of risk remaining in the business unit after transfer pricing.
This mixture of funding can be reflected in the transfer pricing in 1 of 2 almost equivalent ways. One way is that the business units can be charged their debt rate for 100% of their funding requirement. The required capital can be held by the ALM desk in the form of additional assets that can be sold to cover any losses in the business unit. These additional assets are safe assets (e.g., government bonds). Because there is the possibility that they will be sold to cover losses, these assets must be bought by raising equity from shareholders. As these are safe assets, they only give a return equal to the debt rate (e.g., 6%); however, the shareholders demand a high return equal to the hurdle rate for capital, e.g., 16%. The difference (e.g., 10%) is charged to the business units who create the risk requiring the capital to be held. The total charge to the business unit is then the full amount of the asset multiplied by the debt cost, plus the amount of equity times the required excess return:
Cost = A × rd + E × (re – rd)
This approach was illustrated in the balance sheet in Figure 15-6.
An alternative approach is to assume that the capital is held by each business unit rather than by the ALM desk. In this model, the business units are funded with a mixture of debt and equity. For example, a $100 loan requiring $8 of economic capital would be funded with $92 of debt and $8 of equity. The total charge to the unit would then be the amount of debt times the debt rate, plus the amount of equity times the equity rate:
Cost = D × rd + E × re
This is equivalent to the previous formulation:
Cost = D × rd + E × re
= D × rd + E × rd – E × rd + E = re
= A × rd + E × (re — rd)
Figure 15-8 shows the balance sheet modified to have the equity held in the business units. Notice that the real assets and liabilities are the same. In this version, the ALM desk raises a mixture of debt and equity to fund each business unit. For assets such as loans, the amount of the fictitious debt is reduced and replaced by a transfer of equity.
For liabilities, the reasoning is less obvious because the amount of fictitious debt is increased. One way to think about this is that for a bank to take in deposits, it must first raise equity to cover the possibility of losses due to operating risks. It can then raise deposits. It then invests the deposits and the equity in the interbank market. When the deposits are to be repaid, the bank will be able to use the interbank investments to pay back the investors and pay any unusual costs due to operating risks. If there were no significant operating losses, the bank would be able to repay the equity holders, plus give them the profits from the deposit-taking business.
FIGURE 15-8 Balance Sheet Showing Alternative Treatment for Capital
Notice that in the new balance sheet, the equity owed to the shareholders is tagged closely to each business unit, and the only equity “belonging” to the ALM desk is $2 to cover economic capital required for the ALM-mismatch risks.
The approach taken gives balancing accounts for each business unit. However, the balance sheet for the ALM unit is not an exact reflection of the balance sheet for the bank. The bank has $169 billion of real assets and liabilities, but the ALM balance sheet shows $147 billion.
This is because we used the net amount of the assets and liabilities in the trading group rather than the gross amounts. If we use the gross amounts, and enter economic capital for deposits as a negative asset, we can create the balance sheet in Figure 15-9, in which the total assets and liabilities of the ALM desk equal the total real assets and liabilities of the bank ($169).
FIGURE 15-9 Balance Sheet Reflecting All the Bank’s Assets on the ALM Desk
In dealing with transfer pricing, we have described the conceptual framework that is used to set transfer prices and measure risk. In the conceptual framework, all the funds flow through an ALM desk and the “invisible hand” of the transfer-pricing rules sets all the transactions to be at market rates. This works for measuring risk and profitability, but it is not how funds are handled in reality. Let us now look at the real organizational units within a bank that are responsible for operating the ALM system.
Although many groups throughout the bank are involved with asset liability management, the main ones are as follows:
• The Senior Risk Committee
• The Asset Liability Committee (ALCO)
• Asset Liability Manager
• The ALCO Support Group
• The Money-Market Funding Desk
• Financial Accounting Unit
• Business Units
We will look at their roles and responsibilities in detail.
The senior risk committee monitors all the risks of the bank. As a result, they are interested in the risk arising from the ALM book and the correlation of that risk with other risks, especially interest-rate positions on the fixed-income trading desk.
Many executives on the risk committee are also in the ALCO. The ALCO is typically chaired by the CFO, head trader, or in smaller banks by the CEO. The members of the ALCO typically include the heads of the business units. The ALCO typically meets monthly to review the structural interest-rate and liquidity positions of the bank. They also review transfer-pricing policy and decisions on administered rates, such as changes in the bank’s prime rates or retail fixed-deposit rates. If the committee wishes to change the risk profile significantly, they will agree with the CFO to carry out large capital-markets transactions, such as issuing bonds or securitizing and selling parts of the balance sheet. Alternatively, the committee may agree that the business units should change the risk profile by modifying the mix of products that are offered to customers, e.g., by changing rates and marketing to encourage or discourage different types of products.
Smaller-scale tactical changes are delegated to the asset liability manager. In general, the structural interest-rate position should be mostly hedged, unless the ALCO decides that the expected return from “playing the gap” is sufficient to give a good return on the economic capital consumed by the risk.
The asset liability manager (AL manager) is a mid- or senior-level trader who has the daily responsibility of maintaining the interest-rate and liquidity profile targeted by the ALCO. This is done based on risk reports from the ALCO support group and is carried out by ordering trades such as swaps and interbank loans for the ALM book. The ALM book is an accounting entity holding all the fictitious assets and liabilities. The orders for the book may be placed directly with the market or, to increase efficiency, they are more commonly placed with the bank’s own trading desks. The desks then execute the trades and transfer the security to the ALM book. In some institutions, the ALM book is managed as part of the bond desk, but more typically in large institutions, the ALM book is managed separately. The personnel staffing the ALM desk may be the same as those on some of the bond desks, but they are conceptually different.
If the ALM manager did nothing to hedge the ALM book, it would receive the spread between the long and short interbank rates, and it would run the risk of earnings compression if the rates moved. Because it is now the center of ALM interest-rate risk, it attracts the charge for the economic capital consumed by that risk, and should earn the hurdle rate on the capital. As the rates for the fictitious transactions are the rates at which the bank could borrow or lend in the interbank market, the ALM desk could theoretically hedge itself to be risk free with no cost. To hedge itself, the desk can use the usual fixed-income transactions such as buying or shorting bonds, or use derivatives such as futures, swaps, and options.
The ALCO support group is a team of analysts reporting to the AL manager. They provide weekly risk reports to enable the ALCO and AL manager to maintain the target profile. They also carry out analyses to estimate the effective maturity of the indeterminate-maturity products such as checking accounts. This information on duration is used to create the risk reports and is also supplied to the financial accounting group to determine the rates for funds-transfer pricing. In addition to identifying the sources of risk, the ALCO support group should suggest hedges to be approved by the AL manager and the executives on the ALCO. Hedges may include market transactions or changes in customer products.
Conceptually, for transfer-pricing purposes, we said that all requests for funds from the business units are made to the ALM desk. In practice, daily requests for short-term funds are made directly from the business units to the money-market funding desk. For accounting purposes, they are later recorded as if they went to the ALM desk and then the ALM desk requested funds from the money-market desk.
The money-market desk is therefore responsible for sourcing and placing short-term funds to fulfil the cash requirements of the bank. This is a very reactive role, in which the desk receives funds or requests for funds from the rest of the bank, and goes to the overnight interbank market to place or borrow the funds at the best available rates. The money-market funding desk operates on a very quick timescale of hours to ensure that the daily funding requirements are balanced, whereas the time scale for managing the structural interest-rate position of the ALM book is several weeks.
The financial accounting group reports the profitability of each business unit, and therefore needs to be supplied with the rules for transfer pricing, the effective transfer rates to be applied, and the details of which transactions have happened.
The individual business units, such as retail lending, are the ones creating the interest-rate mismatches. The business units have great personal interest in the transfer rates because they affect the units’ profitability and the rates they can give to their customers. Changing the rates changes the profitability and the volume of business for each type of product. This change in volume affects the AL risk profile. For example, if the bank found that the basis risk for prime was consuming too much economic capital, it could ask the retail business to better balance the assets and liabilities based off of prime. This could be done as a decision in the ALCO, and by a capital charge made for the basis risk.
In this chapter, we explained how matched-funds-transfer pricing can be use to bring clarity to risk and profitability measurement. This concludes the discussion of ALM. Next, we turn our attention to credit risk, starting with an explanation of what it is and why it needs to be measured.