Transfer Pricing

14.1 Objectives of Transfer Pricing

14.2 Overview of "Matched-Maturity" Transfer Pricing
14.3 Advantages of the Matched-Maturity Transfer Pricing Method
14.4 Selecting a Transfer Pricing Yield Curve
14.5 Target versus Actual Credit Rating of the Bank
14.6 Applying the 80/20 Rule to Transfer Pricing
14.7 Use of Guaranteed Product Spreads
14.8 Advantages and Disadvantages of the 80/20 Rule
14.9 Prepayment Risk
14.10 The Liquidity Commitment Spread
14.11 Spread or Basis Risk
14.12 A Survey of Transfer Pricing Adjustments
14.13 Transfer Pricing Items Without Maturities: DDA and Equity
14.14 Hedging Strategy for the Large Transaction Book
14.15 Summary

319
320 Chapter 14

Internal funds transfer pricing is practiced to some degree by every bank.

Some have elaborate mainframe systems with automated data feeds from
their applications systems. At the other extreme, some bankers denigrate
the concept, but practice it nevertheless. They do so whenever they con
sider whether to make a loan or buy an investment security. They do so
whenever they think about the spread they are earning on a commercial
loan. They do so whenever they are considering the margin implications of
their retail CD pricing strategy.
Every banker should understand the basic principles of transfer pricing.
They are an essential part of any type of profitability measurement, pricing
analysis, or risk-adjustment for any balance sheet product. Those that criti
cize the topic are usually only upset about others who get carried away by
theoretical or implementation complexities. We agree. Those that desire
"precise" transfer pricing are violators of the 80/20 rule!
As with the other aspects of profitability measurement, there are a myr
iad of different, legitimate perspectives on this topic. In the context of this
book, the main objective of internal funds transfer pricing is to risk-adjust
the net interest income or margin for a line unit or product for profitability
analysis purposes. Emphasizing the principles elaborated on in Section
13.1 will provide important guidance in developing the procedures de
scribed in this chapter. The discussion that ensues is not the "only" way;
rather it is one of many possible paths. Other perspectives may be found in
Banker's Treasury Management Handbook (Binder, Barrett F., ed., Boston,
MA: Warren, Gorham and Lamont, 1988).

14.1 OBJECTIVES OF TRANSFER PRICING

There are four main objectives in any complete transfer pricing system:
remove interest rate risk from line units and products, reflect "risk-ad
justed" spreads over long-term interest rate cycles, centralize all interest
rate mismatches of the bank in one (ALCO) unit, and provide consistent
guidance in product pricing analyses.

Remove Interest rate risk from line units and products. Line unit
managers may have some discretion of product pricing, but they have no
control over market yield curve shifts or unusual behavior of index rates,
such as the Prime rate. In the spirit of leaving only controllable variances
in their
Transfer Pricing 321

profitability reports, it is essential that the transfer pricing system remove

all unmanageable interest rate risk from their results. This requires guaran
teeing a net spread to the unit for every product over each budgeting cycle,
generally a one-year period.

Reflect "risk-adjusted" spreads over long-term Interest rate cycles. Even if a

net spread on a product is guaranteed for one year, if that spread shifts
unexpectedly between years, the sudden adverse effect on a line unit could
be substantial. Therefore, determine spread relationships over full interest
rate cycles, making the year-to-year changes in net spreads minimal. How
ever, this practice will result in greater volatility in the ALCO or Treasury
unit where all interest rate mismatches are transferred. This will be dis
cussed later.

Centralize all interest rate mismatches of the bank in one (ALCO)

unit. Transfer pricing does not actually remove interest rate risk from a
bank. It transfers it from one unit to another, a concept explained in the
next sec tion. For now, it is most effective to transfer all interest rate
mismatches in the bank into one unit for analysis and management. This is
usually an organizational unit referred to as the ALCO unit or Treasury
unit.

Provide consistent guidance in product pricing analyses. The detailed

pro cedures to be described for transfer pricing in the profitability system
are not exactly applicable to product pricing analyses or pricing
spreadsheets. There are some critical differences that will be described.
Nevertheless, the general principles are the same and maintaining
consistency with these is essential.

14.2 OVERVIEW OF "MATCHED-MATURITY" TRANSFER

PRICING
The concept of matched-maturity transfer pricing is best illustrated by ex
ample. Consider a "bank" with the balance sheet shown in Table 14.1. The
reader will immediately notice that this "bank" has no equity. Transfer
pricing equity will be discussed in a later section.
322 Chapter 14

Table 14.1
Transfer Pricing Case Study
Matched-Maturity Method

Interest
Maturity Income
Item Unit Balance {Years} Rate {Ex ense}
Asset:
Commercial Loan Corporate $100 1.00 9.0% $9.00
Liability:
Certificate of Deposit Retail $100 0.25 5.0% (5.00)

Total Bank: 4.0% $4.00

Note: Assume that each transaction is a "buller or zero coupon ttem, where all
interest and principal is paid at maturity. All items are fixed rate transactions. The
dollar net interest income and expense figures will occur only H interest rates
remain stable over the course of one year.

Bankers over the years have devised numerous shortcuts for detennin ing
how to split the bank's total 4% net interest margin between the two line
units. Briefly, some of the more popular method are as follows:

Do nothing. By this approach, there is no internal transfer pricing and

there is no attempt to divide or allocate the total bank net interest income
among the line units. Generally, for budgeting activities, the net interest
income budget is housed in a finance or Treasury unit. Line units are only
ex pected to budget product outstandings and portfolio rates or yields.

Single market (or single pool) rate. This is the simplest method for
deter mining a net interest spread or net interest income for a line unit.
Unfortu nately, no single market rate can ever be appropriate. Yet, it is
surprising how many banks do a crude form of transfer pricing using the
federal funds rate or a moving average of a short-term LIBOR or CD rate.

Multiple pool rate(s). This is basically a "gap" approach to transfer

pricing where each product is classified in one of several pools with
differing ma turity ranges. Indeed, alignment of this approach with the
bank's interest rate gap bucket definitions has significant advantages. The
transfer rate assigned to the product is the respective pool rate for its
maturity range.
Transfer Pricing 323

This approach is really not at all bad. Keeping its limitations in mind,
acceptable results can be obtained with a very modest effort.
The co-terminus, or matched-maturity, method may be considered to be
a much more detailed and refined version of the multiple pool approach.
Its basic tenet is that every incremental customer transaction should be
"matched" with a corresponding hypothetical internal funds transfer. This
matching concept should include the following attributes:
■ The transfer funds should mirror the expected cash flow pattern of
the original transaction, including amortizations and/or prepayments.
■ The interest rate assigned to the transfer funding should be consistent
with the marginal cost of a large block of wholesale funding at the
bank's cWTent marginal funding rate(s) for the cash flow pattern ex
pected.
Using the example from Table 14.1, to apply the matched-maturity
method, we must determine the marginal cost of funds for each maturity
listed. Assume that the relevant rates are 6.0% for three-month funds and
7.0% for one-year funds. With this information, we can generate the nec
essary funds transfers as shown in Table 14.2.

Table 14.2
Matched-Maturity Funds Transfers
for the Case Study Situation In Table 14.1

ASSETS LIABILITIES
Corporate Banking:
100 One-Year Loans at 9.0% 100 One-Year Transfer Liabilities at 7.0% •
Retail Banking:
100 Three-Month Transfer Asset at 6.0% ** 100 Three-Month CDs at 5.0%
ALCO Unit:
100 One-Year Transfer Asset at 7.0%* 100 Three-Month Transfer Liabillties at 6.0% **

Note: • and •• denote matched pairs of funds transfers.

Notice that for each actual customer transaction, there is a matched pair
of funds transfer entries. For the one-year loan, a transfer liability is
booked for the same amount and maturity. This transfer liability carries the
one-year transfer rate of 7.0%. An offsetting transfer asset entry is made
324 Chapter 14

on the books of the ALCO unit. It will have the same balance, maturity
date, and interest rate as the transfer liability. This matched pair of transfer
entries ensures that when the books of all units are consolidated, all inter
nal funds transfers are eliminated.
An analogous pair of entries is booked in response to the three-month
CD. In this case, a three-month transfer asset is entered for Retail Banking
and an offsetting transfer liability is created for the ALCO unit. The result
is that each unit now has a "balanced" balance sheet and a "locked-in" net
interest spread for each product. Most importantly, the basic interest rate
mismatch being undertaken by this bank is now transferred into the ALCO
unit, the area responsible for interest rate risk management, and the mis
match spread is accurately quantified. Management should always under
stand what portion of their net interest margin is attributable to such
interest rate mismatches. These "earnings" are very volatile. The final
division of the overall bank's margin of 4% is diagrammed in Figure 14.1.

Figure 14.1
Matched-Maturity Transfer Pricing
I0'J, r-----------------------

9'J,

S'J,
2% LOAN SPREAD

7'J,
YIELD CURVE

6'J,

1% MISMATCH SPREAD

1% CD SPREAD

J'J, L _J --'- , _. j J

0 0.25 0.50 0.75 1.00 1.2!1

MATURITY (YEARS)

It is critical to understand that the mismatch spread of 2% is by no

means assured beyond the initial three-month period where the CD rate is
known. The risk of this balance sheet mismatch will become apparent at
each rollover date of the CD over the life of the loan.
Transfer Pricing 325

14.3 ADVANTAGES OF THE MATCHED-MATURITY TRANSFER

PRICING METHOD
For the simple example just described, the advantages of the matched-ma
turity transfer pricing method are substantial. They include the following:

The marginal spread for each product Is accurately measured. In theory, for
each new transaction booked by the bank, a concurrent transfer rate is
assigned. The spread between the yield of the transaction and the transfer
rate represents the true profitability contribution of that product to the
bank's overall net interest margin, and therefore, is the correct spread to
incorporate in product or unit profitability analyses.

The earnings attributable to interest rate mismatching is correctly identified.

Bankers rarely appreciate the magnitude of their spreads that are truly at
the mercy of the markets and shifts in the yield curve. In our experience,
when this is elucidated using such a transfer pricing system, most senior
managers gain a much greater sensitivity and concern over the magnitude
of these "earnings at risk," and they usually act to manage it more conser
vatively. It is only with a matched-maturity approach to transfer pricing
that any reasonable estimate of the magnitude of these mismatch earnings
can be accomplished.

Each product spread is independent of any other balance sheet element

This is a critical point. Many, many bankers mentally "allocate" certain
deposit transactions to be the funding source for certain types of loans for
the purpose of estimating their "lending spreads." For example, they might
reason that MMDA deposits "fund" Prime-based loans. This approach
means that any spread that should rightfully be assigned to the efforts ex
pended in gathering advantageous MMDA deposits is usurped as a subsidy
for the lending ctivities of the bank. Aside from the unfairness aspect,
this reasoning only works as long as the volumes of these "linked" balance
sheet categories stay close to one another.
In contrast, the matched-maturity method is valid at all times, whatever
the relative levels of any loan or deposit category. For example, return to
the case study situation in Table 14.1. Suppose that the Corporate Banking
manager renounced any transfer pricing and "claimed" the retail CDs as
the proper cost of funds for commercial loans. By this logic, the entire
margin of 4% should be allocated to Corporate Banking and none for Re
tail Banking. The lenders would probably show very high current profits,
326 Chapter 14

and might reason to themselves that they could afford to lower their pric ing
to achieve greater market share.
Assume they conclude that they could double their outstandings by low
ering their loan rate to 8% from 9%. Their "analysis" would look some
thing like the upper panel in Table 14.3.

Table 14.3
Contrast Between Faulty and Correct Pricing Analyses

"Faulty" Pricing Analysis:

Original Pro Forma Increment
Loan Rate 9% 8%
"Cost of Funds" 5% 5%
Spead 4% 3%
Volume 100 200 100
Net Interest Income $4.0 $6.0 $2.0

Matched-Maturity Pricing Analysis:

Original Pro Forma Increment
Loan Rate 9% 8%
Transfer Rate 7% 7%
Spread 2% 1%
Volume 100 200 100
Net Interest Income $2.0 $2.0 $0.0

There are two errors in the upper panel. First, it assumes that loan origi
nations are the only source of net interest income, whereas Table 14.2
demonstrates that only a portion derives from lending; the rest is attribut
able to deposit-gathering and balance sheet mismatching. Secondly, it as
sumes that $200 of CDs can be generated at the same. rate as the bank
generated the first $100 of CDs. However, this is easier said than done. As
the lenders are recognizing, doubling the outstandings of any product usu
ally requires some type of major rate concession. The assumption that the
bank can also double its CD volume with NO rate concession at all is
quite unrealistic.
Using the matched-maturity method, it becomes instantly clear that this
pricing alternative would be detrimental to the bank. Outstandings might
double, but there would be no incremental revenues. Indeed, this would be
Transfer Pricing '32.7

the actual result if the bank chose to match-fund the incremental $100 of
loan growth using wholesale funding at the wholesale CD rate of 7%.
Only the matched-maturity method depicts this reality.
Therefore, one of the major advantages of the matched-maturity ap
proach is that it accurately separates the bank's net interest income for
each product and for any balance sheet mismatching. Any of those individ
ual spreads are completely independent of the existence of any of the oth
ers and may be achieved on a stand-alone basis. Referring back to the
original situation in Table 14.1, the bank could offer the commercial loans,
but not accept the CDs, if, for example, it wanted to exit retail banking. It
could achieve the spread of 2% funding with one-year wholesale CDs at
7%. Alternatively, it could fund with three-month wholesale CDs at 6%
and earn the loan spread of 2% and the mismatch spread of 1%.
Conversely, it could offer the retail CDs, but not make commercial
loans, and still earn the CD spread and/or the mismatch spread. Finally, it
could earn the mismatch spread, and not offer any other product! It would
do so by issuing $100 of three-month CDs in the wholesale markets at 6%
and investing the proceeds in a one-year maturity Eurodollar CD or Euro
dollar deposit earning 7%. Presto! It has a risky 1% net interest margin on
the mismatch with no commercial loans or retail CDs.

14.4 SELECTING A TRANSFER PRICING YIELD CURVE

This is a critical aspect of the matched-maturity method. All product
spreads will be based on the difference between the product's rate or yield
and a corresponding point on this transfer pricing yield curve. Generally, a
CD or LIBOR yield curve will be selected because these rates generally
best reflect the characteristics of bank marginal funding costs. These are
more appropriate than a Treasury yield curve, because banks cannot di
rectly fund themselves in the wholesale markets at the Treasury rate.
What are the attributes of an "ideal" yield curve for transfer pricing
activities? It would reflect a realistic, all-in cost of large blocks of whole
sale funding consistent with the target credit rating of the bank. Unfortu
nately, there are several issues with this concept that require resolution.

Funding versus Investment rate. In the real world, there may be a signifi
cant rate difference between the cost of a large block of wholesale borrow
ings and the yield earned on a large block of new investment securities.
Therefore, one could argue that the transfer rate associated in funding a
loan as a "cost of funds" is quite different from the transfer rate associated
328 Chapter 14

with the "earnings credit" given to a block of new deposits. Actually, the
difference may not be too large. Most banks tend to invest in government
or AAA-rated securities at yields substantially less than their cost of
wholesale funding (since so few banks can borrow at a AAA-rate). How
ever, for transfer pricing, focus on an incremental investment rate consis
tent with the (target) credit rating of the bank. This might be an A or BBB
corporate bond yield. Most banks' actual investment purchases are of ex
tremely high credit quality for liquidity management purposes, or to ''bal
ance out" or "average out" the credit risk of the loan portfolio. However,
the hypothetical "earnings credit" that should be given to the bank's de
posits should reflect the yield related to the overall "average" credit rating
of all earning assets. This would be a rate significantly higher than a
Treasury or AAA yield.
All this notwithstanding, the vast majority of bankers have a much eas
ier time accepting a funding perspective rather than an investment perspec
tive in considering transfer pricing methods. This will tend to give a slight
"windfall" to the deposit gatherers, but probably does not violate the 80/20
rule.

Identifying "All-In" Cost Adjustments

There are several adjustments that must be considered to quoted CD mar
ket rates before they can be used for transfer pricing. They include:

360-day rate basis. Secondary market CD, LIBOR, and federal funds
rates are quoted on a 360-day year basis that must be converted to an
actual, or 365-day, basis for internal bank use. Also, many banks will run
their sys tems on a monthly interval, necessitating converting a "bond
equivalent" semiannual coupon concept into a monthly coupon-equivalent
interest rate.

FDIC Insurance premium. All domestic deposits are assessed a BIF or

SAIF insurance premium. Quoted CD rates must be increased by a yield
equiva lent amount to adjust for this. For consistency, this implies that all
deposit generating units should also be allocated their share of the bank's
insur ance expense in the cost allocation component of any profitability
system. Thus, for a hypothetical branch that only raises deposits but makes
no loans, it will bear the BIF insurance premium for its deposits as an
interest expense, but this will be exactly offset by the transfer earnings
credit that has been "grossed up" by the insurance premium rate.

Required reserves. When they exist, Federal Reserve Bank reserve

require ments must also be considered. However, if CD rates are being
used, and
Transfer Pricing 329

there are no reserve requirements on CDs, this issue can be ignored in

generating the yield curve. Also, if LIBOR is used, then reserves are not
relevant.

Commissions. If wholesale funding sources were used, the bank would

ex pect to pay a normal level of commissions or fees to the dealers or bro
kers. These should be factored into the yield curve as well.

14.5 TARGET VERSUS ACTUAL CREDIT RATING OF THE BANK

There is one final adjustment to quoted market rates that must be dis
cussed. It relates to the bank's credit rating in the marketplace as discussed
in Chapters 10 and 11. The concepts of LIBOR and secondary CD rate
quotes apply to "prime" or top-tier banking companies. All others will face
an additional spread over these rates that generally increase for longer ma
turities.
This issue is most critical for long-term (over one year) maturities
where the bank's funding spreads (over Treasuries) should always be a
major focus of management attention. In a sense, these spreads represent a
market assessment of the overall riskiness of the bank and its debt. Larger
banking companies have widely known ratings published by the rating
agencies. These are often the most accessible guides for establishing realis
tic borrowing spreads for such entities. Another good source for large bank
holding companies are the various investment houses that will provide
borrowing spread quotes upon request.
However, a serious dilemma arises here. Suppose that for years a bank
has operated as a high quality, AA-rated bank. However, over the past
couple of years, it has run into serious credit quality problems, such that
its formal rating has slipped into a BBB category. Management has im
proved its credit policies and strengthened its credit administration prac
tices. It is absolutely dedicated to re-establishing its AA stature as soon as
possible. Therefore, it claims that its current BBB-rating is only tempo
rary. Should the bank use a "target" AA yield curve for transfer pricing
purposes (since it is anticipating regaining this rating), or should it use a
current BBB yield curve? The difference in transfer rates could be as high
as 100 basis points!
This issue of the "target" versus "actual" credit rating in determining
transfer rates has become a raging controversy in recent years because so
many banking companies have been severely downgraded in the late
330 Chapter 14

1980s and early 1990s. There are two viewpoints that each have powerful
arguments.

Viewpoint #1: Actual credit rating. In a world where market discipline is ex

ercised and rational economic forces can prevail, using current actual
credit ratings should always prevail. In such a "rational" world, the proof
is in the proverbial pudding. In other words, the markets are reflecting
management's proven behavior in allowing the bank's credit quality to
deteriorate. The cost of funds will not improve merely on the basis of
promises, but only after the markets are convinced that the bank's behav ior
has really changed and increased credit quality can be realistically an
ticipated.

Viewpoint #2: Target credit rating. If management is truly committed to im

proving the bank's credit quality and is only allowing loans consistent
with a long-run AA rating onto its books, then this is the correct "incre
mental" cost of funds and the bank should employ the AA yield curve.
There are two other arguments that are commonly expressed in support of
this approach:
■ Using a BBB yield curve would create a huge windfall in the profit
ability of the deposit gathering activities because of the "artificially"
high earnings credits they would receive. This would allow the de
posit pricers to increase the bank's posted deposit rates in an effort to
gain market share from the competition. However, any incremental
deposit dollars raised would certainly not be invested at a BBB yield.
Hence, the transfer spreads being advantageously assigned to incre
mental deposit dollars would not reflect actual total bank incremental
earnings. This violates a tenet of transfer pricing that it should gener
ally reflect actual "match-funded" bank spreads.
■ While viewpoint #1 is legitimate in a world where BBB banks actu
ally use BBB funding, the reality is that the vast majority of BBB
banks never use a single dollar of BBB funds. One of the most
important effects of FDIC deposit insurance and the "too big to fail"
corollary is that they obviate market discipline for insured deposito
ries. Banks and thrifts simply do not pay their rational economic cost
of funds because of the protection that FDIC insurance provides.
Moreover, for large blocks of incremental, long-term funding, there
are officially established avenues of subsidized funding available. For
example, the Federal Home Loan Banlcs routinely provide AA or
better long-term funds to banks and thrifts. It is not unusual for the
Transfer Pricing 331

FHLBs to offer "specials" in the form of long-term funds at or below

LIBOR or even below Treasuries! Therefore, while it may be theo
retically correct to use a BBB yield curve, in the real world, it just
does not actually happen.
In our experience, it is only realistic to use the actual credit rating at the
bank holding company level, where FDIC insurance and subsidized fund
ing are not available and market discipline is freely exercised. At the bank
level, it is extremely difficult to get management to accept a BBB cost of
funds if the bank never borrows a single dollar at a BBB rate. Also, adopt
ing a BBB yield curve for a bank (as opposed to a bank holding company)
would severely distort deposit pricing activities. Therefore, we have gener
ally accepted the use of an AA yield curve for bank transfer pricing activi
ties.
Two points must be emphasized. First, the bank's return to AA quality
must be a high probability. Second, the right marginal cost of funds for
AA credits is an AA cost of funds. ff asset quality is below AA, a higher
cost of funds should be used.

14.6 APPLYING THE 80/20 RULE TO TRANSFER PRICING

In considering the implementation of a matched-maturity transfer pricing
system, banks often become anxious about the amount of data, analysis,
and capital and human resources required to design, program, operate, and
maintain it. Hypothetically, a bank could attempt to perform a transfer
pricing analysis on every separate loan and deposit account in the bank.
For an amortizing automobile or mortgage loan, funds transfers would be
required for the expected principal paydown pattern for every month of the
life of the loan. This can quickly require an unbelievable amount of data
processing resources to analyze and report!
Here is another place where the application of the 80/20 rule can bring
sanity out of chaos. For most product groups, transfer spreads should be
determined only for the bank's total portfolio of each group, and not for
the unique profiles of originations and maturities within each line unit.
Notice the stipulation that a bankwide spread be determined for each
product group. We are not advocating determining a single transfer rate
for an entire product group.
332 Chapter 14

14.7 USE OF GUARANTEED PRODUCT SPREADS

Going one giant step further, we also advocate guaranteeing the spread
for each product for the entire budget year. Why do this? It is the only
way to be consistent with our profitability principle that variances should
only arise from controllable events. Many shifts in transfer pricing spreads
are not controllable by line unit managers. Examples include a shift in the
spread between the Prime rate and its optimal funding mix (as discussed in
Chapter 9), the effect of prepayments on mortgage loans, or general disin
termediation effects of market rate shifts on retail deposits. None of these
are controllable by line managers, but they all could have substantial ef
fects on current transfer pricing "maturity" assumptions and the resulting
product spreads.
Another major advantage of this approach is that, for most products,
transfer pricing analyses need only be updated once per year in preparation
for the budgeting cycle. The exception is an important one. We assign
"up-to-the-minute" transfer rates for large transactions (over $500,000).
These are generally either very large loan drawdowns, purchases of invest
ment securities, or the issuance of a large block of debt. In either case, the
spread in these instances can be a very sensitive function of market yield
curve conditions at the precise time the transaction is effected. For these
large items, a log is maintained in the Treasury unit of the pertinent trans
fer rate for each, and that transfer rate is permanently linked to that spe cific
asset or liability transaction.
Returning to the situation for most products, there is one more aspect of
this once-per-year updating that is crucial. The guaranteed spread for a
coming year is a four-year moving average spread, consisting of three
years of actual history and one year of forecast based on the bank's simu
lation model. This long-term averaging serves the useful role of further
smoothing year-to-year performance volatility, without hindering the
bank's overall interest rate risk management activities.
As a specific example, return to the example of Prime-indexed loans.
Chapter 9 showed that the variance minimizing funding portfolio consists
approximately of 40% federal funds, and 60% 90-day LIBOR tractor funds
(which is simply a 90-day moving average of the 90-day LIBOR rate).
Historical data from 1975 through the first four months of 1992 were ana
lyzed for a hypothetical product that is priced to yield Prime "flat" (i.e., no
spread over Prime). The calculated annual spreads generated against the
"optimal" transfer pricing portfolio are shown in Table 14.4. Notice that
there are consecutive years where there can be large year-to-year differ
ences in the spread.
Transfer Pricing 333

Table 14.4
Spreads Between Prime Rate and ..Optimal" Funding Portfolio

Four-Year Moving
Year Annual Averages
1975 1.03%
1976 1.31
19n 1.04
1978 0.80 1.05%
1979 1.09 1.06
1980 1.61 1.14
1981 1.99 1.37
1982 1.86 1.64
1983 1.38 1.71
1984 1.40 1.66
1985 1.60 1.56
1986 1.43 1.45
1987 1.33 1.44
1988 1.60 1.49
1989 1.60 1.49
1990 1.80 1.58
1991 2.49 1.87
Overall Average 1.49%

Also shown in Table 14.4 are the four-year moving averages of the
annual data points. Using the annual spreads, the largest year-to-year
change in spread was 0.69%, compared to the average spread of 1.49%.
With the moving averages, the largest change between two years was only
0.29%. Conceptually, the use of the moving averages not only gives line
managers a more stable bottom line or SVA, it also reinforces the notion
that their pricing strategies should take longer term interest rate cycles into
consideration. Pricing should not be based solely on current, perhaps un
usual, market rate relationships.
334 Chapter 14

14.8 ADVANTAGES AND DISADVANTAGES OF THE 80/20 RULE

The savings to the bank of employing the recommendations in this chapter
can be staggering. We offer one rough quantification of the benefit. Sup
pose that two banks have decided to build new transfer pricing systems.
Both have about 100 line units and the same full-line retail and corporate
product set. One decides that precision and detail are to be achieved at any
cost, the other opts for the recommendations just presented.
To estimate the difference in the "size" of the respective system data
storage requirements, we can make the crude estimates displayed in Table
14.5.

Table 14.S
Calculation of Hypothetlcal System Complexity Levels

Detailed 80/20 Rule

A roach A roach
Number of Line Untts Analyzed: 100 1
x Number of Product Groups Analyzed: 100 40
x Frequency of Analysis (Per Year): 12- 1
x Average Number of Origination Periods/Product 20 3
x Average Number of Maturity Buckets/Product: 10 2

= Number of Storage Elements Needed: 24,000,000 240

The 80/20 rule results in a system that is about 100,000-fold less com
plex! Aside from the clear logistical advantages of the simpler approach,
there is the critical advantage of the ease of understanding and gaining a
good intuitive "feel" for a data set numbering 240 versus one numbering
over 20 million by managers with a nontechnical background.
There are two disadvantages of this simplified approach. First, the four
year moving average spreads are not directly applicable to product pricing
analyses. Any spreadsheet used for understanding the profitability implica
tions of product pricing strategies should have current period transfer
rates only. A different situation exists for indexed products, such as Prime
based loans. The authors believe that VERY long-term spreads between
the index rate and the transfer funding portfolio should be employed if the
Transfer Pricing 335

product being analyzed has a maturity greater than one or two years. This
is evident from Table 14.4. Notice that in 1991, the spread for this index
increased to 249 basis points, well above its long-term average of about
150 basis points. Unless management is convinced that this wider spread is
indicative of a permanent new spread relationship between the Prime rate
and other market rates, pricing spreadsheets should reflect the long-term
value of 150 basis points.
The second disadvantage is that the use of four-year moving averages in
the profitability system does complicate the total bank net interest income
reconciliation process. This would be a major problem if the profitability
system were being used for total bank interest rate risk management analy
ses. Fortunately, it usually is not. Long-term, moving average spreads will
definitely distort the estimation of mismatch earnings for the ALCO unit
IN THE SHORT RUN. However, these short-term errors will all wash out
over extended periods. That is, a four-year moving average of the calcu
lated mismatch earnings will be useable. The authors believe that this issue
I

is quite a small sacrifice to make for the advantages listed and recommend
that others give these arguments serious consideration.

14.9 PREPAYMENT RISK

Chapter 9 introduced the seriousness of prepayment risk on asset and li
ability management analyses. Other types of embedded options features
were discussed as well. There are two fundamental methods for incorpo
rating and reflecting prepayment risk in transfer pricing systems. (Of
course, there is always the third option of simply ignoring prepayment
risk, referred to as the ostrich strategy.) The two methods may simplisti
cally be referred to as the "after-the-fact" and "before-the-fact" ap
proaches.

After-the-fact approach. This is applicable to larger-sized transactions,

es pecially where an economic prepayment penalty is charged to the bor
rower. Transfer rates are assigned based on the contractual amortization or
maturity schedule. If and when a prepayment occurs, the original transfer
funds are "sold" back to the ALCO unit with a mark-to-market prepay
ment loss (or gain) passed to the line unit in the form of a cost allocation.
In theory, the unit will collect a similar amount from the customer at the
time of the prepayment, such that the unit's profit and loss statement will
not be adversely affected. This prepayment fee collected by the ALCO
unit through the cost allocation can be used to offset by the ALCO unit
336 Chapter 14

manager actually to prepay any matched funding it may have originally

undertaken to fund the loan.

Before-the-fact approach. This is recommended for loan products

where there is generally no prepayment penalty assessed. These include
mortgage loans and installment loans. The concept is incrementally to
increase the funds transfer rates by an amount that will compensate the
ALCO unit over the expected average life of such loans for the
prepayments that may occur. A methodology referred to as the "option-
adjusted spread" (OAS) calculation provides a usable estimate. In
detennining the transfer rates, the contractual amortization patterns should
be adjusted for the long-term expected average prepayment rate. This will
significantly lower the "ex pected maturity" of the loans, and thus the
starting transfer rates. The OAS spread is then added onto the basic transfer
rates as an adjustment. These spreads are readily available from investment
firms for mortgage loans, including adjustable rate mortgages (ARMs).
Please keep in mind that even ARMs have significant OAS adjustments.
This may not be apparent at first. ARM prepayments do pick up when
fixed mortgage rates drop because of refinancing activity into fixed mort
gages or conversions. Also, ARMs have annual and lifetime caps that may
limit the extent of any interest rate resets, such that the bank's margin is
"squeezed" in periods of extreme rate volatility. The ARM OAS adjust
ment compensates the bank for these risks.
A question often asked at this point is: "Shouldn't the bank buy some
prepayment protection with the proceeds of this OAS spread being col
lected from the line units for these risks?" Indeed, this question should be
seriously discussed in the ALCO. There is no simple answer. Unfortu
nately, there is also no "pure" off balance sheet hedging instrument avail
able, even if the bank wanted to hedge itself. However, the Federal Home
Loan Banks do make available long-term fixed or floating rate advances
with prepayment (or "call option") privileges. Also, larger banking firms
can issue callable debt in the wholesale markets. It may well be prudent to
spend some of these OAS "dollars" on such protection. In any event, the
responsibility and risk involved in this decision should be borne by the
ALCO unit, not by line units.

14.10 THE LIQUIDITY COMMITMENT SPREAD

Another source of funding risk that should be considered relates to the
mismatch between the repricing sensitivity period of a transaction and its
Transfer Pricing 337

ultimate maturity. The best example of this (again) is Prime-based loans.

The weighted average sensitivity of the "optimal" funding mix of 40%
Federal Funds and 60% 90-day CD tractor is only about 27 days. How
ever, the average maturity of most Prime-based loans can often be five
years or more.
To understand the issue, it is easiest to consider the following question:
If five-year Prime-based loans were the only earning assets in a bank, what
type of funding strategy should it adopt to "match fund" these loans? Our
answer is NOT to use 40% federal funds and 60% 90-day CD or LIBOR
tractors. Why not? After all, this would indeed best address the interest
rate sensitivity aspect of this bank, but it would leave the bank woefully
exposed to liquidity risk. (This is an appropriate place to scream, "Re
member Continental Illinois!")
The concept of matched funding must encompass ALL risk dimensions:
maturity as well as rate sensitivity. Of course, for fixed-rate loans, matur
ity and sensitivity are pretty much the same. But, for floating rate loans,
the difference is significant.
The correct hypothetical answer to the question posed is to raise 60%
five-year floating rate debt indexed to three-month CDs or LIBO and
staggered into three pools (or tranches) that reset each month in sequence;
and 40% five-year fixed rate debt swapped into a federal funds equivalent
all-in cost The net characteristics of such a funding strategy would match
both the rate sensitivity and the maturity characteristics of the loans being
funded.
How would this elaborate strategy affect the all-in cost of funding? We
estimate that it would raise the cost of funds about 20 to 30 basis points
above the cost of "plain vanilla" federal funds and CD or LIBOR funding.
One can think of this incremental cost as a sort of "commitment fee" being
paid to the markets to guarantee that the funding will remain in place for
the full five-year term. Any bank that funds five-year Prime-based loans
only with short-term funds is accepting the huge risk that should any un
expected problem arise as to its credit quality, there could be a future
"run" or "flight" away from this bank's name by the wholesale markets,
triggering a liquidity crisis, or even insolvency, for the bank.
Including this aspect of funding into the transfer pricing system need
not be complicated using the 80/20 rule. We have adopted the approach
that a "generic" liquidity commitment spread of 25 basis points will be
applied to any transfer rate for any product group that has a rate reset less
than one year, but expected maturity greater than one year. Table 14.6
summarizes those product groups that commonly receive this increment to
their transfer funding rates. For deposits, the increment increases its trans-
338 Chapter 14

fer spread and its profitability. In essence, those deposit categories are be
ing compensated for the liquidity protection they provide.

Table 14.6
Products Associated with a Liquidity "Commitment" Spread

Assets
Prime-Based Commercial Loans
Other Money Market Rate-Indexed Commercial Loans
Floating Rate Consumer Loans:
Home Equity Lines of Credit
Variable Rate Credit Cards
Guaranteed Student Loans
Adjustable Rate Mortgages

Llabllltles
T-Bill Indexed or Floating Rate Retail
Deposits: Money Market Deposit
Accounts Interest-Bearing Checking
Accounts
Variable Rate Individual Retirement Accounts
Variable Rate Retail CDs (Over One Year)
Floating Rate Long-Term Debt

14.11 SPREAD OR BASIS RISK

The final source of transfer pricing risk to be considered is again related to
Prime-based loans. It also pertains to any other item that is indexed to a
rate other than the bank's transfer pricing CD or LIBOR yield curve, such
as Treasury Bills or the COFI (Cost of Funds Index). We are referring here
to the topic of spread or "basis" risk. The Prime rate or the Treasury Bill
auction rate are not perfectly correlated with typical transfer pricing (CD
or LIBOR) rates. This additional risk should not be ignored.
To introduce this topic, review Figure 9.2. Notice on the x-axis that
while there is some reduction in spread volatility using the "optimal" fund-
Transfer Pricing 339

ing mix of 40% federal funds and 60% 90-day LIBOR tractor, there is still
a substantial amount of volatility remaining. The standard deviation of this
residual "basis" or spread risk is about 48 basis points. Recall from the
same figure, as well as from Table 14.4, that the average spread was 149
basis points.
Using a normal distribution assumption, these values provide an indica
tion of the dispersion or volatility characteristics of the Prime rate funding
spreads as shown in Table 14.7.

Table 14.7
Probability Distribution of Prime Rate Spreads
Assuming a Normal Distribution Function

Assumptions:
Average (Mean) Spread 149 Basis Points
Standard Deviation 48 Basis Points
With a Probability of: Spread Will Be Greater Than: and Less Than:
68.3%
101 Basis Points 197 Basis Points
95.4%
53 Basis Points 245 Basis Points

The 68.3% probability refers to the range of spreads calculated as the

mean (average) spread plus or minus one standard deviation. The 95.4%
line refers to a range of the mean spread plus or minus two standard devia
tions. Put another way, there is less than a 5% chance (about one year in
every 20) that the spread will be less than 53 basis points or greater than
245 basis points. These statistics are quite consistent with the one observa
tion of 249 basis points for the 17-year period analyzed (Table 14.4).
Hopefully, this discussion has left the impression that there is still quite
a substantial level of margin risk, even with an "optimal'' funding portfo
lio. Indeed, there is! From the 68.3% line, it is clear that more than 30% of
the time, the Prime rate's spread will be either lower than 101 basis points
or more than 197.
How should such risk be incorporated in the transfer pricing rates for
Prime-based loans? Some banks leave this risk with the units that make
such loans. They establish a transfer funds rate based on the "optimal"
funding mix, ignoring both the liquidity "commitment" spread presented in
the last section as well as this spread risk. The attitude that goes along
340 Chapter 14

with this approach is something like: 'The line units are creating this basis
risk for the bank, so they should bear the consequences."
While this approach certainly has some legitimacy, it is not compatible
with the "no uncontrollable variances" principle. Line managers have no
control over the detailed behavior of the Prime rate. Indeed, in many
banks, it is the executive management in the persons of the CEO and
either the Chief Credit Officer or the Chief Financial Officer who deter
mine the specific timing of Prime rate changes. Therefore, one could make
the perverse argument that the CEO should bear the consequences of this
spread risk!
Our recommendation is to "charge" for this spread risk, then guarantee
whatever spread remains. A simple approach is to charge one standard
deviation. Hence, for Prime-based loans, a charge of 48 basis points would
be assessed for this risk. For products based upon other indices, studies
similar to that described in Chapter 9 and summarized in Figure 9.2 would
be conducted. (There is nothing magic about using one standard deviation.
If this seems too onerous, then one-half standard deviation can just as eas
ily be adopted.)
Even simpler would be to adopt a "generic" spread risk increment of
about 25 basis points for any indexed product. This would be applied to
any product with an optimal funding portfolio standard deviation of 20
basis points or more.

14.12 A SURVEY OF TRANSFER PRICING ADJUSTMENTS

A brief survey of some common bank products and the transfer pricing
considerations appropriate for each is presented in Table 14.8. The column
labeled "Rate Sensitivity" refers to the initial basic estimated maturity or
rate reset characteristic that is the focus of all presentations on "matched
maturity" transfer pricing, which was discussed in Sections 14.2 and 14.3.
Notice that an OAS adjustment is made only for those products that
display "economic" prepayment risk. Auto loans have relatively stable pre
payment rates that are not sensitive to interest rate movements. The esti
mated rate of such prepayments will be dominated by "demographic"
factors, such as the sale of the car, or an insurance settlement after an
accident. This can be factored into the initial estimate of the "effective"
paydown pattern used to establish the basic sensitivity characteristic by
scheduling out the contractual amortization pattern and adjusting it for an
estimated demographic prepayment rate of a little above I% per month for
a "seasoned" portfolio. Demographic prepayments occur steadily and do
Transfer Pricing 341

Table 14.8
Survey of Transfer Pricing Elements

Rate Prepayment Liquidity Spread

Product Sensitlviti OAS Commitment Risk
LOAN PRODUCTS:
Prime-Based Loans Yes No Yes Yes
Capped" Prime Loans
11
Yes Yes Yes Yes
Fixed Rate Commercial Loans Yes After-Fact No No
Fixed Rate Mortgage Loans Yes Yes No No
Adjustable Rate Mortgages Yes Yes Yes Yes
Guaranteed Student Loans Yes No Yes Yes
Credit Card (Fixed Rate) Yes No No No
Automobile Loans Yes No No No

DEPOSIT PRODUCTS:
Non-Int Bearing DDA Yes No No No
Interest DDA (Floating Rate) Yes No Yes Yes
Money Market Depostt Account Yes No Yes Yes
MMDA (Wtth a Rate Floor")
11 Yes Yes Yes Yes
Retail CDs Yes No No No

not change because of interest rate shifts. Economic prepayments occur

specifically because interest rates have lowered enough to drive borrowers
to refinance.
The "after-fact" notation for fixed rate commercial loans signifies that
there is usually a full, explicit prepayment penalty fee. Therefore, it is
appropriate to make an "after-the-fact" mark-to-market adjustment for any
associated transfer funding (or real funding). This is especially applicable
if the bank has assigned "up-to-the-minute" transfer rates for large loan
drawdowns.
The OAS adjustment for the "floored" l\1M:DA product is to reflect the
potential value of a rate floor on any deposit product for the depositor, or
"squeeze" on the bank's margin. It would lower the "earnings credit" as
signed to MMDAs through the transfer pricing system. Likewise, "capped"
Prime loans should also be assessed an OAS adjustment.
342 Chapter 14

Interestingly, there are home equity lines of credit that are indexed to
the Prime rate that have interest rate floors at or near 10%. Such a product
should have an OAS spread added to its basic transfer pricing spread in
recognition of the value to the bank of that feature.

14.13 TRANSFER PRICING ITEMS WITHOUT MATURITIES:

DOA AND EQUITY
Many of the concepts presented in Chapters 8 and 9 are pertinent here. For
noninterest-bearing demand deposits, the "core" and "volatile" portions
can be estimated as described in those chapters. The volatile portion
should be assigned a short-term effective maturity transfer rate, perhaps a
30-day tractor. The core portion should receive the "generic" long-term
tractor assumption as described in Section 9.3. Assuming that the bank has
selected a four-year tractor for its generic long-term category, then a typi
cal transfer pricing assumption might be: 20% 30-day tractor transfer rate
and 80% four-year tractor transfer rate.
Likewise, for equity, the generic long-term tractor provides a conven
ient assumption. For purists, one could argue that the duration of net eq
uity should determine the transfer rate maturity assumption.

14.14 HEDGING STRATEGY FOR THE LARGE TRANSACTION

BOOK
Section 14.7 stipulated that for large transactions (over $500,000), we
track up-to-the-minute transfer rates. Typically, the "up-to-the-minute"
concept applies only to the underlying Treasury rate quote. To this, the
bank's target credit rating spread-over-Treasuries is added to generate the
full transfer rate. In many regional banks, these large transactions are man
aged for interest rate and liquidity risk separately from the rest of the bal
ance sheet. We think this is quite prudent.
For such a "large transaction book," generally managed by the Treasury
unit, there are excellent hedging and risk management strategies available.
This is because these types of transactions tend to be few in number, very
explicit in their terms, and have full economic prepayment adjustments or
accurately priced call privileges. Examples include large commercial loan
takedowns, purchases of investment securities, and the issuance of large
blocks of debt or wholesale CDs.
Transfer Pricing 343

One durable strategy is described here. It recognizes the need to main

tain both long- and short-term casn flows within the safety zones. It also
shows how both duration and gap analyses can complement one another.
The basic procedure for a book of large, fixed rate transactions is as
follows:

Analyze quarterly cash flows using gap analysis and duration (IRE)
analysis. If the book contains a large number of short-term (under one-
year) trans actions, quarterly buckets may not be detailed enough. Some
banks use weekly short-term buckets. However, quarterly buckets work
quite well for transactions over one year. The strategy will be to use gap
analysis to measure all mismatches under two years. Mismatches beyond
two years will be analyzed with the interest rate elasticity (IRE) concept
from Chap ter 8.

Hedge all gap mismatches under two years with financial futures. The
fu tures markets are ideal for hedging time periods of two years or less. A
range of weighted gap mismatches should be established. Whenever the
gap profile under two years becomes larger than the target range (or safety
zone), that is a signal to the Treasury staff to execute some incremental
futures transactions. The safety zone should be small enough to be pru
dent, but large enough to prevent large numbers of one-contract trades.

Hedge the net IRE of all cash flows over two years with swaps. A
swap book is used to manage the IRE of all long-term cash flows. For large
books, this management process will entail a substantial number of trans
actions, establishing a pattern of regular participation in the markets. With
this constant presence, Treasury staff will become quite proficient at buy
ing and selling their positions. Again, a conservative, yet practical safety
zone should be established for the large transaction book's IRE.
Also, the Treasury unit manager should not overlook the liquidity safety
zone! Limits should be placed on the use of short-term wholesale CDs (or
commercial paper in the case of holding company activities). Whenever
the limit is reached, the treasurer should be issuing long-term debt that can
be swapped into whatever rate reset characteristic is desired.

14.15 SUMMARY
Transfer pricing is not a simple subject. However, it is possible to adhere
to the profitability principles presented in Chapter 13. In particular, the
344 Chapter 14

80/20 rule decreases the complexity of transfer pricing systems by incred

ible degrees! This allows the ALM staff to get beyond the basic sensitivity
aspect of the matched-maturity approach, and to deal with other risk topics
such as OAS adjustments, liquidity commitment fees, and "basis" or
spread risk.
As radical as some of the suggestions in this chapter may appear, we
would respectfully encourage readers to consider them well. They simplify
an extremely complex topic. Even if only employed during the develop ment
of initial prototype versions of what may be intended to be much more
"sophisticated" transfer pricing systems, we believe they will dra matically
facilitate the understanding and acceptance of these critical con cepts by a
very broad cross section of bank management and staff.