Roebers MA BMS
Roebers MA BMS
M ASTER ’ S T HESIS
Supervisors:
B. R OORDA (UT)
Author:
R. J OOSTEN (UT)
T. R OEBERS
D. F OKKEMA (EY)
P. V ERSTAPPEN (EY)
Abstract
With term premia present in the yield curve, banks have incentives to
create mismatches between term structures of cash flows and with this, ex-
pose themselves to interest rate risk. Especially in the current period of
historically low interest rates and rising pressure of competition, the con-
sequences of a return to pre-crisis interest rate levels could be disastrous
if this mismatch is too big. Regulators also acknowledge this problem,
for which they come to introduce new guidelines to manage and quan-
tify interest rate risk in the banking book (IRRBB) in a more standardized
manner.
We examine the impact of a bank’s interest rate risk appetite on its re-
turn on equity, as well as give insight in the impact of a direct capital charge
for IRRBB. We do this by creating a model that reallocates the exposures to
balance sheet items. Our model is a stylized reflection of an average, small
Dutch bank and optimizes the return on equity of a bank while being sub-
ject to interest rate risk, liquidity and capital constraints originating from
the Basel accords. In order to provide a precise calculation of the interest
rate measures, the balance sheet items are allocated to detailed subclasses
based on fixed interest rate periods. We quantify IRRBB by the change in
net interest income (NII) and the change in economic value of equity (EVE)
resulting from a set of alternative interest rate scenarios. Subsequently,
banking instruments are subject are subject to optionality, creating uncer-
tainties in future cash flows. We analyze the impact of changes in two
sources of optionality embedded in banking instruments on a bank’s inter-
est rate risk exposure.
Our findings show the added value of the introduced alternative in-
terest rate scenarios and the importance of the complementary use of the
two interest rate risk measures in controlling earnings and economic value
volatility. Furthermore, we illustrate that the impact of a decrease in term
transformation by lowering thresholds on interest rate risk measures on a
bank’s interest rate spread. We find a decrease in interest rate risk-taking
when a direct capital charge for IRRBB would be implemented in the form
of a capital indicator based on the EVE. Finally, our findings indicate that
even small changes in the duration of core non-maturity deposits and the
magnitude of the prepayment rate cause relatively big fluctuations in a
bank’s interest rate risk exposure. With this, we lay out that the interest
rate risk exposure is highly sensitive to changes in client behavior, making
interest rate risk management an even more dynamic process.
v
Acknowledgements
This thesis is the final assignment in completing my Master Financial
Engineering and Management at the University of Twente. The last six
months I had the pleasure of writing my thesis during an internship at the
FS Risk department at EY, where I have worked alongside a lot of enthu-
siastic and helpful colleagues. I want to use this section to thank a few
people for making this thesis possible.
First of all, I want to thank my colleagues at EY for their input and
healthy distractions. In particular, I want to thank Diederik Fokkema and
Philippe Verstappen for their guidance, support and flexibility, both per-
sonally and professionally, during my time as an intern.
Furthermore, I want to thank Berend Roorda, who guided me as my
first supervisor on behalf of the University of Twente. The lectures, the
guidance and the opportunity to work as a student assistant at the Fi-
nance for Engineers module contributed largely to the experience I have
gained during my Master. I am also grateful for the guidance and lectures
of Reinoud Joosten, who acted as my second supervisor. Both supervi-
sors provided me with good conversations and extensive feedback, which
allowed me to improve my work.
With this thesis, my time as a student comes to an end. Here, a spe-
cial thanks is in place to my (former) roommates from the Bentrot, who,
amongst others, made this period a time that I will never forget.
Last but not least, I want to thank Suzanne, my family and my friends
for their mental support during the last months. I am very grateful that
you are always there for me, even during unforeseen setbacks.
Contents
Abstract iii
Acknowledgements v
1 Introduction 1
1.1 Problem Context . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Research Objective . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Research design . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Literature Review 5
2.1 Definition and Origins . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 Banking Book Versus Trading Book . . . . . . . . . . 5
2.1.2 Definition . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.3 Components Of Interest Rate Risk . . . . . . . . . . . 6
2.1.4 Composition Of Interest Rates . . . . . . . . . . . . . 7
2.2 Interest Rate Risk and Bank Stability . . . . . . . . . . . . . . 9
2.3 IRRBB Regulation . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3.1 Bank For International Settlements . . . . . . . . . . . 10
2.3.2 The Basel Regulation . . . . . . . . . . . . . . . . . . . 10
2.3.3 New Developments In IRRBB Regulation . . . . . . . 12
2.4 Interest Rate Risk Measures . . . . . . . . . . . . . . . . . . . 13
2.4.1 Gap Analysis . . . . . . . . . . . . . . . . . . . . . . . 14
2.4.2 Duration of Equity . . . . . . . . . . . . . . . . . . . . 14
2.4.3 Economic Value Perspective . . . . . . . . . . . . . . . 16
2.4.4 Earnings Perspective . . . . . . . . . . . . . . . . . . . 16
2.4.5 Regulatory Scope . . . . . . . . . . . . . . . . . . . . . 17
2.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3 The Model 21
3.1 Model Objective . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.1 Asset and Liability Mix . . . . . . . . . . . . . . . . . 21
3.1.2 Interest Rate Swaps . . . . . . . . . . . . . . . . . . . . 23
3.1.3 The Objective Function . . . . . . . . . . . . . . . . . 24
3.2 Model Definition . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2.1 The Balance Sheet Definition . . . . . . . . . . . . . . 25
3.2.2 Model Constraints . . . . . . . . . . . . . . . . . . . . 26
3.3 Measuring Interest Rate Risk . . . . . . . . . . . . . . . . . . 27
3.3.1 Interest Rate Floor . . . . . . . . . . . . . . . . . . . . 27
3.3.2 ∆Economic Value of Equity . . . . . . . . . . . . . . . 29
3.3.3 ∆Net Interest Income . . . . . . . . . . . . . . . . . . 32
viii
4 Results 39
4.1 Short-Term Versus Long-Term Funding . . . . . . . . . . . . 39
4.2 Parallel Versus Non-Parallel Shocks . . . . . . . . . . . . . . 41
4.3 Short-Term Versus Long-Term Focus . . . . . . . . . . . . . . 42
4.4 Including Capital Requirements . . . . . . . . . . . . . . . . . 45
4.5 Improving Our Balance Sheet . . . . . . . . . . . . . . . . . . 46
4.6 Change in Capital Requirements . . . . . . . . . . . . . . . . 47
4.7 Change in Optionality . . . . . . . . . . . . . . . . . . . . . . 48
4.7.1 Stability of Deposits . . . . . . . . . . . . . . . . . . . 48
4.7.2 Prepayment Behavior of Mortgagors . . . . . . . . . . 49
4.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
D Risk Measures 67
D.1 Capital Requirements . . . . . . . . . . . . . . . . . . . . . . . 67
D.1.1 Total Capital Ratio . . . . . . . . . . . . . . . . . . . . 67
D.1.2 Leverage Ratio . . . . . . . . . . . . . . . . . . . . . . 68
D.2 Liquidity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
D.2.1 Liquidity Coverage Ratio . . . . . . . . . . . . . . . . 69
D.2.2 Net Stable Funding Ratio . . . . . . . . . . . . . . . . 69
Bibliography 79
ix
List of Figures
3.1 Interest rate shock for the Euro in a steepener interest rate
curve scenario. . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2 Estimated gold storage cost based on gold future prices (No-
mura, 2016). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3 EVE factors for NHG mortgages buckets. . . . . . . . . . . . 32
3.4 Change in value of bullet loan versus mortgage with pre-
payment rate of 5% in a parallel up scenario. . . . . . . . . . 33
3.5 Impact of interest rate swaps on ∆EVE. . . . . . . . . . . . . 34
A.1 Euro interest rate shock scenarios set out by the Basel Com-
mittee on Banking Supervision. . . . . . . . . . . . . . . . . . 58
A.2 Base and alternative interest rate scenarios. . . . . . . . . . . 58
List of Tables
List of Abbreviations
Chapter 1
Introduction
I wrote this thesis during an internship at the Financial Services Risk de-
partment of EY, located in Amsterdam. This department is specialized in
both qualitative and quantitative financial risk and compliance challenges.
One of today’s main topics playing a role in new regulation is interest rate
risk in the banking book. EY supports, among others, banks in organizing
the implantation of new interest rate risk regulations.
This chapter provides context to the thesis’ subject, the thesis’ objective
and elaborates on the structure of the thesis.
What would be the impact of stricter regulation on interest rate risk in the bank-
ing book and how could a bank improve its balance sheet given its interest rate risk
appetite?
1. (a) What is interest rate risk in the banking book and how does it
relate to profitability?
(b) What are the regulatory developments regarding interest rate
risk in the banking book and what other regulatory requirements
are applicable to a bank’s balance sheet?
(c) How can the interest rate risk exposure of a balance sheet be
quantified?
2. (a) How does a typical balance sheet of a small Dutch bank look
like?
(b) How can the impact of setting a bank’s interest rate risk appetite
be illustrated and how can this be used to create an improved
balance sheet allocation?
3. (a) How severely does a bank’s interest rate risk appetite affect its
earnings?
(b) What is the impact of stricter capital requirements on a bank’s
interest rate risk taking and what the impact of changes in key
modeling assumptions on the interest rate risk exposure of a
bank?
Chapter 2
Literature Review
2.1.2 Definition
The theory of financial intermediation attributes a number of activities,
commonly referred to as qualitative asset transformation. THese activi-
ties are seen as the core activities of a retail bank, and include taking on
credit risk, liquidity provision and maturity transformation. The latter
evolves in most cases as a result of liquidity provisions when long term
fixed-rate loans are funded using short-term deposits (Bhattacharya and
Thakor, 1993). With term premia present in the yield curve, banks have in-
centives to create maturity gaps, i.e., a mismatch between term structures
of cash flows. Hereby, banks expose expose themselves to interest rate risk
(Memmel, 2011).
Before we include the banking book aspect, we first take a look at the
definition of interest rate risk. One definition, often used in academic liter-
ature, is the following:
Definition 2.1. Interest rate risk encompasses all risks that are directly
or indirectly induced by uncertainty about future interest rates (Hellwig,
1994).
Definition 2.2. Interest rate risk is the potential loss from unfavorable changes
in interest rates on a bank’s profitability and market value of equity.
In this thesis we use the definition of interest rate risk in the banking
book provided by the BCBS (2016). This definition resembles the previous
definition, but stresses the direct effect of adverse fluctuations in the yield
curve on earnings and capital.
Definition 2.3. Interest rate risk in the banking book refers to the current
or prospective risk to a bank’s capital and to its earnings, arising from the
impact of adverse movements in interest rates on its banking book.
2. Basis risk. One complication of interest rate risk is that there are dif-
ferent reference rates. These interest rates tend to move together, but
are not perfectly correlated (Memmel, Seymen, and Teichert, 2016).
Basis risk describes the impact of relative changes in interest rates for
interest rate bearing instruments with the same term structure but
different interest rate indices. For instance, a basis risk exposure will
arise if the spread between three-month Treasury and three-month
LIBOR changes. This change will affect the net interest margin of a
bank as a result of changes in the spreads received or paid on instru-
ments that are repriced at that time. In the previous section, we stated
that the exposure to interest rate risk equals zero if the maturity of as-
sets perfectly matches the payments of the funding. We assumed here
that the interest rate indices for the payments are the same. Whether
this is not the case, there is still a basis risk component that can cause
exposure.
3. Option risk arises from alternative levels and terms of cash flows as a
result of optionality. Interest rate levels can trigger events embedded
in banking products. Common examples of banking products with
embedded optionality are redemption of deposits and prepayment
of loans. Also automatic optionality, for example the change in value
of certain interest rate derivatives, belongs to this type of risk.
1. The base of the interest rate is the risk-free rate, the return that can be
obtained without assuming any risks (Hull, 2012).
8 Chapter 2. Literature Review
4. The credit spread can be divided into two premiums, a general mar-
ket credit spread and an idiosyncratic credit spread. The general market
credit spread represents the spread associated with the risk premium
required by market participants for a given credit quality and is typi-
cally the required yield of a debt instrument from a party with a spe-
cific credit rating over a risk-free alternative. The idiosyncratic credit
spread is the premium for the credit quality of the specific individual
borrower and the risks associated with the credit instrument. The id-
iosyncratic credit spread takes into account other information as well,
such as risks from the sector, geographical location of the borrower
and risks associated with the credit instrument (BCBS, 2016).
The required return for instruments valued at amortize cost, e.g. con-
sumer or corporate loans, are based on two components BCBS (2016):
1. A funding rate, which is the cost of funding the loan and consists of
a reference rate plus a funding margin. The reference rate is an ex-
ternally set benchmark rate, such as the London interbank offer rate
(LIBOR). To come to a bank’s own funding rate, the funding margin
is added.
from term transformation. On the other hand, he found that the inter-
est rate margin is not affected much by the exposure to interest rate risk,
which makes it interesting to look at it from a model perspective.
by the BCBS. This measurement system, also known as the Basel Capital
Accord, the 1988 Accord or simply Basel I, called for a minimum capital
ratio of eight percent of a bank’s risk-weighted assets, and had to be im-
plemented by 1992. In 1995, the framework was refined to address also
market risk in addition to credit risk, via an amendment to the Capital
Accord. This amendment also made it possible for banks to make use of
internal models to determine their adequate market risk capital require-
ments.
In June 2004, after a consultation period of almost six years, the Re-
vised Capital Framework, better known as Basel II, was introduced. This
framework consists of three pillars, a structure which is still being used
in the Basel regulation. The minimum requirements are captured in the
first pillar. The second pillar treats the supervisory review of the capital
adequacy and internal processes of a bank. Standards of effective use of
disclosure to strengthen market discipline belong to the third pillar (Hull,
2012). The objective of Basel II was to improve the reflection of underlying
risk by regulatory capital and capture risks from innovation in the finan-
cial industry. Furthermore, the new framework sought to encourage and
reward improvements in risk measurement and controls. After the intro-
duction of Basel II, the BCBS started to focus more on the trading book in
addition to the banking book. A new amendment was issued governing
the treatment of risk measurements of banks’ trading books in 2005, which
was integrated in Basel II in 2006 (BCBS, 2006).
During the crisis, the need for increasing supervision and more se-
vere capital requirements rose. Financial institutions were too leveraged
and their capital buffers were inadequate. The absence of these standards
in combination with poor internal risk management resulted in practices
such as the mispricing of credit and liquidity risk and excess credit growth
(Baldan and Zen, 2013). As a response to the need for more supervision,
the BCBS introduced a first set of principles to manage liquidity risk in
September 2008. In 2009, new documents were issued in order to further
strengthen Basel II. These packages of documents mostly contained treat-
ments for complex securitization positions, off-balance sheet vehicles and
trading book exposures.
The financial crisis shed a light on the risks taken by banks. Often,
banks were not able to impose losses on their capital buffers (Baldan and
Zen, 2013). Inevitably, the BCBS announced higher capital standards for
international active banks in 2010. This reform in the design of capital
and liquidity was the basis of Basel III. In addition to a higher percentage
common equity to cover potential losses, the leverage ratio, capital con-
servation buffer and counter cyclical capital buffer were introduced. Also
liquidity risk is covered more comprehensively through the introduction
of the liquidity coverage ratio (LCR) and net stable funding ratio (NSFR),
see Appendix E. Moreover, global systematic important banks (G-SIBS) are
exposed to extensive additional capital and supervision.
12 Chapter 2. Literature Review
limit capital arbitrage between banking book portfolios and trading book
portfolios, which are subject to different accounting standards. Although
the motivation is logical, it is hard to create a standardized framework that
captures the interest rate risk exposure, because of the heterogeneity in
customers and risk appetite and optionality in banking products. This
challenge was also reflected in the time it took to publish a first consul-
tation paper. The BCBS spent no less than three years to publish its first
consultation in which it made an attempt at standardizing IRRBB a little
bit further by consulting on two options for regulatory treatments for IR-
RBB: a standardized Pillar 1-approach and an enhanced Pillar 2-approach
(BCBS, 2015). Due to feedback from the banking industry, the BCBS ac-
knowledged that including IRRBB in Pillar 1 would be less appropriate,
because of the heterogeneous nature of IRRBB (BCBS, 2016).
In April 2016, the BCBS presented the enhanced Pillar 2-approach in
which it continued to create a more standardized criterion to identify out-
liers by pleading for improved development of interest rate shocks, key
behavioral and modeling assumptions and internal validation processes
for internal measurement systems and models used for IRRBB. New in
this enhanced Pillar 2-approach is the more standardized disclosure of the
change in economic value of equity (EVE) based on standardized interest
rate shock scenarios. More on the new interest rate risk disclosures can be
found in Appendix F. In addition to previously prescribed shifts, a set of
non-parallel shifts is added for the EVE measure. Finally, some more mod-
eling restrictions are introduced for non-maturity deposits (BCBS, 2016).
With the introduction of new IRRBB guidelines, the introduction of an ex-
plicit capital framework for interest rate risk in the banking book seems
averted for the time being. The recurring debate of a standardized versus
the Pillar 2-approach was died down until the next regulatory attempt will
introduce itself. Until then, the task is left to the national regulators to de-
termine whether banks hold an appropriate amount of capital for this type
of risk.
per se indicate perfectly matched cash flows, but it indicates small changes
in interest rate will cause no change in portfolio value (Hull, 2012).
n
1 dB X cf i · e−y·ti
D=− · = ti · ( ) (2.1)
B dy i=1
B
∆B = −D · B · ∆y (2.2)
Where:
D = Duration
B = Bond price
y = Interest rate
ti = Time of cash flow i
cfi = Cash flow i
The duration ignores the curvature in the relative change curve of the
value of the portfolio. This can partially be overcome by capturing the
convexity, the slope of the change as result of interest rate changes as can
be seen in Formula 2.3. The change in value can be calculated using For-
mula 2.4 (Hull, 2012).
Pn
1 d2 B cf · t2 · e−yti
C = − · 2 = i=1 i i (2.3)
B dy B
1
∆B = −D · B · ∆y + · B · C · (∆y)2 (2.4)
2
Generally, duration is used in two common used measures: the duration
of equity and the price value of a Basis point (PV01). The duration method
can be generalized to use in determining the price sensitivity of all inter-
est rate dependent instruments on a balance sheet. Because the duration
of both assets and liabilities can be calculated, the duration of the equity
can be constructed, as the definition of economic value of equity is the
economic value of the assets minus that of the liabilities. The duration of
equity gives an indication about the value change as a result of relatively
small changes in the yield curve. Using the duration of equity for a one
basis point parallel change will result in the PV01.
The convexity expansion for the duration can be used to calculate the
effect of relatively large shifts in the yield curve on the bond price. Still, du-
ration only considers parallel shifts in the yield curve, because of the gen-
eralization of cash flows over time. In an environment of historically low
interest rates non-parallel shifts in the yield curve should be considered. In
addition to yield curve risk, basis risk cannot be measured using this ap-
proach. Furthermore, durational measures ignore change in cash flow as
a result of optionality affected by interest rate changes (De Nederlandsche
Bank N.V., 2005). Many banking products have embedded optionality trig-
gered by interest rates, which causes alternative expected cash flows. This
makes it important to include optionality in determining interest rate risk
in the banking book. Finally, duration is a static measure, meaning it does
16 Chapter 2. Literature Review
Where:
∆EVE i,c = Change in EVE under interest rate scenario i in currency c
2.5 Conclusions
In this chapter, we answered our first sub-questions, by determining the
definition of interest rate risk in the banking book and its components,
together with the components of the required return for banking products.
We furthermore treated the relationship between interest rate risk taking
and bank earnings. Finally, we treated the developments in interest rate
risk regulation and the measures to quantify interest rate.
• What is interest rate risk in the banking book and how does it relate to prof-
itability?
• What are the regulatory developments regarding interest rate risk in the
banking book?
Interest rate risk has increased focus on the regulator’s agenda, due to the
increased competition and the current low interest rate environment. The
BCBS strives to more standardization, but IRRBB remains, due to hetero-
geneity and the tailored approach that is needed, part of Pillar 2. How-
ever, new guidelines on measuring and disclosing IRRBB are introduced
recently. These guidelines have been used as much as possible in support-
ing the IRRBB calculations in our model.
• How can the interest rate risk exposure of a balance sheet be quantified?
While measuring interest rate risk in the banking book, it is of essence that
all interest rate risk components are covered. We laid out the commonly
2.5. Conclusions 19
Chapter 3
The Model
In this chapter, the model we use to improve the balance sheet, and with
this, illustrate the impact of changes in regulations and model assumptions
is introduced. Section 3.1 elaborates on the items for our stylized balance
sheet as well as introduces the objective function of the model. Section 3.2
explains the balance sheet definition and key constraints. Section 3.3 treats
how we include the EVE and NII measures and elaborates on our choice
for the interest rate floor. This chapter concludes with Section 3.4, in which
we explain how we come to our starting exposures and decision space.
- (i=1) Cash and balances with central banks - (j=1) Due to banks
- (i=2) Loans and advances to banks
Retail:
Government bonds: - (j=2) Demand deposits
- (i=3) Government bonds (AA- +) - (j=3) Savings deposits
- (i=4) Government bonds (A- to A+) - (j=4) Term deposits
and liabilities except for mortgages are assumed to be bullet payments. For
all balance sheet items, the interest payment and payment of the principal
are calculated monthly.
To avoid making assumptions about the redemption of mortgages, the
repayment of the principal for the mortgages is modeled solely through a
conditional prepayment rate of five percent of the principal amount at the
time of the cash flow. This value in line with the CPRs of the RMBSs used
to construct our stylized balance sheet (Dolphin Master Issuer B.V. (2016)
and Goldfish Master Issuer B.V. (2016)).
For a more detailed description of the balance sheet items, see Ap-
pendix E.
1986). The second factor that can be distinguished is credit valuation ad-
justment (CVA), for which also capital should be held (BCBS, 1993). The
calculation of the required capital for interest rate swaps is treated in Ap-
pendix C.
~ old + ~xnew · II
NII = ~xold · II ~ new − ~yold · IE
~ old − ~ynew · IE
~ new + ~o · SE
~ (3.1)
Where:
xold (y old ) = Asset (liability) exposure after repayment
xnew (y new ) = New acquired amount of the asset (liability)
o = Notional interest rate swaps
II = Interest income
IE = Interest expense
SE = Expected swap payment - fixed coupon
Table 3.3 summarizes the main revenues and costs of a bank’s income
statement. Due to a point in time optimization, our objective function does
3.2. Model Definition 25
not consider depreciations nor expected capital gains, making the depreci-
ations and expected capital gains zero. Following from this, our objective
function can be defined as:
δ (NII − LL − Coperating )
Return on equity = (3.2)
Equity
Where:
δ = Tax factor
LL = Loan loss provisions
C operating = Operating costs
X
Total liabilities and equity = yi,t = 1 (3.4)
x− +
i,t ≤ xi,t ≤ xi,t (3.6)
− +
yi,t ≤ yi,t ≤ yi,t (3.7)
o− +
t ≤ ot ≤ ot (3.8)
Here, x− +
i,t and xi,t respectively represent the lower and upper bound of
asset i.
1. The cash flows: loans are subject to prepayment risk. In our model
only mortgages are subject to this risk and this is included in the
model through a stress factor of the CPR.
2. The discount factor: the discount factor changes due to changes in
the interest rate.
CF i,T,k = (tk − tk−1 ) · Ci,T · Ni,T + Ni,T · Ik=K , for k ∈ 1..K (3.15)
Where:
tk = the timing of cash flow k
Ci,T = the coupon of asset i with maturity T
Ni,T = the principal of asset i with maturity T
Ik=K = 1 if k equals K, otherwise 0
Mortgages are subject to prepayments. Therefore, the cash flows of
mortgages include a constant prepayment rate and are determined as fol-
lows:
CF i,T,k = (tk − tk−1 ) · (Ci,T + CPR) · Ni,T,k−1 + Ni,T,K · Ik=K , for k ∈ 1..K
(3.16)
Where the remaining principal amount evolves through:
The payments are then slotted over nineteen time buckets as described
by the BCBS (2016). Ultimately, the commercial spreads should be sub-
tracted from the cash flows. An alternative given by the BCBS (2016) is to
discount using the original spreads. Since we use a hypothetical balance
sheet, we are unable to determine the precise spreads of our positions. In-
stead, we use the current yield curve of the asset or liability as input for
the discount factor. By doing so, we assume that the current commercial
margins equal the commercial margins of the current portfolio.
The second step is to calculate the economic value of an asset under
all interest rate scenarios. The economic value of asset i with maturity T
under scenario s can be defined as the sum of the discounted cash flows:
K
X
EV s,i,T = DF s,i,k · CF s,i,T,k (3.18)
k=1
3.3. Measuring Interest Rate Risk 31
The third step is to determine the percentage of change per asset per
maturity under all interest rate scenarios. This percentage change serves
as a factor which can be scaled by the exposure in order to calculate the
change in economic value of an asset subclass under an alternative in-
terest rate scenario, due to the linear relationship between the change in
economic value of the asset and the principal of the asset. Because of this
linearization, the model has only to calculate the change in economic value
of an asset once, which increases the speed of the simulation. The factor of
asset i and maturity T for scenario s is calculated by:
EV i,T,0 − EV i,T,s
∆EV i,T,s = (3.20)
EV i,T,0
Finally, the change in the economic value of equity under scenario s
is calculated by taking the sum of the product of the ∆EV-factors and the
corresponding exposures plus the decrease in fair value of the automatic
interest rate options.
I X
X 19 J X
X 19
∆EVE s = ∆EV i,T,s · xi,T − ∆EV j,T,s · yj,T + KAO (3.21)
i=1 b=1 j=1 b=1
Where:
xi,T (yi,T ) = Amount of asset (liability) i (j) with maturity T
KAO = The decrease in fair value of automatic interest rate options
Where the decrease in fair value of the interest rate swaps in scenario i
is calculated by:
n
X m
X
KAO i = ∆FVAO oi − ∆FVAO qi (3.22)
o=1 q=1
Where:
KAO i = the add-on on the EVE-measure for scenario i
FVAO i = the fair value of the automatic option for scenario i
n(m) = number of options sold (bought)
Figure 3.3 illustrates the EVE-factors for NHG mortgages under the six
interest rate scenarios. In Figure 3.4 the effect of the parallel up-scenario on
the economic value of two loans is illustrated. It can be seen that due to the
earlier repayments of the mortgage, the change in economic value an asset
is less, since the effective duration is less than that of a bullet bond. The
impact of the alternative interest rate scenarios for the interest rate swaps
is calculated likewise and is illustrated in Figure 3.5.
32 Chapter 3. The Model
Another thing to notice is that, although the initial shocks are symmet-
ric, the impacts of the shocks are not symmetric. One reason for this is that
positive shocks, i.e., shocks that increase the yield curve, tend to have less
impact than downward shocks. This can be seen in Table 3.4, where the
change in present value of a payment of 100 in ten years is illustrated. This
effect is also illustrated in Figure 3.3. Furthermore, the scenarios are also
subject to different factors of optionality which triggers a change in cash
flow. Finally, the interest rate floor is only triggered at short-term negative
shocks, which causes a change in magnitude of the shock and can cause
asymmetrical outcomes as well.
set out by the BCBS as a vital measure (BCBS, 2016). The ∆NII measures
the change in interest income as a result of parallel interest rate shocks
within a certain time period. Since the ∆NII is calculated by using the
difference of the base rate and the shocked rate, it can be proved that the
change is approximately independent of the underlying interest rate of the
asset or liability and linear to the principal as pointed out by the BCBS
(2015). As with the ∆EVE, we determine the percentage change in net
interest income given fluctuations in the yield curve for every subclass and
use the exposure to scale it.
Consider an asset with principal N that reprices at time t. Until time t,
the asset will yield a rate rt . After repricing the asset will generate a yield
similar to the forward rate and will roll over over the time period from t to
H. For simplicity, the difference in cash flows is received at H as was done
in the calculations of the (BCBS, 2015). The total net interest income over
horizon H can be written as the sum of the net interest income over the
interest fixed period from 0 to t and sum of the net interest income over
period t to H:
h rH ·H−rt ·t i
NII = N · eFtH ·(H−t) − 1 = N · e H−t ·(H−t) − 1 = N · erH ·H−rt ·t − 1
(3.24)
The present value of the net interest income between t and T can be written
as:
The difference in present value of the net interest incomes can now be cal-
culated by combining Formula 3.25 and Formula 3.27.
3.4. Simulation Input 35
I X
X 19 J X
X 19 K
X
∆NII s = ∆II i,T,s · xi,T − ∆IE j,T,s · yj,T + ∆SE o · ok (3.30)
i=1 b=1 j=1 b=1 k=1
Where:
xi,T (y i,T ) = Amount of asset (liability) i (j) with maturity T
ok = Notional of interest rate swap k
∆II = Factor for change in interest income
∆IE = Factor for change in interest expense
∆SE = Factor for change in swap payment
Issuer B.V., 2016). Based on the sum of size of the underlying mortgage,
corresponding to roughly 15 percent of the total Dutch mortgages debt
outstanding (CBS), combined with a geographical coverage of the Nether-
lands, we think the RMBSs’ underlying mortgages are representative for a
Dutch mortgage portfolio. The distribution of the mortgages over repric-
ing dates are determined by using the remaining interest rate fixed period
of the underlying mortgages. Because the performance report of the non-
NHG RMBS does not specify the repricing per loan-to-value (LTV), we use
a single distribution of mortgages over repricing dates for the non-NHG
mortgages. The starting exposures, returns and regulatory risk factors are
listed in Appendix E.
3.5 Conclusions
In this chapter, we introduced the model objective, restrictions and imple-
mentation of risk measures as well as discussed the input sources used to
construct our stylized balance sheet. With this, we answer the next set of
sub-questions:
• How does a typical balance sheet of a small Dutch bank look like?
3.5. Conclusions 37
• How can the impact of setting a bank’s interest rate risk appetite be illus-
trated and how can this be used to create an improved balance sheet alloca-
tion?
For illustrating the impact of setting the interest rate risk appetite, our
model reallocates the assets and liabilities in an optimal manner in order to
optimize the interest rate spread given a defined decision space. This de-
cision space is based on a realistic shift given a horizon of one year. Here,
not only the impact, but also the balance sheet allocation is monitored in
order to illustrate possible shifts to improve the balance sheet allocation.
39
Chapter 4
Results
For our starting position, we use a set of ten interest rate swaps to hedge
the parallel interest rate risk exposure to more appropriate values. We use a
solver to determine the allocation of interest rate swap to hedge the parallel
shocks in terms of both EVE and NII. As a result, the total capital ratio in
4.2. Parallel Versus Non-Parallel Shocks 41
the new starting balance sheet is slightly lower due to the extra risk weights
of the swaps. Furthermore, the interest rate spread is affected due to the
payment of swap rates. We observe a lower interest rate risk exposure
for parallel and steepener interest rate shocks by swapping long repricing
cash flows, i.e., 25 years, to three months repricing cash flows. However,
the equity volatility as a result of changes in the short-term rate rises, due
to an increase in short-term fixed interest periods. The difference in net
interest income arising from the non-maturity deposits is now covered by
the floating interest income of the swaps, resulting in a significantly lower
∆NII.
The limits on the other ratio’s correspond to the starting values of the
balance sheet:
rate risk appetite as non-parallel shifts could still have a great impact on
the equity of a bank.
( A ) RoE ( B ) ∆NII
on long-term measures, such as the EVE, can result in high earnings volatil-
ity of up to an NII of six percent of the bank’s equity in our model. This
means that in case of one of the two parallel alternative interest rate sce-
narios, the return on equity can deviate with six percent points a year.
Focusing entirely on earnings-based measures results in low earnings
volatility over the evaluation horizon by definition. Using the one year
NII-measure as control measure results in a relatively high return on eq-
uity of 5.79 percent. Our model looks for a reallocation of its assets and
liabilities which maximizes the interest rate spread. From a credit risk per-
spective, our model allocates the assets to maximize its return given the
allowed credit risk-weight exposure of the assets. The same is the case for
the interest rate risk exposure. Our model allocates the balance sheet items
such that the maximum impact given adverse interest rate scenarios re-
mains within limits. Therefore, it makes sense that the model exposes our
bank to the maximum EVE threshold of 15 percent in two shocks scenar-
ios in order to pursue its maximum yield, as can be seen in Table 4.3. The
impact of the NII on the return on equity is less than the impact of limiting
the EVE, as it only takes into account the mispricing in the first year, which
our bank hedges with one year swaps. Since these one year swaps affect
the term transformation only for a small amount, the effect on the return
on equity is not substantial. Furthermore, we evaluate our bank’s two year
NII in this scenario, which has a value of 3.80 percent of its equity, indicat-
ing that focusing on one year NII and the maximum EVE threshold of 15
percent can still result in a high middle-term earnings volatility. This also
suggests that keeping the one year NII at the desired level could become
costly due to the rollover of the hedging instruments against unfavorable
rates.
In the previous examples we illustrated the impact of using only one
of the two interest rate measures and laid out the need of combining both
constraints in the model. In Figure 4.3 we improve the balance sheet using
increasing limits on the EVE and NII. We use the balance sheet allocation
of the best previous outcome as input for the next run. We plotted the
outcome in a surface plot to illustrate the impact of the two measures on
the return on equity of our bank. Although the interest rate swaps have no
fair value, the annual fixed interest payments do have a negative impact on
44 Chapter 4. Results
the return on equity. Given the maximization of short-term profits, i.e., the
maximization of the return on equity, we can conclude a clear incentive
for the bank to take on more interest rate risks in terms of both earnings
and economic value perspective, despite the option to hedge its exposure
freely in terms of value. We observe a slightly downward sloping curve
for limiting the EVE, indicating that further decreasing the change in the
economic value is relatively more expensive in terms of interest rate spread
than applying a looser EVE. The earnings volatility can be hedged cheaply
compared to the economic value volatility, since only the cash flows of the
first year are taken into account. However, we observe higher potential
rollover costs in case of a higher EVE, as the gaps of maturities exceeding
one year are bigger.
As a great part of the Pillar 2 capital is now covered in the total capital
ratio, we decrease the limit of the total capital ratio with weighted interest
rate risk exposure of 15% of its capital, making the the total capital ratio
46 Chapter 4. Results
( A ) RoE ( B ) ∆EVE
12.09 percent. Now, our hypothetical bank must allocate its equity over
the different TREA building blocks. The first thing to notice in Figure 4.4a
is that by adding a capital charge for IRRBB, increasing the EVE limit has
only effect to a certain point on the return on equity. This indicates that
given a total capital ratio of 12.09 percent the maximum EVE threshold
of 15 percent is not sought, since the capital is allocated over other risk
exposures, mainly by increasing its credit risk exposure. Figure 4.4b shows
the actual EVE values for the simulation, which confirms the observation
that the maximum threshold is not sought.
rate risk exposure the maximum is not sought, leaving more capital to be
allocated to credit risk exposure. The new available exposure on our TREA
is filled with credit risk exposure by increasing the lending of more risky
mortgages. Finally, we see a decrease in mortgages with a loan-to-value of
less than 60 percent, as the NHG mortgages yield similar returns against
less credit risk exposure.
( A ) RoE ( B ) ∆EVE
change in economic value is bigger when the average duration of core de-
posits is longer. Subsequently, we use these factors as new input in order to
visualize the impact on balance sheet level for all six shocks to determine
the impact on the disclosed EVE.
Figure 4.7b illustrates the EVE values for a constant balance sheet under
different NMD repricing assumptions. Our hypothetical bank is funded
by 28.20 percent non maturity deposits, which is low compared to the
49.50 percent overnight deposits to total EU household savings (Euro Area
Statistics). Despite this, the EVE is affected heavily by a change in dura-
tion, in particular for the parallel shocks. A low duration of core deposits,
i.e., less stable deposits, increases the magnitude of the alternative interest
scenarios on balance sheet level, as deposits reprice earlier, which increases
the gap between the repricing maturities of assets and liabilities. We fur-
thermore observe that for an average duration of core deposits of two years
and less, deposits are found not efficient compared to other short-term
funding sources, as our model decreased its position instead of allocating
new business. At this point, the benefits of the deposits, i.e., the low in-
terest expenses, do not weight the disadvantages, i.e., the instability in the
form of a high stressed cash-outflow factor and relatively low duration.
Moreover, more stable funding with a higher duration against the same
interest expense is possible in the form of covered bonds.
( B ) CPR
4.8 Conclusions
This chapter answers the remaining sub-questions:
• How severely does a bank’s interest rate risk appetite affect its earnings?
We included the weighted interest rate risk exposure in the total risk ex-
posure amount. Here, the total capital ratio is used as a scalar for capital
requirements. We observed that the interest rate risk taking is more af-
fected by an increase in capital requirement than other TREA components.
This indicates that, given the capital indicator of 12.5 times the EVE used,
the return per allocated capital is higher for credit risk exposure.
52 Chapter 4. Results
Chapter 5
5.1 Conclusions
We illustrated the effect of including non-parallel shifts in the yield curve
in the EVE calculation. Although the required capital for interest rate risk
is not affected solely by parallel shocks yield curves, banks are obliged to
disclose parallel shocks of 200bps to the public to promote greater trans-
parency and comparability. We observed that disclosing values of paral-
lel shocks does not reflect the interest risk taking of a bank properly, as
our bank could expose itself heavily to non-parallel shocks. As a result,
the objective in form of the return on equity was not affected much by
further restricting exposures to parallel shocks only. However, including
non-parallel shocks in the simulation resulted in a decrease in exposure to
non-parallel scenarios for our hypothetical bank by matching the repricing
maturities more properly. As a result, the return on equity also decreased
by roughly 1.1 percent for conservative IRRBB limits (EVE of four percent)
and 0.5 percent for a more risky IRRBB appetite (EVE of 15 percent) com-
pared to restricting only parallel shocks.
Going forward, we showed that a balance sheet allocation considering
only long-term measures while optimizing the short-term returns could
54 Chapter 5. Conclusion, Discussion and Further Research
Using the proposed model, a bank can get insights in a more profitable
balance sheet allocation giving restrictions on prudential measures, visu-
alize the impact of interest rate risk limits on its potential expected return,
as well as determine the impact of possible changes in client behavior as
result of behavioral optionality embedded in banking products.
As interest rates are currently lower than ten years ago, the swap expenses
tend to be less compared to real swap expenses of a bank. Subsequently,
interest rate swaps are traded over the counter and can be customized to
a bank’s need. By assuming a set of only ten interest rate swaps, only the
notional amounts between swap maturities can be determined. In order
to make it more representative, interest rate swap exposures could be in-
cluded in more detail. Moreover, we assumed our bank can influence its
new acquired mortgages with a high level of precision. In real life, banks
are less able to determine this acquisition. However, a bank is able to hedge
its interest rate exposure more precisely to match repricing dates using cus-
tomized notional amounts. This partly compensates for the discussed in-
efficiency, as swap rates resemble the change in interest fixed period for
mortgages.
Our model indicates an optimal distribution of capital given a maxi-
mum level of risk exposure. Yet, it does not advise on the level of capital,
nor we included other forms of capital than Tier 1 capital. Moreover, we
do not include transaction costs, which a bank does pay in switching posi-
tions.
We included the main sources of embedded optionality in our model.
Other forms of optionality, for instance embedded options to extend the
duration or change interest rate characteristics, are not covered, but can
cause a change in cash flows and interest rate fixed periods under interest
rate scenarios.
57
Appendix A
F IGURE A.1: Euro interest rate shock scenarios set out by the
Basel Committee on Banking Supervision.
Appendix B
Distribution of non-maturity
deposits
All core deposits are uniformly distributed such that the average repric-
ing time equals the maximum average repricing time described in the fi-
nal guidelines. Non-core deposits are seen as not stable, and are therefore
slotted in the overnight bucket. The final distribution is illustrated in Fig-
ure B.2.
Appendix B. Distribution of non-maturity deposits 61
Appendix C
Despite that the fair value of the interest rate swaps at starting point in
the simulation equals zero, interest rate swaps do have an impact on risk
measures. In this section, we will elaborate on the impact of interest rate
swaps on the ∆EVE, ∆NII and the additional capital requirements as a
result of counterparty credit risk and credit value adjustment.
future earnings (PFE) multiplied by a factor alpha, which equals 1.4, writ-
ten mathematically:
Since we use fixed par rates, resulting in a fair value of the interest rate of
zero, the replacement cost are zero. The potential future exposure can be
calculated using the formula:
type
PFE = SF IR · |δi · dIR
i · MF i | (C.2)
Where:
SF = supervisory factor, for interest rate swaps equal to 0.5%.
δi = supervisory delta.
dIR
i = adjusted notional.
MF type
i = maturity factor, which is 1 for maturities higher than one year.
exp(−0.05 · Si ) − exp(−0.05 · Ei )
dIR
i = T radeN otional · (C.3)
0.05
Where:
Si = begin date of swap i
Ei = end date of swap i
√ s X hedge
X X hedge
K = 2.33· h· ( 0.5 · wi · (Mi · EAD total
i − Mi · Bi − wind · Mind · Bind )2 + 0.75 · wi2 · (Mi · EAD total
i ) − Mi · Bi )2
i ind i
(C.4)
Appendix D
Risk Measures
• Interest rate risk exposure, 12.5 (i.e. the inverse of the minimum 8%
risk-based capital requirement) times the EVE-measure.
The TCR of the proxy banks used for constructing our stylized balance
sheet variates between 15.5 and 19.3 percent. A note should be made that
these values only include exposures captured in pillar 1 and so do not re-
flect all risks, hence the relatively big difference compared to the regulatory
minimum.
68 Appendix D. Risk Measures
Capital measure
Leverage ratio = ≥ 3% (D.2)
Exposure measure
D.2 Liquidity
The first known literature on liquidity risk dates from 1876, when Knies
(1976) stressed the need for cash buffers to compensate for possible gaps
between cash in and out flows of banks in case the precise maturity could
not be controlled.
Liquidity risk encompasses the risk a bank may be unable to meet short
term financial demands and results from size and maturity mismatches of
assets and liabilities (Bessis, 2011). A bank with long-term commitments
and short-term funding generates cash-flow deficits. The liquidity risk re-
sults from insufficient resources to fund the long term assets, as a result
of a decrease in available funds. Liquid assets, such as cash and govern-
ment bonds, protect banks from market tension, as they can be used as
alternative source of funding for short term obligations. The cost of liq-
uidity for banks often refers to the cost of maintaining liquidity ratios at a
an adequate level, i.e. generate more stable long-term funding or holding
sufficient HQLA.
Since the global financial crisis of 2008, the focus of the BCBS on liquid-
ity risk has increased. Despite their adequate capital levels, many banks
experienced difficulties during the early "liquidity phase" of the financial
crisis in 2007 as a result of not managing liquidity risk in the right man-
ner. Before the crisis, the asset markets were resilient and funding could be
acquired against low costs. A rapid reversal in market conditions showed
the volatility of liquidity, resulting in severe stress scenarios under banks.
According to Ferrari and Ruozi (2009), the insufficient liquidity of banks’
is rather the result of the financial crisis than the cause of it. Instead, these
authors state the crisis comes forth from the lack of proper principles of a
healthy and prudent management together with pursuing too much short-
term profits.
To promote both the short- and long-term resilience of banks’ liquidity
risk profiles and in order to manage and monitor liquidity risk, the BCBS
introduced two liquidity measures (BCBS (2013a), BCBS (2014a)): the liq-
uidity coverage ratio (LCR) and the net stable funding ratio (NSFR). The
D.2. Liquidity 69
Stock of HQLA
LCR = ≥ 100%
Total net cash outflows over the next 30 calender days
(D.3)
Appendix E
Government Bonds
Bonds issues by sovereigns. Subclass: Credit rating and Maturity.
Retail Mortgages
This asset class includes loans secured by residential property and can be
divided in subclasses based on loan-to-value and maturity.
Other Assets
This asset class encompasses other non-interest bearing assets and is there-
for excluded from interest rate risk calculations.
Demand Deposits
Funds held in an account from which deposited funds can be withdrawn
at any time without any advance notice to the depository institution. Sub-
class: maturity.
72 Appendix E. Balance Sheet Definition
Savings Deposits
Interest-bearing deposits that provides a modest interest rate. Sub-class:
maturity.
Term Deposits
Deposits that has a fixed term. Subclass: maturity.
Secured Funding
Secured funds, on the shorter term (<1Y) commercial paper, on the longer
term bonds.
Other Liabilities
This asset class encompasses other non-interest bearing liabilities and is
therefor excluded from interest rate risk calculations.
Equity
Banks’ shareholders’ equity. This class is held stable over the simulation
and is excluded from interest rate calculations by regulation.
73
74
TABLE E.2: Asset starting exposure and risk factors.
Asset Start RW <1Y RW >1Y HQLA CI-Factor RSF <6M RSF 6M-1Y RSF ≥1Y
Cash and equivalents 1.0% 0% 0% 100% 0 0%
NHG 25,8% 5% 5% 0% 0,7% 50% 50% 65%
Loan-to-value <60% 22,8% 30% 30% 0% 0,7% 50% 50% 65%
Loan-to-value <80% 18,0% 35% 35% 0% 0,7% 50% 50% 65%
Loan-to-value <102% 17,3% 55% 55% 0% 0,7% 50% 50% 65%
Government bonds (AA-) 1,0% 0% 0% 100% 0% 5% 5% 5%
Government bonds (A- to A+) 1,4% 20% 20% 85% 0% 15% 15% 15%
According to
Loans and advances to banks 7,7% 20% 50% 0% 0 50% 50%
maturity
Other assets 4,9% 60% 60% 35% 0% 100%
Starting Weights Cash Outflow Factor ASF Factor <6M ASF Factor 6M-1Y ASF Factor ≥1Y
Due to banks 0,7% 100% 0 50% 50%
Demand deposits 12.0% 5%, 10% 90% 90% 100%
Savings deposits 16,2% 5%, 10% 95% 95% 100%
Term deposits 15,9% 100% 95% 95% 100%
Secured funding 45,6% 100% 0 50% 100%
Other liabilities 5,2% 0% 50%
Equity 4,5% 0% 100% 100% 100%
E.3. Balance sheet input 75
Appendix F
Bibliography
Cox, J. C., Jr. J. E. Ingersoll, and S. A. Ross (1985). “A theory of the term
structure of interest rates.” In: Econometrica: Journal of the Econometric
Society 69, pp. 385–407.
De Nederlandsche Bank N.V. (2005). Guidelines on interest rate risk in the
banking book.
Delta Lloyd Bank (2015). Delta Lloyd Bank, 2015 Annual report.
Dolphin Master Issuer B.V. (2016). Dolphin Monthly Portfolio and Performance
Report.
Drehmann, M., S. Sorensen, and M. Stringa (2006). “Integrating credit and
interest rate risk: A theoretical framework and an application to banks’
balance sheets. In Bank for International Settlements Conference on
“Risk management and regulation in banking”. In: Basel 6, pp. 1–33.
Dunn, K. B. and J. McConnell (1981). “Valuation of GNMA Mortgage-Backed
Securities.” In: The Journal of Finance 36, pp. 599–616.
EBA (2015). Guidelines on the management of interest rate risk arising from non-
trading activities.
English, W.B. (2002). “Interest rate risk and bank net interest margins”. In:
BIS Quarterly Review 12.
European Parliament and the Council of the European Union (2013). Reg-
ulation No 575/2013 of the European Parliament and of the Council.
EY (2016). Insights from EY European IRRBB Survey 2016 for banks.
Ferrari, R. and R. Ruozi (2009). “Liquidity risk in banks: economic and reg-
ulatory aspects profiles (Translation)”. In: Banche e Banchieri 2, pp. 85–
101.
Flannery, M. J. (1983). “Interest Rates and Bank Profitability: Additional
Evidence: Note.” In: Journal of Money, Credit and Banking 15, pp. 355–
362.
Fraser, D.R., J. Madura, and R.A. Wigand (2002). “Sources of bank interest
rate risk”. In: The Financial Review 37, pp. 351–367.
Goldfish Master Issuer B.V. (2016). Goldfish Monthly Portfolio and Perfor-
mance Report.
Hellwig, M. (1994). “Liquidity provision, banking, and the allocation of
interest rate risk”. In: European Economic Review 38.7, pp. 1363–1389.
Hull, J. (2012). Risk Management and Financial Institutions. John Wiley &
Sons.
Jones, D. (2000). “Emerging problems with the Basel Capital Accord: Reg-
ulatory capital arbitrage and related issues.” In: Journal of Banking and
Finance 24, pp. 35–58.
Koch and MacDonald (2014). Bank management. Nelson Education.
Kraft, D. (1994). “Algorithm 733: Tompfortran modules for optimal control
calculations”. In: ACM Transactions on Mathematical Software 20, pp. 262–
281.
Maes, K. (2005). “Interest Rate Risk in the Belgian Banking Sector”. In: Na-
tional Bank of Belgium.
Markowitz, H. (1952). “Portfolio selection”. In: The Journal of Finance 7,
pp. 77–91.
BIBLIOGRAPHY 81